Corpus

Learning about the causal structure of the world is a fundamental problem for human cognition. Causal models and especially causal learning have proved to be difficult for large pretrained models using standard techniques of deep learning. In contrast, cognitive scientists have applied advances in our formal understanding of causation in computer science, particularly within the causal Bayes net formalism, to understand human causal learning. In the very different tradition of reinforcement learning (RL), researchers have described an intrinsic reward signal called 'empowerment' which maximizes mutual information between actions and their outcomes. Empowerment may be an important bridge between classical Bayesian causal learning and RL and may help to characterize causal learning in humans and enable it in machines. If an agent learns an accurate causal world model, they will necessarily increase their empowerment, and increasing empowerment will lead to a more accurate causal world model. Empowerment may also explain distinctive features of children's causal learning, as well as providing a more tractable computational account of how that learning is possible. In an empirical study, we systematically test how children and adults use cues to empowerment to infer causal relations and design effective causal interventions. This article is part of the theme issue 'World models in natural and artificial intelligence'.

Here we explore whether adults and children differ in the way they searchthrough a semantic network, and the role those differences may play in hypothesisgeneration. Participants generated sequential guesses about a novel causal relation.A week later, they completed a similarity spacing task and we measured the average“distance” between guesses. We find that adults show greater dependenciesbetween sequential guesses than preschoolers, and generate a less diverse set ofoptions. These findings may support the idea that development can be viewed asanalogous to simulated annealing strategies in machine learning that start “hot” (inearly childhood), generating wider and more variable searches, and eventually cool(in adulthood) to generate narrower searches.

A long-standing finding in the causal learning literature is that adults struggle to identify conjunctive causal rules, where an effect requires the simultaneous presence of multiple causes, while performing better in disjunctive settings. However, most demonstrations of this ``conjunctive handicap'' rely on passive observation paradigms with limited evidence, where learners have no control over evidence generation. This paper asks whether this bias persists when adults are granted agency through active exploration. Using a modified ``blicket detector'' task, adult participants freely intervened to identify causal objects under conjunctive or disjunctive rule structures. We show that active exploration substantially improves adults' conjunctive causal reasoning, although conjunctive rules still require more tests to infer than disjunctive rules. We further compare human performance to a range of large language models in the same setting. While some state-of-the-art models approach human-level performance on hypothesis inference accuracy, they often exhibit less efficient exploration strategies and similar conjunctive-disjunctive performance gaps.

Data for The Star Machines study described in our paper "Empowerment Gain and Causal Model Construction: Children and adults are sensitive to controllability and variability in their causal interventions "

A long-standing finding in the causal learning literature is that adults struggle to identify conjunctive causal rules, where an effect requires the simultaneous presence of multiple causes, while performing better in disjunctive settings. However, most demonstrations of this ``conjunctive handicap'' rely on passive observation paradigms with limited evidence, where learners have no control over evidence generation. This paper asks whether this bias persists when adults are granted agency through active exploration. Using a modified ``blicket detector'' task, adult participants freely intervened to identify causal objects under conjunctive or disjunctive rule structures. We show that active exploration substantially improves adults' conjunctive causal reasoning, although conjunctive rules still require more tests to infer than disjunctive rules. We further compare human performance to a range of large language models in the same setting. While some state-of-the-art models approach human-level performance on hypothesis inference accuracy, they often exhibit less efficient exploration strategies and similar conjunctive-disjunctive performance gaps.

Here we explore whether adults and children differ in the way they search through a semantic network, and the role those differences may play in hypothesis generation. Participants generated sequential guesses about a novel causal relation. A week later, they completed a similarity spacing task and we measured the average “distance” between guesses. We find that adults show greater dependencies between sequential guesses than preschoolers, and generate a less diverse set of options. These findings may support the idea that development can be viewed as analogous to simulated annealing strategies in machine learning that start “hot” (in early childhood), generating wider and more variable searches, and eventually cool (in adulthood) to generate narrower searches.

= .60; Experiment 2). Consequently, these results tentatively suggest that assessments involving pictures may underestimate children's capacities in some contexts. To accurately measure developing capacities in children from diverse backgrounds, it is critical that assessment tools are appropriately adapted to environmental contexts. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

. We found that for "win-relevant" variables, children chose easier settings when playing to win than when playing for fun. By contrast, for "win-irrelevant" variables, children generally preferred similar settings across conditions. Children also judged "win-relevant" variables as more important to winning than "win-irrelevant" variables and judged both as irrelevant to having fun. These results suggest that playing to win and playing for fun are distinct motivational contexts to which children can appropriately adapt their decisions during play. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

Curiosity is adaptive, enhances learning, and reduces uncertainty. Social curiosity is defined as the motivation to gain information about the actions, relationships, and psychology of others. Little is known about the developmental and evolutionary roots of social curiosity. Here, across three comparative studies, we investigate whether chimpanzees ( n = 27) and young children (4–6 years old, n = 94) show particular interest in social interactions among third parties. Chimpanzees and children preferred to watch videos of social interactions compared with videos of a single conspecific (Experiment 1) and young children and male chimpanzees even paid a material cost to gain social information (Experiment 2). Finally, our results show that boys become more curious about negative social interactions whereas girls become more curious about positive social interactions as they develop, while chimpanzees demonstrated no preference for negative versus positive social interactions (Experiment 3). Taken together, these findings suggest that social curiosity emerges early in human ontogeny and is shared with one of our two closest living relatives, the chimpanzees.

Implications draw on the history of transformative information systems from the past.

Abstract Almost all of human infants’ experience and learning takes place in the context of caregiving relationships. This essay considers how infants understand the care they receive. We begin by outlining plausible features of an “intuitive theory” of care. In this intuitive theory, caregiving has both a distinctive foundational structure and distinctive features that differentiate it from other social relationships. We then review methods and findings from research on infants’ understanding of people and social relationships. We propose that even before infants can use language, they may understand caregiving as an abstract intuitive theory with some features in common with how adults think about caregiving. In particular, infants understand care relationships as intimate, altruistic, and asymmetric. We review work that starts to shed light on this proposal, including the findings that infants distinguish between intimate relationships and merely positive ones and that they have asymmetric expectations of responses to distress in intimate relationships between large and small individuals. The proposal that infants can make these inferences has societal and political implications for how we structure caregiving in early life.

For most people, most of the time, caring for others, particularly close others-children and parents, wives and husbands-is at once one of the most meaningful, important, and morally compelling things we ever do-and one of the most difficult. With a few exceptions, however, this foundational human capacity has been oddly invisible in the social and human sciences. In this volume of Dædalus, “The Social Science of Caregiving,” we aim to at least begin to remedy this. We include essays ranging across a wide landscape of the social sciences and sciences, from biology and psychology to philosophy, political science, and policy. The collection derives from an interdisciplinary project of the same name at the Center for Advanced Study in the Behavioral Sciences (CASBS) at Stanford University, co-led by the issue editors in collaboration with Alison Gopnik's lab at the University of California, Berkeley.Several overarching themes emerge from these essays. First, there is the wide-ranging nature of human care, both in terms of the cared-for and the carers. The canonical biologically grounded case may be care for offspring, but this rapidly extends to care for elders, for the ill, for distant others in a “community of fate,” and even care for the dead.1 A strikingly wide variety of people may be carers with different relationships to the cared-for. Again, the canonical biological example may be mothers, given that relationships of care have a long evolutionary history, particularly in mammals and birds. But these essays emphasize the ways that men and women, as well as a diverse group of paid and unpaid people, are involved in caregiving.2 Similarly, a remarkably wide range of institutions, with very different histories, roles, and structures, is involved in caregiving. This ranges from more obvious government institutions and policies to religious traditions and institutions, to formal unions, geographical neighborhoods, and historic Black community centers.3We might ask then, what unites this disparate range of relationships and phenomena? What makes them all examples of care? A few themes emerge here as well. One is that care is intrinsically asymmetrical; it depends on the idea that the carer has resources that the cared-for person does not. Second, care has an intrinsically altruistic character: it involves the carer donating resources to the cared-for, regardless of return, and doing so precisely because the cared-for lacks the necessary resources. This is particularly vivid in unpaid care relationships, such as family relationships. But even when care is paid labor, it has this kind of altruistic element. In most cases, the caregiver is paid by someone other than the cared-for person, either another caregiver or an institutional source of care. And psychologically, paid caregivers often feel altruism toward the people they care for, and indeed this is a source of meaning and satisfaction. These features of care make it very different from the kind of standard social and economic transactional relationships, such as those between employers and employees, buyers and sellers, or cooperative partners, that can be characterized in terms of a social contract between two equivalent autonomous agents. They also differentiate care from power relationships, which involve similar asymmetries between those with more resources and those with less, but assume that the consequence of such asymmetries is that the less powerful agent will serve the interests of the more powerful one. These distinctive features of care may indeed have contributed to the neglect of these relationships in standard economic and political accounts.Other features of care are more variable but nevertheless seem to be important in many cases. Care often seems to involve local attachments, whether these are the classic emotional bonds of attachment theory or more abstract relationships between members of a particular community, such as the Black institutions discussed by Maisha T. Winn and Nim Tottenham in their essay, or even the relationships we have with those who are no longer alive, as Phil Ford, Jacob G. Foster, and J. F. Martel describe in their contribution to this volume.4 On the other hand, care can also take on a kind of universality in religious or philosophical contexts, as Zachary Ugolnik and Eric Schwitzgebel discuss in their respective essays. Similarly, there are interesting questions about the motivations and objectives of care. In the simplest case, carers might be motivated to increase the objective well-being of the cared-for-what economists would call their objective utilities-for example, by feeding an infant or giving medication to an elder. But in other cases, the carer may be more concerned with the subjective utilities of the cared-for-what the cared-for thinks of as their own best interests rather than what the carer might consider to be best for them. The case of elders makes this contrast vivid: what should a carer do about a parent who is determined to eat meals that are objectively bad for him or to continue living in a house that may no longer be physically safe? A third form of care involves neither type of utility but rather tries to donate resources in a way that confers autonomy on the cared-for. Rather than trying to fulfill particular desires or goals, subjective or objective, the carer may be working to give the cared- for enough resources to fulfill those goals themselves. This sort of care is especially vivid in cases like adolescence or illness. But it may also be involved in, for example, the decisions of a richer community that aims to care for one with fewer resources.We have organized these essays into roughly three groups: one that focuses on biological and psychological perspectives, another that addresses more abstract philosophical and sociological themes, and a third that is concerned with policy questions. The first set of essays examines the psychological and social underpinnings of care both for children and elders. Ashley J. Thomas, Christina M. Steele, Alison Gopnik, and Rebecca R. Saxe consider how infants themselves understand and identify caregivers, with empirical results that suggest that even surprisingly young infants make inferences about care.5 Seth Pollak and Megan Gunnar review the substantial literature on the crucial effects of early care and nurturance on later development, an area where there has been extensive empirical work, and discuss its broader implications.6 Monica E. Ellwood-Lowe, Gabriel Reyes, Meriah L. DeJoseph, and Willem E. Frankenhuis explore the particular issues that arise in low-income families and discuss the ways that different environments might shape caregiving practices, while preserving the basic structure of care.7 Winn and Tottenham look to Independent Black Institutions (IBIs) established in the late 1960s as sources of insight.8 They explain how three pillars of Black education across IBIs (Identity, Purpose, and Direction) map onto beneficial practices identified in the psychological and neuroscience literature on care and development, such as exposing children to caregivers beyond simply their parents and teachers by including elders, school employees, and other alloparents. Toni Schmader and Katharina Block consider the question of why people might choose to take on or fail to take on the role of carers, with men as a particularly striking example, showing that paradoxically, cultures with more gender equality may make it more difficult for men to take on such roles.9The essay by Claire M. Growney, Caitlin Zaloom, and Laura L. Carstensen and the one by Elizabeth Fetterolf, Andrew Elder, Margaret Levi, and Ranak B. Trivedi argue for a new model of care for the elderly in which the need for autonomy and usefulness of the cared-for has equal standing with their need for assistance.10 For Growney, Zaloom, and Carstensen, changes in real estate markets, zoning, and planning are essential to create and sustain age-diverse neighborhoods that enable elders to help in the care of younger people, and the young to aid the old in turn. Fetterolf, Elder, Levi, and Trivedi focus on the necessary, if stressful, negotiations between the person in need of care, their family members, the in-home carers, the health experts, and those who pay the bills. The introduction of technology into these relationships can ease some of the human burdens of care but can also introduce conflicts. The authors document both.The second set of essays looks at more abstract aspects of care. These essays focus on the interrelated issues concerning the care of others, the divine, the dead, and AI agents.11 They also explore how these approaches can inform our daily life and offer insights into what we value in human care. Notably, these authors provide different types of care that are meaningful in their particularity and, at once, potentially expandable based upon that foundation of meaning. Schwitzgebel compares the Golden Rule (do unto others as you would do unto yourself) to what he calls extending your “concern for those nearby to more distant people” as different philosophical strategies aimed at generating care more broadly.12 Ultimately, assuming we care for those already close to us, he argues that extending that concern can be a more effective strategy to guide our actions than starting with our own preferences and projecting those preferences upon others.Ugolnik addresses the underlying importance of religious practices and institutions upon caregiving.13 Comparing Buddhist and Christian narratives of care, he argues the divine often cares and is cared for, elevating care to a sacred action. According to these traditions, care is thus a divine activity in which humans participate by engaging in caregiving practices, an experience that extends beyond the giver and receiver of care into a wider network of relationships. Ford, Foster, and Martel examine how we care for the dead, offering a theoretical approach to acknowledging the material and cultural links between the past and the present, highlighting our dynamic relationships with those who are no longer living.14 Brian Christian looks at our conceptions of artificial agents both as potential carers and reflections of human care.15 He emphasizes the long history of human thought about relationships to and among these artificial agents, well before such agents actually existed. But, of course, these considerations are particularly salient as AI becomes more sophisticated and plays a more central role in our lives.Collectively, this philosophical set of essays broadens our understanding of various types, models, and motivations of care and caregiving, whether we are involved in a caregiving relationship with the dead, the divine, or the artificial. The authors also provide substantive insights we can apply when supporting care in our political economy, the focus of the remaining essays.The third set of essays argues for care as a social and governmental responsibility that comes with costs, yes, but also individual and collective benefits. This means, first, the recognition of the obligations of the members of a society to those beyond their family and friends. Second, it requires attention to the infrastructure of care: the laws, institutions, organizations, and financing to support these obligations. Finally, these essays claim that a caring society is based on a complex network of relationships that includes not only family and friends but also paid caregivers, medical professionals, insurance agents, nonprofit and religious organization personnel, and, inevitably, government bureaucrats.The essays go beyond the neighborhood community and the home. Their concern is transformation of the contemporary political economy. Robert H. Frank advocates increased public investment, supported by a small change in taxation policy.16 The effect would reduce wasteful consumption while using those dollars to fund collectively beneficial outcomes for society at large. Elizabeth Garlow and Anne-Marie Slaughter make the case for a worldview of care in which human relationships and connections take precedence over approaches grounded in narrow self-interest.17 They draw on scientific evidence and current practices to demonstrate the viability and superiority of policies informed by the worldview of care. Gregg Gonsalves and Amy Kapczynski use the history of both the successes and failures of public health to argue for major reform of the infrastructure and public financing necessary for what they call “the social life of care.”18 Its inception, however, will depend on effective social mobilization, a question Levi addresses in her essay.19 The fact that all of us need care and so many of us provide care forms the basis for generating an expanded and inclusive community of fate. One effect would be the capacity for collective action. Another would be a venue for civil, if heated, debate about the most appropriate policies.Finally, Jane Hirshfield's poem “O, Responsibility” and Roz Chast's cartoon about the paradoxes of elder care capture the subtle and revelatory insights that only art can provide.20The Social Science of Caregiving is an ambitious project; indeed, it is just the kind of project CASBS at Stanford University thrives on. The workshops and the essays gathered in this issue of Dædalus represent small and initial steps toward assembling what we know, what we need to know, and what we need to do. Bringing together multiple disciplines reveals the diversity of forms of care and caregiving across history and place. But it also clarifies what all successful care and caregiving have in common: a commitment to the autonomy and well-being of the cared-for, respect by means of both recognition and appropriate compensation for those providing the care, and the establishment of supportive societal and public institutions.This issue and the larger project would not have been possible without the generous support of the Templeton World Charity Foundation and the Alfred P. Sloan Foundation. We also thank the University of California, Berkeley for supporting Alison Gopnik's lab and the team at the Center for Advanced Study in the Behavioral Sciences (CASES) at Stanford University for hosting the project and its convenings. We’re grateful for their institutional support and the contributions of all participants convened at the workshops, especially the authors and artists whose work appears in this volume. Finally, we thank the Academy team, including Phyllis Bendell, Key Bird, and Peter Walton, for their care and precise editing in ushering forward this issue.

There has been extensive discussion about gender gaps in representation and career advancement in the sciences. However, psychological science itself has yet to be the focus of discussion or systematic review, despite our field's investment in questions of equity, status, well-being, gender bias, and gender disparities. In the present article, we consider 10 topics relevant for women's career advancement in psychological science. We focus on issues that have been the subject of empirical study, discuss relevant evidence within and outside of psychological science, and draw on established psychological theory and social-science research to begin to chart a path forward. We hope that better understanding of these issues within the field will shed light on areas of existing gender gaps in the discipline and areas where positive change has happened, and spark conversation within our field about how to create lasting change to mitigate remaining gender differences in psychological science.

Unlike other primates, young children have been shown to exhibit seemingly irrational overimitation—faithfully copying unnecessary steps from a demonstration. We tested 3- to 5-year-old children (N = 39) and capuchin monkeys (N = 21) on a causal reasoning task in which a sequence of two actions was demonstrated, followed by reward production. We manipulated (1) causal plausibility and (2) the degree and nature of demonstrator intentionality, to explore the hypothesis that children—but not capuchins—integrate information about demonstrator intent and causal relations to infer which actions are necessary. We compared both species’ behavior to Bayesian computational models with the same varying social and physical expectations. Our results suggest that both species can learn from causal demonstrations, and that their copying behavior is affected by both the demonstration’s causal plausibility and the demonstrator’s communicative cues, but that children may be unique in interpreting communicative cues as having pedagogical intent.

Learning about the causal structure of the world is a fundamental problem for human cognition. Causal models and especially causal learning have proved to be difficult for Large Models using standard techniques of deep learning. In contrast, cognitive scientists have applied advances in our formal understanding of causation in computer science, particularly within the Causal Bayes Net formalism, to understand human causal learning. In the very different tradition of reinforcement learning, researchers have described an intrinsic reward signal called “empowerment” which maximizes mutual information between actions and their outcomes. “Empowerment” may be an important bridge between classical Bayesian causal learning and reinforcement learning and may help to characterize causal learning in humans and enable it in machines. If an agent learns an accurate causal world model they will necessarily increase their empowerment, and increasing empowerment will lead to a more accurate causal world model. Empowerment may also explain distinctive empirical features of children’s causal learning, as well as providing a more tractable computational account of how that learning is possible. In an empirical study, we systematically test how children and adults use cues to empowerment to infer causal relations, design effective causal interventions and appropriately generalize to new contexts.

Many childhood assessments rely on picture stimuli, but children in diverse early environments possess varying amounts of experience with pictures. Two preregistered experiments, conducted in 2022-2023, investigated whether picture assessments are valid across diverse contexts. Low-to-middle-income children (n = 192; 2-7 years; 85 females; all Black) in their first month of formal schooling in Mombasa County, Kenya, an early environment with relatively few pictures, performed more accurately on an object vocabulary task than a picture vocabulary task (β = .07, p < .001) (Experiment 1). Middle-to-high-income children (n = 96; 2-3 years; 52 females; predominantly White and Asian) in the San Francisco Bay Area, an early environment with relatively more pictures, performed similarly on object and picture vocabulary tasks (β = .02, p = .60) (Experiment 2). Consequently, these results tentatively suggest that assessments involving pictures may underestimate children’s capacities in some contexts. To accurately measure developing capacities in children from diverse backgrounds, it is critical that assessment tools are appropriately adapted to environmental contexts.

We study how children develop a causal curriculum to achieve a challenging goal that is not solvable at first. Adopting the Procgen environments that include various challenging game tasks, we found that 5- to 7-year-old children actively used their current level competence to determine their next step in the curriculum and made improvements to their performance during this process as a result. This suggests that children treat their level competence as an intrinsic reward, and are motivated to master easier levels in order to do better at a more difficult one, even without explicit reward. However, our findings also suggest that children’s self-designed curricula may not always be the most effective design. Rather, repeatedly practicing on the difficult target task may be sufficient. Notably, when constrained to stay on the target task instead of crafting their own curricula, more children actually succeeded and made greater progress in the game, suggesting that children perceive a curriculum as beneficial, even when focusing on a singular difficulty might prove more effective.

Humans build causal models to navigate their environments, act effectively, and pursue goals. Prior work has examined causal controllability and variability separately, showing that even young children are capable causal learners who seek novelty, surprise, and confounded evidence. However, it remains unclear whether they prioritize controllability and variability when both are available. We presented children (ages 5–10) and adults with three virtual machines: one offering controllability without variability, one offering variability without controllability, and one combining both properties through systematic input-output relationships. Across age groups, participants overwhelmingly preferred the machine with both controllability and variability when asked to perform various new tasks, generalizing and applying its abstract functional structure to different inputs and modalities. For further details, please refer to our Philosophical Transactions A paper titled "Empowerment Gain and Causal Model Construction: Children and adults are sensitive to controllability and variability in their causal interventions.”

Prior research has shown that American children (ages 3 to 8) explore uncertain options at strikingly high rates — even when it comes at a cost to reward maximization, when explicitly instructed to seek rewards, and whether they are choosing for themselves or others. In contrast, adults explore significantly less, showing greater sensitivity to cost. What drives this developmental difference? In a series of studies, we test whether children's exploration is motivated by information value rather than reward value and whether adults also incorporate information value into their decisions. We also examine the role of time horizon in the explore-exploit tradeoff. Our findings reveal that both children and adults consider information value and time horizon when deciding whether to explore or exploit.

Intrinsic motivation plays a crucial role in shaping exploration and learning, yet its specific contributions to causal discovery remain underexplored. This study examines the impact of three intrinsic motivation metrics—entropy, information gain, and empowerment—on causal learning outcomes. Across two experiments, participants engaged in interactive tasks requiring them to infer causal structures through exploration. Results indicate that information gain and empowerment significantly predict learning success, whereas broad, undirected exploration (entropy) does not. These findings suggest that learners optimize causal discovery by prioritizing actions that maximize information and control, rather than engaging in indiscriminate exploration. Our study offers insights into how strategic exploration facilitates causal reasoning and how these principles can be applied to machine learning.

What drives exploration? Understanding intrinsic motivation is a long-standing challenge in both cognitive science and artificial intelligence; numerous objectives have been proposed and used to train agents, yet there remains a gap between human and agent exploration. We directly compare adults, children, and AI agents in a complex open-ended environment, Crafter, and study how common intrinsic objectives: Entropy, Information Gain, and Empowerment, relate to their behavior. We find that only Entropy and Empowerment are consistently positively correlated with human exploration progress, indicating that these objectives may better inform intrinsic reward design for agents. Furthermore, across agents and humans we observe that Entropy initially increases rapidly, then plateaus, while Empowerment increases continuously, suggesting that state diversity may provide more signal in early exploration, while advanced exploration should prioritize control. Finally, we find preliminary evidence that private speech utterances, and particularly goal verbalizations, may aid exploration in children. Our data is available at https://github.com/alyd/humans_in_crafter_data.

We investigate young children's capacity for "causal relational reasoning": the ability to use relational reasoning to design novel interventions and bring about novel outcomes. In two experiments, we show that 24-30-month-old toddlers and three-year-old preschoolers use relational reasoning in a causal problem-solving task. Even toddlers rapidly inferred relational causal rules and applied this knowledge to solve novel problems--thus demonstrating both surprisingly early competence in relational reasoning and sophisticated causal inference. In both experiments, children observed a handful of trials in which a mechanistically opaque machine made objects larger or smaller. When prompted to solve a new problem, they used the machine to change the relative size of a novel object - even though its appearance and absolute size differed from previous observations, and even though subjects had never seen the machine generate objects of the required size before. This suggests that children quickly inferred abstract causal relations and then generalized these relations to determine which intervention would bring about the novel outcome required to solve the problem. These findings suggest a close link between early relational reasoning and active causal learning and inference.

Decisions about how to divide resources have profound social and practical consequences. Do explanations regarding the source of existing inequalities influence how children and adults allocate new resources? When 3- to 6-year-old children (N = 201) learned that inequalities were caused by structural forces (stable external constraints affecting access to resources) as opposed to internal forces (effort), they rectified inequalities, overriding previously documented tendencies to perpetuate inequality or divide resources equally. Adults (N = 201) were more likely than children to rectify inequality spontaneously; this was further strengthened by a structural explanation but reversed by an effort-based explanation. Allocation behaviors were mirrored in judgments of which allocation choices by others were appropriate. These findings reveal how explanations powerfully guide social reasoning and action from childhood through adulthood.

Abstract To learn from metaphors, children must not only understand metaphors, but also appreciate their informativeness. Metaphors based on abstract commonalities (e.g., functional metaphors like “Eyes are windows”) typically have greater potential to generate novel inferences, and thus generally allow for more learning, than metaphors based on more superficial commonalities (e.g., perceptual metaphors like “Eyes are buttons”). However, previous research shows that preschoolers prefer perceptual metaphors over functional metaphors. Consequently, we ask whether providing additional context can shift preschoolers’ and adults’ metaphor preferences. In Experiment 1, pedagogical context increases preferences for functional metaphors in adults, but not preschoolers. In Experiment 2, providing explanations for similarities between the concepts in metaphors increases preschoolers’ preferences for functional metaphors. In Experiment 3, preschoolers differentiate between functional and perceptual metaphors when explicitly asked to communicate functional or perceptual information. These findings suggest that providing explanations for metaphors helps preschoolers appreciate their informativeness.

Early childhood researchers frequently use learning materials and assessments involving pictures, across different cultures and contexts. However, there is variation in when and how children across cultures and contexts begin to understand and learn from pictures. While children growing up in high-income contexts often have more experience with picture books and other kinds of two-dimensional visual symbols, children growing up in low-income, rural contexts in low- and middle-income countries often have less experience with pictures and other kinds of visual symbols. The current research leverages variation in picture experience within a geographical region to investigate whether previous picture experience is related to toddlers' (1) performance on a picture-based word learning task, and (2) referential understanding, controlling for maternal education, number of toys, caregiver talk, and caregiver play. One hundred and twenty-eight toddlers in urban and rural western Kenya (n = 64 per area), who had varying amounts of picture experience, participated in a picture-based word learning task. Preregistered analyses with the entire sample showed no relation between picture experience and performance on a picture-based word learning task, or between picture experience and referential understanding. However, exploratory analyses found a positive association between picture experience and performance on the picture-based word learning task in the urban sample, but not the rural sample. We found no association between toddlers' referential understanding and picture experience, in either sample. We discuss how these results may inform the efficacy of learning materials and the validity of assessments used with children from diverse global backgrounds.

Abstract Previous work suggests that preschoolers often misunderstand metaphors. However, some recent studies demonstrate that preschoolers can represent abstract relations, suggesting that the cognitive foundations of metaphor comprehension may develop earlier than previously believed. The present experiments used novel paradigms to explore whether preschoolers (N = 200; 4–5 years; 100 males, 100 females; predominantly White) can understand metaphors based on abstract, functional similarities. In Experiment 1, preschoolers and adults (N = 64; 18–41 years; 25 males, 39 females; predominantly White) rated functional metaphors (e.g., “Roofs are hats”; “Tires are shoes”) as “smarter” than nonsense statements (e.g., “Boats are skirts”; “Pennies are sunglasses”) in a metalinguistic judgment task (d = .42 in preschoolers; d = 3.06 in adults). In Experiment 2, preschoolers preferred functional explanations (e.g., “Both keep you dry”) over perceptual explanations (e.g., “Both have pointy tops”) when interpreting functional metaphors (e.g., “Roofs are hats”) (d = .99). In Experiment 3, preschoolers preferred functional metaphors (e.g., “Roofs are hats”) over nonsense statements (e.g., “Roofs are scissors”) when prompted to select the “better” utterance (d = 1.25). Moreover, over a quarter of preschoolers in Experiment 1 and half of preschoolers in Experiment 3 explicitly articulated functional similarities when justifying their responses, and the performance of these subsets of children drove the success of the entire sample in both experiments. These findings demonstrate that preschoolers can understand metaphors based on abstract, functional similarities.

Although play is common in many cultures and species, existing theories of play ultimately account for only small subsets of all play behaviors. Here, we contrast play with two well-understood forms of behavior—exploration and exploitation—and investigate the hypothesis that at least some forms of play have objectives that are distinct from both information gain and reward. In both a foraging task reminiscent of classical decision-making studies and a more naturalistic setting (interacting with a commonly available playset), we analyzed US children’s behavior when given a prompt guiding them to explore, exploit, play, or do whatever they wanted to do. We found in both experiments that children’s behavior differed significantly and systematically across conditions. These results suggest that play does not merely involve seeking information or reward but is a distinct mode of behavior with other motivations. In particular, when they were instructed to play, children preferentially generated new goals and attempted to accomplish them.

Almost all of human infants’ experience and learning takes place in the context of caregiving relationships. This paper considers how infants understand the care they receive. We begin by outlining plausible features of an “intuitive theory” of care. On this intuitive theory, caregiving has both a distinctive foundational structure and distinctive features that differentiate it from other social relationships. We then review methods and findings from research on infants’ understanding of people and social relationships. We propose that even before infants can use language, they may understand caregiving as an abstract intuitive theory with some features in common with the adult theory. In particular, infants understand care relationships as intimate, altruistic, and asymmetric. We review work that starts to shed light on this proposal, including that infants distinguish between intimate relationships and merely positive ones, and that they have asymmetric expectations of responses to distress in intimate relationships that occur between large and small individuals. The proposal that infants can make these inferences has societal and political implications for how we structure caregiving in early life.

Many childhood assessments rely on picture stimuli, but children in diverse early environments possess varying amounts of experience with pictures. Two preregistered experiments, conducted in 2022-2023, investigated whether picture assessments are valid across diverse contexts. Children from Mombasa County, Kenya in their first month of formal schooling (n = 192; 2-7 years; 85 females; all Black) performed more accurately on an object vocabulary task than a picture vocabulary task (β = .07, p < .001) (Experiment 1). Children from the San Francisco Bay Area (n = 96; 2-3 years; 52 females; predominantly White and Asian) performed similarly on object and picture vocabulary tasks (β = .02, p = .60) (Experiment 2). Consequently, assessments involving pictures may underestimate children’s capacities, in some contexts.

Decisions about how to divide resources have profound social and practical consequences. Do explanations regarding the source of existing inequalities influence how children and adults allocate new resources? When 3-6-year-old children (N=201) learned that inequalities were caused by structural forces (stable external constraints affecting access to resources) as opposed to internal forces (effort), they rectified inequalities, overriding previously-documented tendencies to perpetuate inequality or divide resources equally. Adults (N=201) were more likely than children to rectify inequality spontaneously; this was further strengthened by a structural explanation but reversed by an effort-based explanation. Allocation behaviors were mirrored in judgments of which allocation choices by others were appropriate. These findings reveal how explanations powerfully guide social reasoning and action from childhood through adulthood.

In this work we investigate how children ages 5-12 perceive, understand, and use generative AI models such as a text-based LLMs ChatGPT and a visual-based model DALL-E. Generative AI is newly being used widely since chatGPT. Children are also building mental models of generative AI. Those haven't been studied before and it is also the case that the children's models are dynamic as they use the tools, even with just very short usage. Upon surveying and experimentally observing over 40 children ages 5-12, we found that children generally have a very positive outlook towards AI and are excited about the ways AI may benefit and aid them in their everyday lives. In a forced choice, children robustly associated AI with positive adjectives versus negative ones. We also categorize what children are querying AI models for and find that children search for more imaginative things that don't exist when using a visual-based AI and not when using a text-based one. Our follow-up study monitored children's responses and feelings towards AI before and after interacting with GenAI models. We even find that children find AI to be less scary after interacting with it. We hope that these findings will shine a light on children's mental models of AI and provide insight for how to design the best possible tools for children who will inevitably be using AI in their lifetimes. The motivation of this work is to bridge the gap between Human-Computer Interaction (HCI) and Psychology in an effort to study the effects of AI on society. We aim to identify the gaps in humans' mental models of what AI is and how it works. Previous work has investigated how both adults and children perceive various kinds of robots, computers, and other technological concepts. However, there is very little work investigating these concepts for generative AI models and not simply embodied robots or physical technology.

This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 4,300 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children (ages three to five) and to adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while GPT-o1, GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-o1, performs better in tasks involving simple surface-level visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of extrinsic spatial properties in the physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.

Three preregistered experiments, conducted in 2021, investigated whether English-speaking American preschoolers (N = 120; 4-6 years; 54 females, predominantly White) and adults (N = 80; 18-52 years; 59 females, predominantly Asian) metonymically extend owners' names to owned objects-an extension not typically found in English. In Experiment 1, 5- and 6-year-olds and adults extended names to owned objects over duplicates (d = 0.34 in children; d = 1.13 in adults). In Experiment 2, 5- and 6-year-olds and adults extended names to owned over borrowed objects (d = 1.37 in children; d = 4.34 in adults). Experiment 3 replicated this finding with 4-year-olds (d = 0.43). Thus, English-speaking preschoolers can acquire semantic generalizations, even those not present in their language.

What drives children to explore and learn when external rewards are uncertain or absent? Across three studies, we tested whether information gain itself acts as an internal reward and suffices to motivate children's actions. We measured 24-56-month-olds' persistence in a game where they had to search for an object (animal or toy), which they never find, hidden behind a series of doors, manipulating the degree of uncertainty about which specific object was hidden. We found that children were more persistent in their search when there was higher uncertainty, and therefore, more information to be gained with each action, highlighting the importance of research on artificial intelligence to invest in curiosity-driven algorithms. RESEARCH HIGHLIGHTS: Across three studies, we tested whether information gain itself acts as an internal reward and suffices to motivate preschoolers' actions. We measured preschoolers' persistence when searching for an object behind a series of doors, manipulating the uncertainty about which specific object was hidden. We found that preschoolers were more persistent when there was higher uncertainty, and therefore, more information to be gained with each action. Our results highlight the importance of research on artificial intelligence to invest in curiosity-driven algorithms.

Abstract Caring for the young and the old, the fragile and the ill, is central to human thriving, and has played a fundamental role in human evolution. Yet care has been largely invisible in political economy and it does not fit the prevailing philosophical, political, and economic frameworks. Care typically emerges in the context of close personal relationships, and it is not well suited to either utilitarian or Kantian accounts of morality, or to “social contract” accounts of cooperation. Markets and states both have difficulty providing and supporting care, and as a result, care is overlooked and undervalued. I sketch alternative ways of thinking about the morality and politics of care and present alternative policies that could help support carers and those they care for.

Although adults use metaphors to guide their thinking and reasoning, less is known about whether metaphors might facilitate cognition earlier in development. Previous research shows that preschoolers understand metaphors, but less is known about whether preschoolers can learn from metaphors. The current preregistered experiment investigated whether adults ( n = 64) and 3- and 4-year-olds ( n = 128) can use metaphors to make new inferences. In a between-subjects design, participants heard information about novel artifacts, conveyed through either only positive metaphors (e.g., “Daxes are suns”) or positive and negative metaphors (e.g., “Daxes are suns. Daxes are not clouds.”). In both conditions, participants of all ages successfully formed metaphor-consistent inferences about abstract, functional features of the artifacts (e.g., that daxes light up rather than let out water). Moreover, participants frequently provided explanations appealing to the metaphors when justifying their responses. Consequently, metaphors may be a powerful learning mechanism from early childhood onward.

Much discussion about large language models and language-and-vision models has focused on whether these models are intelligent agents. We present an alternative perspective. First, we argue that these artificial intelligence (AI) models are cultural technologies that enhance cultural transmission and are efficient and powerful imitation engines. Second, we explore what AI models can tell us about imitation and innovation by testing whether they can be used to discover new tools and novel causal structures and contrasting their responses with those of human children. Our work serves as a first step in determining which particular representations and competences, as well as which kinds of knowledge or skill, can be derived from particular learning techniques and data. In particular, we explore which kinds of cognitive capacities can be enabled by statistical analysis of large-scale linguistic data. Critically, our findings suggest that machines may need more than large-scale language and image data to allow the kinds of innovation that a small child can produce.

Board, card, or video games have been played by virtually every individual in the world population, with both children and adults participating. Games are popular because they are intuitive and fun. These distinctive qualities of games also make them ideal as a platform for studying the mind. By being intuitive, games provide a unique vantage point for understanding the inductive biases that support behavior in more complex, ecological settings than traditional lab experiments. By being fun, games allow researchers to study new questions in cognition such as the meaning of "play'' and intrinsic motivation, while also supporting more extensive and diverse data collection by attracting many more participants. We describe both the advantages and drawbacks of using games relative to standard lab-based experiments and lay out a set of recommendations on how to gain the most from using games to study cognition. We hope this article will lead to a wider use of games as experimental paradigms, elevating the ecological validity, scale, and robustness of research on the mind.

Much discussion about large language models and language-and-vision models has focused on whether these models are intelligent agents. We present an alternative perspective. We argue that these artificial intelligence models are cultural technologies that enhance cultural transmission in the modern world, and are efficient imitation engines. We explore what AI models can tell us about imitation and innovation by evaluating their capacity to design new tools and discover novel causal structures, and contrast their responses with those of human children. Our work serves as a first step in determining which particular representations and competences, as well as which kinds of knowledge or skill, can be derived from particular learning techniques and data. Critically, our findings suggest that machines may need more than large scale language and images to achieve what a child can do.

Play is important in many cultures and species, but the basic motivations behind play remain unclear. In two preregistered experiments, we presented 5-to-10-year-old children (n = 124) with a novel game. We compared their choices about how to best configure the game when either playing for fun or playing to win, exploring: 1) how sensitive children are to the factors that are most relevant for playing to win; and 2) how their preferences when playing for fun relates to their preferences when playing to win. We found that for “win-relevant” features, children chose easier settings when playing to win than when playing for fun. By contrast, for “win-irrelevant” features, children generally preferred similar settings across conditions. Children also judged “win-relevant” factors as more important to winning than “win-irrelevant” factors, and judged both as irrelevant to having fun. These results suggest that playing to win and playing for fun are distinct motivational contexts to which children can appropriately adapt their decisions during play.

Early childhood researchers frequently use learning materials and assessments involving pictures, across different cultures and contexts. However, there is variation in when and how children across cultures and contexts begin to understand and learn from pictures. While children growing up in high-income contexts often have more experience with picture books and other kinds of two-dimensional visual symbols, children growing up in low-income, rural contexts in low- and middle-income countries often have less experience with pictures and other kinds of visual symbols. The current research leverages variation in picture experience within a geographical region to investigate whether previous picture experience is related to toddlers’ 1) performance on a picture-based word learning task, and 2) referential understanding, controlling for maternal education, number of toys, caregiver talk, and caregiver play. 128 toddlers in urban and rural western Kenya (n = 64 per area), who had varying amounts of picture experience, participated in a picture-based word learning task. Preregistered analyses with the entire sample showed no relation between picture experience and performance on a picture-based word learning task, or between picture experience and referential understanding. However, exploratory analyses found a positive association between picture experience and performance on the picture-based word learning task in the urban sample, but not the rural sample. We found no association between toddlers’ referential understanding and picture experience, in either sample. We discuss how these results may inform the efficacy of learning materials and validity of assessments used with children from diverse global backgrounds.

Emotions are closely and intricately connected to intentions and desires in our adult understanding of mind. In full-fledged adult psychology, an important feature of desires and intentions is that they are “directed at” objects and events. In fact, in adult psychology this is one thing that distinguishes desires and intentions from mere moods, feelings, or emotions and makes them similar to perceptions and beliefs. Desires carry with them an implication of certain positive or negative attitudes towards objects and events. Various forms of emotional contagion and/or simple associative processes could be at work in these studies, as argued by advocates of a “lean interpretation” of the standard social referencing work. Toddlers are systematically testing the dimensions on which their desires and the desires of others may be in conflict. Taken together, the recent procedural advances in addressing questions to very young children are starting to reveal some of what children understand in the dark ages.

Much discussion about large language models and language-and-vision models has focused on whether these models are intelligent agents. We present an alternative perspective. We argue that these artificial intelligence models are cultural technologies that enhance cultural transmission in the modern world, and are efficient imitation engines. We explore what AI models can tell us about imitation and innovation by evaluating their capacity to design new tools and discover novel causal structures, and contrast their responses with those of human children. Our work serves as a first step in determining which particular representations and competences, as well as which kinds of knowledge or skill, can be derived from particular learning techniques and data. Critically, our findings suggest that machines may need more than large scale language and images to achieve what a child can do.

Developmental psychologists have spent decades devising experiments to test the intelligence and knowledge of infants and children, tracing the origin of crucial concepts and capacities. Moreover, experimental techniques in developmental psychology have been carefully designed to discriminate the cognitive capacities that underlie particular behaviors. We propose that using classical experiments from child development is a particularly effective way to probe the computational abilities of AI models, in general, and LLMs in particular. First, the methodological techniques of developmental psychology, such as the use of novel stimuli to control for past experience or control conditions to determine whether children are using simple associations, can be equally helpful for assessing the capacities of LLMs. In parallel, testing LLMs in this way can tell us whether the information that is encoded in text is sufficient to enable particular responses, or whether those responses depend on other kinds of information, such as information from exploration of the physical world. In this work we adapt classical developmental experiments to evaluate the capabilities of LaMDA, a large language model from Google. We propose a novel LLM Response Score (LRS) metric which can be used to evaluate other language models, such as GPT. We find that LaMDA generates appropriate responses that are similar to those of children in experiments involving social understanding, perhaps providing evidence that knowledge of these domains is discovered through language. On the other hand, LaMDA's responses in early object and action understanding, theory of mind, and especially causal reasoning tasks are very different from those of young children, perhaps showing that these domains require more real-world, self-initiated exploration and cannot simply be learned from patterns in language input.

How does creativity develop? Creativity is a multi-faceted behavior and thus it is difficult to find measures for creativity that are both precise and comparable across development. Here, we examine the development of creativity using a "creative foraging" task. The task measures different facets of creativity which we compare between 4- to 8-year-old children and adults. We find that compared to adults, children spend a higher percentage of their search exploring, and their exploitation phases are less efficient. Moreover, children orient their search to a different and smaller region of the search space, but within that space they produce more unique creative products. Lastly, as children grow up, their creative products become more adult-like and their uniqueness decreases. Together, these results suggest that creative search changes across development, in the search strategy employed, in how the space of possibilities is navigated, and in what ideas are ultimately chosen.

Technological advances are enabling roles for machines that present novel ethical challenges. The study of 'AI ethics' has emerged to confront these challenges, and connects perspectives from philosophy, computer science, law, and economics. Less represented in these interdisciplinary efforts is the perspective of cognitive science. We propose a framework - computational ethics - that specifies how the ethical challenges of AI can be partially addressed by incorporating the study of human moral decision-making. The driver of this framework is a computational version of reflective equilibrium (RE), an approach that seeks coherence between considered judgments and governing principles. The framework has two goals: (i) to inform the engineering of ethical AI systems, and (ii) to characterize human moral judgment and decision-making in computational terms. Working jointly towards these two goals will create the opportunity to integrate diverse research questions, bring together multiple academic communities, uncover new interdisciplinary research topics, and shed light on centuries-old philosophical questions.

For more than three decades, researchers have characterized the dramatic changes in early cognitive development and the learning mechanisms that underlie those changes by analogy to the thinking of professional scientists. This "child-as-scientist" view has emphasized the parallels between: (1) the evidence-based, theoretical nature of both children's and scientists' knowledge, (2) the rational process by which that knowledge is updated and revised, and (3) conceptual change, the often radical alterations to epistemic content that can result from those revisions. In this chapter, we begin by laying out the fundamentals of scientific thinking and reasoning, situating it in an "interventionist" framework of causal reasoning. Next, we review the history of the "child-as-scientist" approach in this context. Then, we outline recent work, open questions, and future directions in research on the development of scientific thinking in infancy, early childhood, and beyond.

Abstract Pretend play universally emerges during early childhood and may support the development of causal inference and counterfactual reasoning. However, the amount of time spent pretending, the value that adults place on pretence and the scaffolding adults provide vary by both culture and socioeconomic status (SES). In middle class U.S. preschoolers, accuracy on a pretence-based causal reasoning task predicted performance on a similar causal counterfactual task. We explore the relationship between cultural environment, pretence and counterfactual reasoning in low-income Peruvian (N = 62) and low-income U.S. (N = 57) 3- to 4-year olds, and contrast findings against previous findings in an age-matched, mixed-SES U.S. sample (N = 60). Children learned a novel causal relationship, then answered comparable counterfactual and pretence-based questions about the relationship. Children's responses for counterfactual and pretence measures differed across populations, with Peruvian and lower-income U.S. children providing fewer causally consistent responses when compared with middle class U.S. children. Nevertheless, correlations between the two measures emerged in all populations. Across cohorts, children also provided more causally consistent answers during pretence than counterfactually. Our findings strengthen the hypothesis that causal pretend play is related to causal counterfactual reasoning across cultural contexts, while also suggesting a role for systematic environmental differences. This article is part of the theme issue ‘Thinking about possibilities: mechanisms, ontogeny, functions and phylogeny’.

Identifying abstract relations is essential for commonsense reasoning. Research suggests that even young children can infer relations such as "same" and "different," but often fail to apply these concepts. Might the process of explaining facilitate the recognition and application of relational concepts? Based on prior work suggesting that explanation can be a powerful tool to promote abstract reasoning, we predicted that children would be more likely to discover and use an abstract relational rule when they were prompted to explain observations instantiating that rule, compared to when they received demonstration alone. Five- and 6-year-olds were given a modified Relational Match to Sample (RMTS) task, with repeated demonstrations of relational (same) matches by an adult. Half of the children were prompted to explain these matches; the other half reported the match they observed. Children who were prompted to explain showed immediate, stable success, while those only asked to report the outcome of the pedagogical demonstration did not. Findings provide evidence that explanation facilitates early abstraction over and above demonstration alone.

chapter Share on Causal Models and Cognitive Development Author: Alison Gopnik University of California at Berkeley University of California at BerkeleySearch about this author Authors Info & Claims Probabilistic and Causal Inference: The Works of Judea PearlFebruary 2022 Pages 593–604https://doi.org/10.1145/3501714.3501746Online:04 March 2022Publication History 0citation7DownloadsMetricsTotal Citations0Total Downloads7Last 12 Months7Last 6 weeks1 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my Alerts New Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

How does creativity develop? Creativity is a multi-faceted behavior and thus it is difficult to find measures for creativity that are both precise and comparable across development. Here, we examine the development of creativity using a “creative foraging” task. The task measures different facets of creativity which we compare between 4- to 8-year-old children and adults. We find that compared to adults, children spend a higher percentage of their search exploring, and their exploitation phases are less efficient. Moreover, children orient their search to a different and smaller region of the search space, but within that space they produce more unique creative products. Lastly, as children grow up, their creative products become more adult-like and their uniqueness decreases. Together, these results suggest that creative search changes across development, in the search strategy employed, in how the space of possibilities is navigated, and in what ideas are ultimately chosen.

To learn from metaphors, children must not only understand metaphors, but also appreciate their informativeness. Metaphors based on abstract commonalities (e.g., functional metaphors like “Eyes are windows”) typically have greater potential to generate novel inferences, and thus generally allow for more learning, than metaphors based on more superficial commonalities (e.g., perceptual metaphors like “Eyes are buttons”). However, previous research shows that preschoolers prefer perceptual metaphors over functional metaphors. Consequently, we ask whether providing additional context can shift preschoolers’ and adults’ metaphor preferences. In Experiment 1, pedagogical context increases preferences for functional metaphors in adults, but not preschoolers. In Experiment 2, providing explanations for similarities between the concepts in metaphors increases preschoolers’ preferences for functional metaphors. In Experiment 3, preschoolers differentiate between functional and perceptual metaphors when explicitly asked to communicate functional or perceptual information. These findings suggest that providing explanations for metaphors help preschoolers appreciate their informativeness.

Despite recent progress in reinforcement learning (RL), RL algorithms for exploration still remain an active area of research. Existing methods often focus on state-based metrics, which do not consider the underlying causal structures of the environment, and while recent research has begun to explore RL environments for causal learning, these environments primarily leverage causal information through causal inference or induction rather than exploration. In contrast, human children - some of the most proficient explorers - have been shown to use causal information to great benefit. In this work, we introduce a novel RL environment designed with a controllable causal structure, which allows us to evaluate exploration strategies used by both agents and children in a unified environment. In addition, through experimentation on both computation models and children, we demonstrate that there are significant differences between information-gain optimal RL exploration in causal environments and the exploration of children in the same environments. We conclude with a discussion of how these findings may inspire new directions of research into efficient exploration and disambiguation of causal structures for RL algorithms.

Recent work in machine learning and cognitive science has suggested that understanding causal information is essential to the development of intelligence. The extensive literature in cognitive science using the ``blicket detector'' environment shows that children are adept at many kinds of causal inference and learning. We propose to adapt that environment for machine learning agents. One of the key challenges for current machine learning algorithms is modeling and understanding causal overhypotheses: transferable abstract hypotheses about sets of causal relationships. In contrast, even young children spontaneously learn and use causal overhypotheses. In this work, we present a new benchmark -- a flexible environment which allows for the evaluation of existing techniques under variable causal overhypotheses -- and demonstrate that many existing state-of-the-art methods have trouble generalizing in this environment. The code and resources for this benchmark are available at https://github.com/CannyLab/casual_overhypotheses.

We investigate individual, developmental, and cultural differences in self-control in relation to children's changing belief in "free will" - the possibility of acting against and inhibiting strong desires. In three studies, 4- to 8-year-olds in the U.S., China, Singapore, and Peru (N = 441) answered questions to gauge their belief in free will and completed a series of self-control and inhibitory control tasks. Children across all four cultures showed predictable age-related improvements in self-control, as well as changes in their free will beliefs. Cultural context played a role in the timing of these emerging free will beliefs: Singaporean and Peruvian children's beliefs changed at later ages than Chinese and U.S. children. Critically, culture moderated the link between self-control abilities and free will beliefs: Individual differences in self-control behaviors were linked to individual differences in free will beliefs in U.S. children, but not in children from China, Singapore or Peru. There was also evidence of a causal influence of self-control performance on free will beliefs in our U.S. sample. In Study 2, a randomly assigned group of U.S. 4- and 5-year-olds who failed at two self-control tasks showed reduced belief in free will, but a group of children who completed free will questions first did not show changes to self-control. Together these results suggest that culturally-acquired causal-explanatory frameworks for action, along with observations of one's own abilities, might influence children's emerging understanding of free will.

Intuitively, children appear to be more exploratory than adults, and this exploration seems to help children learn,. However, there have been few clear tests of these ideas. We test whether exploration and learning change across development using a task that presents a "learning trap." In this task, exploitation-maximizing immediate reward and avoiding costs-may lead the learner to draw incorrect conclusions, while exploration may lead to better learning but be more costly. In Studies 1, 2, and 3 we find that preschoolers and early school-aged children explore more than adults and learn the true structure of the environment better. Study 3 demonstrates that children explore more than adults even though they, like adults, predict that exploration will be costly, and it shows that exploration and learning are correlated. Study 4 shows that children's and adults' learning depends on the evidence they generate during exploration: children exposed to adult-like evidence learn like adults, and adults exposed to child-like evidence learn like children. Together, these studies support the idea that children may be more exploratory than adults, and this increased exploration influences learning.

model, which represents the hypothesis space based on the relative speed of individual monsters. The results of Study 1 (26 five-year-olds, 14 female and 25 six-year-olds, 15 female; predominantly White, fluent in English) provided the first evidence that 5- and 6-year-olds can use cue-abstraction hypothesis space representations when provided with scaffolding. However, Study 2 (65 five-year-olds, 33 female; 67 six-year-olds, 33 female; 68 seven-year-olds, 33 female; predominantly White, fluent in German) showed that young children were best described by the permutation-based model, and that only 7-year-olds, when provided with memory aids, were best captured by the cue-abstraction model. Overall, our results highlight the guiding role of the hypothesis space for active search and learning, suggesting that these two phases might trigger different representations, and indicating a developmental shift in how children represent the hypothesis space. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Humans spend much of their lives engaging with their internal train of thoughts. Traditionally, research focused on whether or not these thoughts are related to ongoing tasks, and has identified reliable and distinct behavioral and neural correlates of task-unrelated and task-related thought. A recent theoretical framework highlighted a different aspect of thinking-how it dynamically moves between topics. However, the neural correlates of such thought dynamics are unknown. The current study aimed to determine the electrophysiological signatures of these dynamics by recording electroencephalogram (EEG) while participants performed an attention task and periodically answered thought-sampling questions about whether their thoughts were 1) task-unrelated, 2) freely moving, 3) deliberately constrained, and 4) automatically constrained. We examined three EEG measures across different time windows as a function of each thought type: stimulus-evoked P3 event-related potentials and non-stimulus-evoked alpha power and variability. Parietal P3 was larger for task-related relative to task-unrelated thoughts, whereas frontal P3 was increased for deliberately constrained compared with unconstrained thoughts. Frontal electrodes showed enhanced alpha power for freely moving thoughts relative to non-freely moving thoughts. Alpha-power variability was increased for task-unrelated, freely moving, and unconstrained thoughts. Our findings indicate distinct electrophysiological patterns associated with task-unrelated and dynamic thoughts, suggesting these neural measures capture the heterogeneity of our ongoing thoughts.

One of the greatest challenges for artificial intelligence is how to behave adaptively in scenarios with uncertain or no rewards. One---and perhaps the only---way to approach such complex learning problems is to build simple algorithms that grow into sophisticated adaptive agents, just like children do. But what drives children to explore and learn when external rewards are absent? Across three studies, we tested whether information gain itself acts as an internal reward and motivates children's actions. We measured 24- to 56-month-olds’ persistence in a game where they had to search for an object (animal or toy), which they never find, hidden behind a series of doors, manipulating the degree of uncertainty about \emph{which specific object} was hidden. We found that children were more persistent in their search when there was higher uncertainty, and therefore more information to be gained with each action, highlighting the importance of research on artificial intelligence to invest in curiosity-driven algorithms.

= 125) children's questions using a combined qualitative and quantitative approach. Children were presented with a hierarchical version of the 20-questions game, in which they were given an array of objects that could be organized into three category levels based on shared features. We then tested whether it is possible to scaffold children's question-asking abilities without extensive training. In particular, we supported children's categorization performance by providing the object-related features needed to ask effective constraint-seeking questions. We found that with both age and scaffolding children asked more effective questions, targeting higher category levels and therefore reaching the solution with fewer questions. We discuss the practical and theoretical implications of these results.

This chapter discusses how the results of the studies bear on the specificity hypothesis, the notion that there are specific relationships between particular semantic and cognitive developments, rather than more general relationships between language and cognition. More significantly, we and others have shown that there are empirical relationships between object permanence skills and disappearance words. So far we have found relationships between disappearance words and object permanence, success/failure words and means-ends understanding, and the naming explosion and categorization. The chapter presents the results of studies investigating the sequencing of semantic and cognitive developments in the one-word stage. Our findings suggest that semantic rather than formal aspects of language development are related to cognitive development and that these semantic developments are related to specific cognitive achievements. A way of testing the specificity hypothesis is to look at the correlations between the age of acquisition of linguistic and cognitive skills.

Lake 2015 presented the Omniglot dataset to study how people generate and classify characters. They proposed a model for one-shot-learning of new concepts based on inferring compositionally structured generative models, and could transfer from familiar concepts to new ones. Their Bayesian-Programming-Language model was able to both classify and generate characters similar to adults. However, adults have years of experience, would a similar model apply to children without this prior knowledge? We introduce a new dataset called Omniglot_Jr., composed of Omniglot letters generated by children aged 3-6. Our results of training BPL with children's data in classification and generation find that BPL achieves higher classification accuracy and generates more adult looking letters. We propose the challenge of reproducing children's distinctive pattern of mistakes. Skills such as character recognition depend on child-like learning; this challenge should help us understand how that learning is possible, and how to simulate it in computational systems.

Adults have been shown to favor stable causal relationships – those that hold robustly across background contexts – in their actions and causal/explanatory generalizations (Vasilyeva et al, 2018). Here we explore how this preference develops. We present results from one study with 141 4-7-year-olds investigating whether children pay attention to causal stability when they explain observations and design interventions in novel contexts. We report developmental shifts in reliance on causal stability in a range of inferential tasks, highlight the important role of perceived average causal strength in determining children’s causal preferences, and discuss the implications of our findings for theories of early causal learning. To our knowledge, this is the first study exploring the role of stability in children’s causal reasoning.

Abstract Self-control is quite difficult—sometimes people are successful, but frequently they are not. So why do people believe that they can choose, by their own free will, to exercise self-control? This chapter summarizes recent research exploring the cultural and developmental origins of beliefs about self-control and free will. It discusses how two factors contribute to the development of children’s beliefs about self-control: culture and first-person experiences. The authors’ studies of four- to eight-year-old children (N = 441; mean age = 5.96 years; range = 3.92–8.90 years) from China, Singapore, Peru, and the United States indicate that self-control beliefs differ across cultures, and that, comparatively, US children hold intuitions that they can freely choose to exercise self-control. Additionally, evidence indicates that the experience of self-control failure impacts beliefs about free will in US children, but that these experience effects are not culturally universal.

This special issue focuses on the relationship between life history and learning, especially during human evolution. 'Life history' refers to the developmental programme of an organism, including its period of immaturity, reproductive rate and timing, caregiving investment and longevity. Across many species an extended childhood and high caregiving investment appear to be correlated with particular kinds of plasticity and learning. Human life history is particularly distinctive; humans evolved an exceptionally long childhood and old age, and an unusually high level of caregiving investment, at the same time that they evolved distinctive capacities for cognition and culture. The contributors explore the relations between life history, plasticity and learning across a wide range of methods and populations, including theoretical and empirical work in biology, anthropology and developmental psychology. This article is part of the theme issue 'Life history and learning: how childhood, caregiving and old age shape cognition and culture in humans and other animals'.

I argue that the evolution of our life history, with its distinctively long, protected human childhood, allows an early period of broad hypothesis search and exploration, before the demands of goal-directed exploitation set in. This cognitive profile is also found in other animals and is associated with early behaviours such as neophilia and play. I relate this developmental pattern to computational ideas about explore-exploit trade-offs, search and sampling, and to neuroscience findings. I also present several lines of empirical evidence suggesting that young human learners are highly exploratory, both in terms of their search for external information and their search through hypothesis spaces. In fact, they are sometimes more exploratory than older learners and adults. This article is part of the theme issue 'Life history and learning: how childhood, caregiving and old age shape cognition and culture in humans and other animals'.

Previous research suggests that preschoolers struggle with understanding abstract relations and with reasoning by analogy. Four experiments find, in contrast, that 3- and 4-year-olds (N = 168) are surprisingly adept at relational and analogical reasoning within a causal context. In earlier studies preschoolers routinely favored images that share thematic or perceptual commonalities with a target image (object matches) over choices that match the target along abstract relations (relational matches). The present studies embed such choice tasks within a cause-and-effect framework. Without causal framing, preschoolers strongly favor object matches, replicating the results of previous studies. But with causal framing, preschoolers succeed at analogical transfer (i.e., choose relational matches). These findings suggest that causal framing facilitates early analogical reasoning.

Although mind-wandering research is rapidly progressing, stark disagreements are emerging about what the term "mind-wandering" means. Four prominent views define mind-wandering as (a) task-unrelated thought, (b) stimulus-independent thought, (c) unintentional thought, or (d) dynamically unguided thought. Although theorists claim to capture the ordinary understanding of mind-wandering, no systematic studies have assessed these claims. Two large factorial studies present participants (N = 545) with vignettes that describe someone's thoughts and ask whether her mind was wandering, while systematically manipulating features relevant to the four major accounts of mind-wandering. Dynamics explains between four and 40 times more variance in participants' mind-wandering judgments than other features. Our third study (N = 153) tests and supports a unique prediction of the dynamic framework-obsessive rumination contrasts with mind-wandering. Our final study (N = 277) used vignettes that resemble mind-wandering experiments. Dynamics had significant and large effects, while task-unrelatedness was nonsignificant. These results strongly suggest that the central feature of mind-wandering is its dynamics.

There has been extensive discussion about gender gaps in representation and career advancement in the sciences. However, psychological science itself has yet to be the focus of discussion or systematic review, despite our field's investment in questions of equity, status, well-being, gender bias, and gender disparities. In the present article, we consider 10 topics relevant for women's career advancement in psychological science. We focus on issues that have been the subject of empirical study, discuss relevant evidence within and outside of psychological science, and draw on established psychological theory and social-science research to begin to chart a path forward. We hope that better understanding of these issues within the field will shed light on areas of existing gender gaps in the discipline and areas where positive change has happened, and spark conversation within our field about how to create lasting change to mitigate remaining gender differences in psychological science.

To successfully navigate in an uncertain world, one has to learn the relationship between cues (e.g., symptoms) and an outcome (e.g., disease). During this learning, it is sometimes possible to actively manipulate the cue values, allowing one to test hypotheses about this relationship directly. Across two studies, we investigated how 5- to 7-year-olds select cue configurations when learning cue-outcome relationships and how they are guided by representations of the hypothesis space regarding these relationships. In our task, children selected which monster pairs to see running in a race, allowing them to learn how two cues (the color and shape of monsters) predicted the relative speed of the monsters; subsequently, they made predictions about the speed of new monsters. Using computational modeling, we compared several models in their ability to capture children’s responses. We found that young children’s search was most consistent with a model that assumed reliance on a hypothesis space represented in terms of the relative speed of individual monsters. However, when memory aids were provided during search, 7-year-olds were best described by a model that assumes reliance on a more efficient, high-level representation that organizes the hypothesis space based on abstracted cue-outcome relationships. Our results highlight the guiding role of hypothesis-space representations for search during learning, suggest that young children already spontaneously abstract hypothesis-space representations, and provide the first evidence for a shift between search and test in terms of the hypothesis-space structures on which children rely when navigating in an uncertain world.

To successfully navigate in an uncertain world, one has to learn the relationship between cues (e.g., symptoms) and an outcome (e.g., disease). During this learning, it is sometimes possible to actively manipulate the cue values, allowing one to test hypotheses about this relationship directly. Across two studies, we investigated how 5- to 7-year-olds select cue configurations when learning cue-outcome relationships and how they are guided by representations of the hypothesis space regarding these relationships. In our task, children selected which monster pairs to see running in a race, allowing them to learn how two cues (the color and shape of monsters) predicted the relative speed of the monsters; subsequently, they made predictions about the speed of new monsters. Using computational modeling, we compared several models in their ability to capture children’s responses. We found that young children’s search was most consistent with a model that assumed reliance on a hypothesis space represented in terms of the relative speed of individual monsters. However, when memory aids were provided during search, 7-year-olds were best described by a model that assumes reliance on a more efficient, high-level representation that organizes the hypothesis space based on abstracted cue-outcome relationships. Our results highlight the guiding role of hypothesis-space representations for search during learning, suggest that young children already spontaneously abstract hypothesis-space representations, and provide the first evidence for a shift between search and test in terms of the hypothesis-space structures on which children rely when navigating in an uncertain world.

Intuitively, children appear to be more exploratory than adults, and this exploration seems to help children learn, but there have been few clear tests of this idea. We test whether exploration and learning change across development using a task that presents a “learning trap.” In this task, exploitation—maximizing immediate reward and avoiding costs—may lead the learner to draw incorrect conclusions, while exploration may lead to better learning but be more costly. In Studies 1, 2, and 3 we find that preschoolers and school-aged children explore more than adults and learn the true structure of the environment better. Study 3 also demonstrates that children explore more than adults even though their predictions about the effects of exploration are similar to those of adults and they predict that exploration will be costly. It also shows that exploration and learning are correlated. Study 4 shows that children’s and adults’ learning depends on the evidence they generate during exploration: children exposed to adult-like evidence learned like adults, and adults exposed to child-like evidence learned like children. Together, these studies support the idea that children may be more exploratory than adults, and this increased exploration influences learning.

Previous work suggests that preschoolers often misunderstand metaphors. However, some recent studies demonstrate that preschoolers can represent abstract relations, suggesting that the cognitive foundations of metaphor comprehension may develop earlier than previously believed. The present experiments used novel paradigms to explore whether preschoolers (N = 200; 4-5 years; 100 males, 100 females; predominantly White) can understand metaphors based on abstract, functional similarities. In Experiment 1, preschoolers and adults (N = 64; 18-41 years; 25 males, 39 females; predominantly White) rated functional metaphors (e.g., “Roofs are hats”; “Tires are shoes”) as “smarter” than nonsense statements (e.g., “Boats are skirts”; “Pennies are sunglasses”) in a metalinguistic judgment task (d = .42 in preschoolers; d = 3.06 in adults). In Experiment 2, preschoolers preferred functional explanations (e.g., “Both keep you dry”) over perceptual explanations (e.g., “Both have pointy tops”) when interpreting functional metaphors (e.g., “Roofs are hats”) (d = .99). In Experiment 3, preschoolers preferred functional metaphors (e.g., “Roofs are hats”) over nonsense statements (e.g., “Roofs are scissors”) when prompted to select the “better” utterance (d = 1.25). Moreover, over a quarter of preschoolers in Experiment 1 and half of preschoolers in Experiment 3 explicitly articulated functional similarities when justifying their responses, and the performance of these subsets of children drove the success of the entire sample in both experiments. These findings demonstrate that preschoolers can understand metaphors based on abstract, functional similarities.

Research in developmental psychology consistently shows that children explore the world thoroughly and efficiently and that this exploration allows them to learn. In turn, this early learning supports more robust generalization and intelligent behavior later in life. While much work has gone into developing methods for exploration in machine learning, artificial agents have not yet reached the high standard set by their human counterparts. In this work we propose using DeepMind Lab (Beattie et al., 2016) as a platform to directly compare child and agent behaviors and to develop new exploration techniques. We outline two ongoing experiments to demonstrate the effectiveness of a direct comparison, and outline a number of open research questions that we believe can be tested using this methodology.

Human beings have a unique life course. We have evolved a much longer and more helpless childhood, and longer old age, than our closest primate relatives. Many adults are involved in caring for children, including not only mothers, but fathers, grandmothers and grandfathers, and unrelated "alloparents". Why are human children so helpless? Why do they need so much support? Why do humans, especially women, keep living long after they stop reproducing? The central idea in this issue is that these "life-history" characteristics help produce other well-known human advantages such as learning and exploration, cooperation and culture. Biologists, anthropologists and psychologists, studying a wide range of animals and cultures, suggest how childhood, caregiving, and learning might go together.

Previous work suggests that children’s ability to understand\nmetaphors emerges late in development. Researchers argue that\nchildren’s initial failure to understand metaphors is due to an\ninability to reason about shared relational structures between\nconcepts. However, recent work demonstrates that causal framing\nfacilitates preschoolers’ relational reasoning. Might causal framing\nalso facilitate preschoolers’ metaphor comprehension? In\nExperiment 1, we presented 128 4- to 5-year-olds with a novel\nmetaphor comprehension task, following a causal warm-up task,\ncontrol warm-up task, or no warm-up task. In the novel\ncomprehension task, preschoolers rated functional metaphors and\nnonsense statements as smart or silly, and provided explanations.\nPreschoolers ranked metaphors as “smarter” than nonsense\nstatements, and a quarter of preschoolers provided functional\nexplanations. There was no effect of warm-up tasks. In Experiment\n2, we validated the metaphor comprehension task with adults.\nOverall, the current work presents a new paradigm that\ndemonstrates preschoolers’ capacity to understand functional\nmetaphors.

Generic language (e.g., “Women are nurturing”; “Women donot like math”) is prominent in child-directed speech, and hasbeen shown to promote essentialist beliefs about the relevantkind, supporting stereotyping and prejudice. Here weinvestigate a theoretically-motivated intervention to break thelink between generics and essentialist assumptions. In a studywith 223 3-8-year-old children who learned about novel socialgroups from generic language, we demonstrate that a structuralconstrual of generics (attributing the category-propertyassociation to stable external constraints) mitigates essentialistassumptions about social categories. We discuss practicalapplications for reducing stereotype endorsement, andtheoretical implications regarding the meaning of genericlanguage and the development of social kind representations.

Contrary to decades of previous research, one recent study suggests that preschoolers can rapidly learn and transfer avariety of relational concepts (e.g., bigger than; smaller than; opposite of) when those relations are operationalized asthe beginning and ending states of causal transformations performed by agents (Goddu, Lombrozo, & Gopnik, in press).Without causal framing, children reverted from relational reasoning to object matching, consistent with findings from manyprevious (non-causal) paradigms. Here, we investigate whether three-year-olds (Experiment 1) and 24- to 30-month-oldtoddlers (Experiment 2) are able to learn and apply relational concepts in a behavioral task where they themselves mustintervene to solve a problem using relational reasoning. Results indicate that children as young as two years of age areable to rapidly learn and generalize relational concepts and use them productively to solve new problems.

Humans possess a remarkable capacity to create and understand abstract representations, such as pictures. U.S. 18-month-olds understand that pictures refer to objects; however, less is known about how this understanding develops. We testthe hypothesis that understanding the representational nature of pictures requires frequent experience with pictures, byworking with rural Kenyan toddlers with few visual symbols in their early environments.We taught rural Kenyan toddlers a novel word (dax) for a picture of a novel object. We then presented the picture and theobject to toddlers and asked them to point to the dax, reasoning that toddlers would select the object if they understoodthat pictures are representations of objects.Surprisingly, only half the sample learned the novel word. Moreover, the toddlers who learned the word selected be-tween the picture and object randomly. We discuss follow-up studies to continue exploring the development of pictorialcompetence across early environments.

Research in developmental psychology consistently shows that children explore the world thoroughly and efficiently andthat this exploration allows them to learn. While much work has gone into developing methods for exploration in machinelearning, artificial agents have not yet reached the standard set by their human counterparts. In this work we propose usingDeepMind Lab as a platform to directly compare child and agent behaviors and to develop new exploration techniques.We tested 60 children aged 4-6 examining two conditions that emulate how current reinforcement learning algorithmslearn using dense and sparse rewards and the children are then asked to find a goal in various mazes. These tasks providedata that can easily be compared to algorithms and we evaluate turn-by-turn moves the children do to what the Intrinsic-Curiosity-Module and Depth-First-Search algorithm do in the same exact maze. We show specifically where and whenchildren differ from the algorithms.

Novel causal systems pose a problem of variable choice: How can a reasoner decide which variable is causally relevant? Which variable in the system should a learner manipulate to try to produce a desired, yet unfamiliar, casual outcome? In much causal reasoning research, participants learn how a particular set of preselected variables produce a particular effect. Here, we investigate 3- to 5-year-olds' ability to select the relevant variable for intervention in a novel causal system. Results demonstrate that even young children can learn which variable is causally relevant from sparse evidence. In particular, children infer that variables that are "difference-making" in one causal system will also be relevant to other, novel, causal problems. If manipulating a causal variable in a particular way leads to one effect, children assume that other manipulations of that variable will lead to other novel effects. (PsycINFO Database Record (c) 2020 APA, all rights reserved).

Previously, research on wishful thinking has found that desires bias older children's and adults' predictions during probabilistic reasoning tasks. In this article, we explore wishful thinking in children aged 3- to 10-years-old. Do young children learn to be wishful thinkers? Or do they begin with a wishful thinking bias that is gradually overturned during development? Across five experiments, we compare low- and middle-income United States and Peruvian 3- to 10-year-old children (N = 682). Children were asked to make predictions during games of chance. Across experiments, preschool-aged children from all backgrounds consistently displayed a strong wishful thinking bias. However, the bias declined with age.

Constructing an intuitive theory from data confronts learners with a "chicken-and-egg" problem: The laws can only be expressed in terms of the theory's core concepts, but these concepts are only meaningful in terms of the role they play in the theory's laws; how can a learner discover appropriate concepts and laws simultaneously, knowing neither to begin with? We explore how children can solve this chicken-and-egg problem in the domain of magnetism, drawing on perspectives from computational modeling and behavioral experiments. We present 4- and 5-year-olds with two different simplified magnet-learning tasks. Children appropriately constrain their beliefs to two hypotheses following ambiguous but informative evidence. Following a critical intervention, they learn the correct theory. In the second study, children infer the correct number of categories given no information about the possible causal laws. Children's hypotheses in these tasks are explained as rational inferences within a Bayesian computational framework.

Constructing an intuitive theory from data confronts learners with a “chicken-and-egg” problem: the laws can only be expressed in terms of the theory’s core concepts, but these concepts are only meaningful in terms of the role they play in the theory’s laws; how can a learner discover appropriate concepts and laws simultaneously, knowing neither to begin with? We explore how children can solve this chicken-and-egg problem in the domain of magnetism, drawing on perspectives from computational modeling and behavioral experiments. We present four- and five-year-olds with two different simplified magnet-learning tasks. Children appropriately constrain their beliefs to two hypotheses following ambiguous but informative evidence. Following a critical intervention they learn the correct theory. In the second study, children infer the correct number of categories given no information about the possible causal laws. Children’s hypotheses in these tasks are explained as rational inferences within a Bayesian computational framework.

What drives human search in situations with sparse rewards? We let preschoolers (age 24-52 months) play a game in which they had to search for an animal and measured their persistence in the absence of explicit rewards. Crucially, we either told children which animal they were searching for (known animal condition), so that they gained no additional information when finding it, or we told them that it could be one out of eight different animals (unknown animal condition), so that they gained additional information when finding it. Although none of the children could actually find an animal, no matter how long they searched, our results indicate that the unknown animal group was more persistent in their search. These results enrich our understanding of children’s motivation in scenarios with sparse rewards and inform research on artificially intelligent agents facing similar environments.

Although mind-wandering research is rapidly progressing, stark disagreements are emerging about what the term “mind-wandering” means. Four prominent views define mind-wandering as 1) task-unrelated thought, 2) stimulus-independent thought, 3) unintentional thought, or 4) dynamically unguided thought. Although theorists claim to capture the ordinary understanding of mind-wandering, no systematic studies have assessed these claims. Two large factorial studies present participants (n=545) with vignettes that describe someone’s thoughts and ask whether her mind was wandering, while systematically manipulating features relevant to the four major accounts of mind-wandering. Dynamics explains between four and twenty times more variance in participants’ mind-wandering judgments than other features. Our third study (n=153) tests and supports a unique prediction of the dynamic framework—obsessive rumination contrasts with mind-wandering. Our final study (n=277) used vignettes that resemble mind-wandering experiments. Dynamics had significant and large effects, while task-unrelatedness was insignificant. These results strongly align with the dynamic conception of mind-wandering.

Previous research has shown that active engagement drives children’s remarkable learning capabilities. We investigated whether preschoolers are “ecological learners”, that is, whether they are able to select those active learning strategies that are most informative in a given task. Children had to choose which of two exploratory actions (open vs. shake) to perform to find an egg shaker hidden in one of four small boxes, contained in two larger boxes. Prior to this game, children either learnt that the egg was equally likely to be found in any of the four small boxes (Uniform condition), or that it was most likely to be found in one particular small box (Skewed condition). Results of Study 1 show that 3- and 4-year-olds successfully tailored their exploratory actions to the different likelihood distributions. Five-year-olds were equally likely to shake first, irrespective of condition, even when incentivized to shake only when needed (Study 2a). However, when the relevance of the frequency training for the hiding game was highlighted (Study 2b and Study 2c), the five-year-olds showed the same behavioural pattern as the younger preschoolers in Study 1. We suggest that ecological learning may be a key mechanism explaining how children can efficiently learn about the world around them.

Representations of social categories help us make sense of the social world, supporting predictions and explanations about groups and individuals. In an experiment with 156 participants, we explore whether children and adults are able to understand category-property associations (such as the association between "girls" and "liking pink") in structural terms, locating an object of explanation within a larger structure and identifying structural constraints that act on elements of the structure. We show that children as young as 3-4 years old show signs of structural thinking, and that 5-6-year-olds show additional differentiation between structural and nonstructural thinking, yet still fall short of adult performance. These findings introduce structural connections as a new type of nonaccidental relationship between a property and a category, and present a viable alternative to internalist accounts of social categories, such as psychological essentialism. (PsycINFO Database Record

= 205), we investigated preschool-aged children's ability to consider others' knowledge when selecting evidence in the service of teaching. Results from Experiment 1 revealed that 4-year-olds reliably selected evidence to rectify others' false beliefs, and provided causal explanations in their teaching, whereas 3-year-olds did not. In Experiment 2, we tie children's evidence selection abilities to theory of mind (ToM) development, above and beyond effects of age and numerical conservation abilities. In Experiment 3, we employed a 6-week training of children's pedagogical evidence selection with a new teaching task, and further explored the relationship between these skills and children's ToM abilities. We qualitatively replicated our results from Experiment 2 and report tentative evidence for a link between the pedagogical training and improvements in ToM. Together, our findings suggest important connections between reasoning about others' minds and evidential reasoning in natural pedagogy during early childhood. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Belief revision can occur at multiple levels of abstraction, including lower-level and higher-order beliefs. It remains unclear, however, how conflicting evidence interacts with prior beliefs to encourage higher-order belief revision. This study explores how 4- and 5-year-olds (N = 96) respond to evidence that directly conflicts with their causal higher-order beliefs. When shown a single event that directly violated a strongly supported prior belief, preschoolers largely maintained their initial higher-order belief. However, when the prior belief was more weakly supported and the counterevidence was stronger, children changed their minds. These findings indicate that young children can revise their higher-order beliefs and, furthermore, do so depending on the strength of both the evidence and their prior beliefs.

We propose a new dataset for evaluating question answering models with respect to their capacity to reason about beliefs. Our tasks are inspired by theory-of-mind experiments that examine whether children are able to reason about the beliefs of others, in particular when those beliefs differ from reality. We evaluate a number of recent neural models with memory augmentation. We find that all fail on our tasks, which require keeping track of inconsistent states of the world; moreover, the models' accuracy decreases notably when random sentences are introduced to the tasks at test. 1

The decision making processes of deep networks are difficult to understand and while their accuracy often improves with increased architectural complexity, so too does their opacity. Practical use of machine learning models, especially for question and answering applications, demands a system that is interpretable. We analyze the attention of a memory network model to reconcile contradictory performance on a challenging questionanswering dataset that is inspired by theoryof-mind experiments. We equate success on questions to task classification, which explains not only test-time failures but also how well the model generalizes to new training conditions.

Adults’ causal representations integrate information about predictive relations and the possibility of effective intervention; if one event reliably predicts another, adults can represent the possibility that acting to bring about the first event might generate the second. Here we show that although toddlers (mean age: 24 months) readily learn predictive relationships between physically connected events, they do not spontaneously initiate one event to try to generate the second (although older children, mean age: 47 months, do; Experiments 1 and 2). Toddlers succeed only when the events are initiated by a dispositional agent (Experiment 3), when the events involve direct contact between objects (Experiment 4), or when the events are described using causal language (Experiment 5). This suggests that causal language may help children extend their initial causal representations beyond agent-initiated and direct contact events.

Bayesian models have been applied to many areas of cognitive science including vision, language, and motor learning. We discuss the implications of this framework for cognitive development. We first present a brief introduction to the Bayesian framework. Bayesian models make assumptions about representation explicit, and provide a detailed account of learning. Furthermore, they can provide an account of developmental transitions and other phenomena in development, such as curiosity and exploration. Drawing on recent work bridging empirical developmental data and modeling, we show that these features of the Bayesian approach provide solutions to problems that elude traditional accounts of learning and raise new areas of investigation.

We present a proposal—“The Sampling Hypothesis”—suggesting that the variability in young children’s responses may be part of a rational strategy for inductive inference. In particular, we argue that young learners may be randomly sampling from the set of possible hypotheses that explain the observed data, producing different hypotheses with frequencies that reflect their subjective probability. We test the Sampling Hypothesis with four experiments on four- and five-year-olds. In these experiments, children saw a distribution of colored blocks and an event involving one of these blocks. In the first experiment, one block fell randomly and invisibly into a machine, and children made multiple guesses about the color of the block, either immediately or after a one-week delay. The distribution of guesses was consistent with the distribution of block colors, and the dependence between guesses decreased as a function of the time between guesses. In Experiments 2 and 3 the probability of different colors was systematically varied by condition. Preschoolers’ guesses tracked the probabilities of the colors, as should be the case if they are sampling from the set of possible explanatory hypotheses. Experiment 4 used a more complicated two-step process to randomly select a block and found that the distribution of children’s guesses matched the probabilities resulting from this process rather than the overall frequency of different colors. This suggests that the children’s probability matching reflects sophisticated probabilistic inferences and is not merely the result of a naïve tabulation of frequencies. Taken together the four experiments provide support for the Sampling Hypothesis, and the idea that there may be a rational explanation for the variability of children’s responses in domains like causal inference.

People can behave in a way that is consistent with Bayesian models of cognition, despite the fact that performing exact Bayesian inference is computationally challenging. What algorithms could people be using to make this possible? We show that a simple sequential algorithm “Win-Stay, Lose-Sample”, inspired by the Win-Stay, Lose-Shift (WSLS) principle, can be used to approximate Bayesian inference. We investigate the behavior of adults and preschoolers on two causal learning tasks to test whether people might use a similar algorithm. These studies use a “mini-microgenetic method”, investigating how people sequentially update their beliefs as they encounter new evidence. Experiment 1 investigates a deterministic causal learning scenario and Experiments 2 and 3 examine how people make inferences in a stochastic scenario. The behavior of adults and preschoolers in these experiments is consistent with our Bayesian version of the WSLS principle. This algorithm provides both a practical method for performing Bayesian inference and a new way to understand people’s judgments.

Although probabilistic models of cognitive development have become increasingly prevalent, one challenge is to account for how children might cope with a potentially vast number of possible hypotheses. We propose that children might address this problem by ‘sampling’ hypotheses from a probability distribution. We discuss empirical results demonstrating signatures of sampling, which offer an explanation for the variability of children's responses. The sampling hypothesis provides an algorithmic account of how children might address computationally intractable problems and suggests a way to make sense of their ‘noisy’ behavior.

Natural pedagogy emerges early in development, but good teaching requires tailoring evidence to learners’ knowledge. How does the ability to reason about others’ minds support early pedagogical evidence selection abilities? In three experiments (N = 205), we investigated preschool-aged children’s ability to consider others’ knowledge when selecting evidence in the service of teaching. Results from Experiment 1 revealed that four-year-olds reliably selected evidence to rectify others’ false beliefs, and provided causal explanations in their teaching, whereas three-year-olds did not. In Experiment 2, we tie children’s evidence selection abilities to Theory of Mind (ToM) development, above and beyond effects of age and numerical conservation abilities. In Experiment 3, we employed a 6-week training of children’s pedagogical evidence selection with a new teaching task, and further explored the relationship between these skills and children’s ToM abilities. We qualitatively replicated our results from Experiment 2, and report tentative evidence for a link between the pedagogical training and improvements in ToM. Together, our findings suggest important connections between reasoning about others’ minds and evidential reasoning in natural pedagogy in early childhood.

In the past thirty years developmental psychologists have developed techniques for investigating the cognitive resources of infants. These techniques show that an infant’s initial representation of the world is richer and more abstract than traditional empiricists supposed. For instance, infants seem to have at least some understanding of distance, of the continued existence of objects which are out of sight, and of the mental states of others. Such results have led philosophers to reconsider the idea – to be found in Plato – that there may be innate constraints on the way we view the world, and to examine the extent to which innate ‘knowledge’ may be revised as a result of learning.

The Swiss psychologist Jean Piaget was the founder of the field we now call cognitive development. His own term for the discipline was ‘genetic epistemology’, reflecting his deep philosophical concerns. Among Piaget’s most enduring contributions were his remarkably robust and surprising observations of children. Time after time, in a strikingly wide variety of domains, and at every age from birth to adolescence, he discovered that children understood the world in very different ways from adults. But Piaget was really only interested in children because he thought they exemplified basic epistemological processes. By studying children we could discover how biological organisms acquire knowledge of the world around them. The principles of genetic epistemology could then be applied to other creatures, from molluscs to physicists. Piaget’s other enduring legacy is the idea that apparently foundational kinds of knowledge were neither given innately nor directly derived from experience. Rather, knowledge was constructed as a result of the complex interplay between organisms and their environment. Piaget saw this view as an alternative to both classical rationalism and empiricism.

We propose a new dataset for evaluating question answering models with respect to their capacity to reason about beliefs. Our tasks are inspired by theory-of-mind experiments that examine whether children are able to reason about the beliefs of others, in particular when those beliefs differ from reality. We evaluate a number of recent neural models with memory augmentation. We find that all fail on our tasks, which require keeping track of inconsistent states of the world; moreover, the models' accuracy decreases notably when random sentences are introduced to the tasks at test.

People use hypothesis-driven search to learn about novel concepts, favoring information sources that reduce uncertaintyacross a set of hypotheses about a target concept. We used childrens information search to investigate their reliance on twotypes of hypothesis spaces: exemplar-based representations or a hierarchical hypothesis space based on cue abstraction.Five- to seven-year-olds learned to rank monsters according to a hierarchical decision rule involving two cues (shape andcolor). Children generated evidence by selecting pairs of monsters and observing which one ranked higher; they werethen tested on whether they learned the decision rule and correct ranking. A comparison of exemplar-based and cue-based Bayesian models revealed that all children made search decisions predicted by the exemplar-based model, but olderchildren could use collected evidence to infer the underlying hierarchical structure. These results suggest a dissociationbetween the representations used to drive search and to make inferences from evidence during development.

Previous research has shown that active engagement driveschildren’s remarkable learning capabilities. We investigatedwhether preschoolers are ecological learners, able to selectthose active learning strategies that are most informative in agiven task. Children (n = 114; 3 to 5 years old) chose betweentwo exploratory actions (opening vs. shaking) to find an eggshaker hidden in one of four small boxes, contained in twolarger boxes. Prior to this game, children learnt that the eggwas equally likely to be in any of the four small boxes(Uniform condition), or that it was most likely to be in oneparticular small box (Skewed condition). Results show that 3-and 4-year-olds, but not 5-year-olds, successfully tailored theirexploratory actions to these different likelihood distributions.We suggest that ecological learning may be a key mechanismexplaining how children can efficiently learn about the worldaround them.

People can behave in a way that is consistent with Bayesian models of cognition, despite the fact that performing exact Bayesian inference is computationally challenging. What algorithms could people be using to make this possible? We show that a simple sequential algorithm “Win-Stay, Lose-Sample”, inspired by the Win-Stay, Lose-Shift (WSLS) principle, can be used to approximate Bayesian inference. We investigate the behavior of adults and preschoolers on two causal learning tasks to test whether people might use a similar algorithm. These studies use a “mini-microgenetic method”, investigating how people sequentially update their beliefs as they encounter new evidence. Experiment 1 investigates a deterministic causal learning scenario and Experiments 2 and 3 examine how people make inferences in a stochastic scenario. The behavior of adults and preschoolers in these experiments is consistent with our Bayesian version of the WSLS principle. This algorithm provides both a practical method for performing Bayesian inference and a new way to understand people’s judgments.

Adults’ causal representations integrate information about predictive relations and the possibility of effective intervention; if one event reliably predicts another, adults can represent the possibility that acting to bring about the first event might generate the second. Here we show that although toddlers (mean age: 24 months) readily learn predictive relationships between physically connected events, they do not spontaneously initiate one event to try to generate the second (although older children, mean age: 47 months, do; Experiments 1 and 2). Toddlers succeed only when the events are initiated by a dispositional agent (Experiment 3), when the events involve direct contact between objects (Experiment 4), or when the events are described using causal language (Experiment 5). This suggests that causal language may help children extend their initial causal representations beyond agent-initiated and direct contact events.

How was the evolution of our unique biological life history related to distinctive human developments in cognition and culture? We suggest that the extended human childhood and adolescence allows a balance between exploration and exploitation, between wider and narrower hypothesis search, and between innovation and imitation in cultural learning. In particular, different developmental periods may be associated with different learning strategies. This relation between biology and culture was probably coevolutionary and bidirectional: life-history changes allowed changes in learning, which in turn both allowed and rewarded extended life histories. In two studies, we test how easily people learn an unusual physical or social causal relation from a pattern of evidence. We track the development of this ability from early childhood through adolescence and adulthood. In the physical domain, preschoolers, counterintuitively, perform better than school-aged children, who in turn perform better than adolescents and adults. As they grow older learners are less flexible: they are less likely to adopt an initially unfamiliar hypothesis that is consistent with new evidence. Instead, learners prefer a familiar hypothesis that is less consistent with the evidence. In the social domain, both preschoolers and adolescents are actually the most flexible learners, adopting an unusual hypothesis more easily than either 6-y-olds or adults. There may be important developmental transitions in flexibility at the entry into middle childhood and in adolescence, which differ across domains.

Extensive research has explored the ability of young children to learn about the causal structure of the world from patterns of evidence. These studies, however, have been conducted with middle-class samples from North America and Europe. In the present study, low-income Peruvian 4- and 5-year-olds and adults, low-income U.S. 4- and 5-year-olds in Head Start programs, and middle-class children from the United States participated in a causal learning task (N = 435). Consistent with previous studies, children learned both specific causal relations and more abstract causal principles across culture and socioeconomic status (SES). The Peruvian children and adults generally performed like middle-class U.S. children and adults, but the low-SES U.S. children showed some differences.

Natural pedagogy emerges early in development (Knudsen &Liszkowski, 2012), but good teaching requires presenting ev-idence specific to learners’ knowledge (Shafto, Goodman, &Griffiths, 2014). How might the development of Theory ofMind (ToM) relate to the ability to select pedagogical evi-dence? We present a training study in which we investigatedthe link between preschool-aged children’s false-belief under-standing and their ability to select evidence for teaching. Ourresults suggest that children with more advanced ToM abili-ties were better evidence selectors, even when controlling foreffects of age and numerical conservation abilities. We alsofound that children who improved more in false-belief under-standing from pre- to post-test performed better on the peda-gogical tasks over the course of the training. Finally, we reporttentative evidence for a link between the pedagogical train-ing and improvements in ToM. Our findings suggest importantconnections between ToM and evidential reasoning in naturalpedagogy in early childhood.

Research on wishful thinking suggests that desires bias adult’s probability judgments. Previous research has yet toexplore if this extends to young children. In Experiment 1, 260 4- and 6-year-olds in the U.S. and Peru played a card game,where selecting a desirable card was unlikely. In Experiment 2, 200 4- to 6-year-old children were shown a bag of plasticeggs; a few contained desirable prizes. Children were asked to make predictions about what card / egg would be randomlyselected. Answers were compared to control conditions in which probability was comparable, but children had no reason todesire a specific outcome. In control conditions, children tended to state that the majority card/ egg would be selected. In theexperimental conditions, children were more likely to state that the desirable (and improbable) card/ egg would be selected.Results suggest that a desire bias extends to children as young as 4.

?Como saben los ninos pequenos lo que saben? Esta pregunta ha inquietado durante largo tiempo a filosofos y psicologos. Ahora, tambien ocupa a los expertos en computacion. Los especialistas en inteligencia artificial estan estudiando la manera en que razonan los ninos con el objetivo de ensenar a las maquinas a comprender el mundo. Dos estrategias complementarias de aprendizaje automatico —timidos intentos de remedar lo que los pequenos hacen de forma natural— han comenzado a transformar la inteligencia artificial.

This study explores the development of free will beliefs across cultures. Sixty-seven Chinese 4- and 6-year-olds were asked questions to gauge whether they believed that people could freely choose to inhibit or act against their desires. Responses were compared to those given by the U.S. children in Kushnir, Gopnik, Chernyak, Seiver, and Wellman (). Results indicate that children from both cultures increased the amount of choice they ascribed with age. For inhibition questions, Chinese children ascribed less choice than the U.S. children. Qualitative explanations revealed that the U.S. children were also more likely to endorse notions of autonomous choice. These findings suggest both cultural differences and similarities in free will beliefs.

Three experiments investigate how self-generated explanation influences children's causal learning. Five-year-olds (N = 114) observed data consistent with two hypotheses and were prompted to explain or to report each observation. In Study 1, when making novel generalizations, explainers were more likely to favor the hypothesis that accounted for more observations. In Study 2, explainers favored a hypothesis that was consistent with prior knowledge. Study 3 pitted a hypothesis that accounted for more observations against a hypothesis consistent with prior knowledge. Explainers were more likely to base generalizations on prior knowledge. Findings suggest that attempts to explain drive children to evaluate hypotheses using features of "good" explanations, or those supporting generalizations with broad scope, as informed by children's prior knowledge and observations.

This study explores the relationship between beliefs aboutself-control and the ability to exercise self-control in 4- to 6-year-old children. Sixty-eight children were asked a series ofquestions to gauge whether they believed that they couldfreely choose to act against their desires or inhibit themselvesfrom performing desired actions. Children were also asked toprovide qualitative explanations for why they could or couldnot exercise free will, and to complete two inhibitory controltasks: forbidden toy and day/night. Choice responses werenegatively correlated with performance on the forbidden toytask, when children performed that task first. There was also anegative correlation between a belief in an internal locus ofcontrol, and success on the forbidden toy measure. Refrainingfrom touching a forbidden toy appears to be correlated to lessbelief in free will. Though this may appear counter-intuitive,it is consistent with cross-cultural research.

Preschoolers use both direct observation of statistical data and informant testimony to learn causal relationships. Can children integrate information from these sources, especially when source reliability is uncertain? We investigate how children handle a conflict between what they hear and what they see. In Experiment 1, 4-year-olds were introduced to a machine and 2 blocks by a knowledgeable informant who claimed to know which block was better at activating the machine, or by a naïve informant who guessed. Children then observed probabilistic evidence contradicting the informant and were asked to identify the block that worked better. Next, the informant claimed to know which of 2 novel blocks was a better activator, and children chose 1 block to try themselves. After observing conflicting data, children were more likely to say the informant's block was better when the informant was knowledgeable than when she was naïve. Children also used the statistical data to evaluate the informant's reliability and were less likely to try the novel block she endorsed than children in a baseline group who did not observe data. In Experiment 2, children saw conflicting deterministic data; the majority chose the block that consistently activated the machine as better than the endorsed block. Children's causal inferences varied with the confidence of the informant and strength of the statistical data, and informed their future trust in the informant. Children consider the strength of both social and physical causal cues even when they disagree and integrate information from these sources in a rational way.

Jacobs et al. [[1][1]] raise some interesting and important questions about our recent paper in which New Caledonian crows were unable to design a novel causal intervention [[2][2]]. These concern the problematic nature of animal/human comparisons, the complexity of our apparatus, other evidence

In our article (Weisberg & Gopnik, 2013), we construct a unifying theory of imaginative processes, in which counterfactuals and pretend games share an underlying cognitive capacity. This theory allows us to explain why children pretend, why they create fantastical pretend worlds, and how pretend play and counterfactual reasoning can help children to learn about the causal structure of reality. In her response, Beck draws a distinction between two types of counterfactuals: general counterfactuals and real-world counterfactuals. General counterfactuals are the broadest class of counterfactuals, encompassing hypothetical futures, fictional events, and pretend games. Real-world counterfactuals are a subset of general counterfactuals, and they are defined by their relationship to the real world: They represent specific alternatives to events that actually happened, or minor modifications to specific elements of reality. We agree with Beck that the class of counterfactuals is broad and that not all types of counterfactuals share the same features. But we wish to add further nuance to this argument by pointing out that it may be difficult if not impossible to draw sharp lines between categories of counterfactuals, because the space of counterfactuals is continuous. Counterfactuals can differ from each other and from the real world along a wide variety of dimensions, and this creates an infinite number of possible worlds to contrast with any given state of the real world (Lewis, 2000). Even if we define real-world counterfactuals as those that are “closest” to the real world because they only change one or a very small set of real-world facts, it does not seem possible to determine where the boundaries of this category are. At what point do real-world counterfactuals become dissimilar enough from reality that they cease to be real-world counterfactuals and become general counterfactuals? We believe that there is no straightforward answer to this question, highlighting the difficulty with drawing distinctions between types of counterfactuals. What are the implications of this argument for future research in this area? One fruitful set of research questions could examine the differences between all logically possible counterfactuals and the ones that people actually consider. Ruth Byrne's systematic work on which counterfactual situations adults tend to construct when invited to reason counterfactually provides a first set of answers to these questions, outlining which dimensions of counterfactual space are most psychologically salient (Byrne, 2005). For example, adults are more likely to construct counterfactuals for events in the recent past than for events in the more distant past, and they are also more likely to regret actions that they have taken rather than instances of inaction. These results point to the importance of temporal distance and intentions in shaping how adults think of the space of counterfactuals. Future work could build on these results to examine similar issues in children. A second set of questions concerns the nature of the psychological demands on creating counterfactuals at various points in the space, which could in turn give us clues about when in development these abilities may appear. Indeed, work by Beck and others suggests that the ability to calculate real-world counterfactuals does not seem to develop until around the age of 7 or 8 (e.g., Beck, Robinson, Carroll, & Apperly, 2006; Rafetseder, Cristi-Vargas, & Perner, 2010). However, some forms of general counterfactuals, notably pretend play, appear early in development. As Beck notes, this age disparity implies that real-world counterfactuals may not underlie children's pretend games or their early-developing causal reasoning abilities, which may instead be more closely tied to general counterfactuals. Although this argument could also be taken to imply that real-world counterfactuals involve different underlying capacities than other forms of counterfactuals or than pretend play, we believe it makes more sense to think of all of these imaginative abilities as united and to look to other variables to explain the developmental lag. As Beck ably points out, real-world counterfactuals place heavier inhibitory demands on children, and we believe that this is the primary source of the developmental difference. Because these counterfactuals differ so little from reality, younger children have trouble inhibiting what they know about current reality to reason about a highly similar counterfactual possibility. Due to these information-processing limitations, it is relatively simple to construct tasks that will lead to failure. A more difficult and interesting question is how to phrase tasks so that even young can demonstrate the competences they have. This line of argument raises a puzzle about exactly how a counterfactual's distance from reality affects children's abilities to think about that counterfactual. As just noted, for real-world counterfactuals, the similarity between reality and the counterfactual scenario makes it more difficult to think counterfactually. But for pretense, the similarity between reality and the pretend game seems to make it easier. The earliest (hence presumably easiest) forms of representational pretend play involve object substitution, especially with objects that share many features in common with their referents. Children who are just beginning to pretend find it easier to produce and comprehend pretend actions which use props that have many similarities to their referents (using a stick as a pencil) than those that are physically different (using a ball as a pencil) or than entirely invisible objects (Bigham & Bourchier-Sutton, 2007; Elder & Pederson, 1978; Jackowitz & Watson, 1980). This developmental sequence strongly implies that similarity between reality and pretense facilities the construction of the kind of counterfactual representations used in pretending. What is the source of this difference between counterfactual thinking for real-world counterfactuals and counterfactual thinking for pretense? One possibility is that pretending and constructing real-world counterfactuals simply involve different mental processes. As noted above, we find this unlikely, both given the many other computational similarities between counterfactual and pretend reasoning and given the difficulty of defining a separate category of real-world counterfactuals. Instead, we believe that there are two other main sources of difference. First, unlike the kinds of counterfactual scenarios used in psychological research, object substitution pretense involves the implicit evocation of a counter-to-fact representation. Children may be able to implicitly reason about close possible worlds before they are able to explicitly do so. Second, reducing the distance between reality and the possible world may make pretending easier because of the presence of a prop. Having a physical object with many of the same features as the pretend object can help children to navigate across the boundary between reality and pretense. No such object exists in the counterfactual cases, which means that the child must construct the representation of the real-world counterfactual without the help of a physical reminder. Whatever the reason, considering the nature of the distance between reality and a counterfactual world may lead to new insights into the development of counterfactual thinking and of the myriad mental processes that rely on this important ability.

We describe a surprising developmental pattern we found in studies involving three different kinds of problems and age ranges. Younger learners are better than older ones at learning unusual abstract causal principles from evidence. We explore two factors that might contribute to this counterintuitive result. The first is that as our knowledge grows, we become less open to new ideas. The second is that younger minds and brains are intrinsically more flexible and exploratory, although they are also less efficient as a result.

A key aspect of theory of mind is the ability to reason\nabout other people's desires. As adults, we know that desires\nand preferences are subjective and specific to the individual.\nHowever, research in cognitive development suggests that a\nsignificant conceptual shift occurs in desire-based reasoning\nbetween 14 and 18 months of age, allowing 18- but not 14-\nmonth-olds to understand that different people can have\ndifferent preferences (Lucas et al., 2014; Ma & Xu 2011;\nRepacholi & Gopnik, 1997). The present research investigates\nthe kind of evidence that is relevant for inducing this shift and\nwhether younger infants can be trained to learn about the\ndiversity of preferences. In Experiment 1, infants younger\nthan 18 months of age were shown demonstrations in which\ntwo experimenters either liked the same objects as each other\n(in one training condition) or different objects (in another\ntraining condition). Following training, all infants were asked\nto share one of two foods with one of the experimenters –\nthey could either share a food that the experimenter showed\ndisgust towards (and the infants themselves liked) or a food\nthat the experimenter showed happiness towards (and the\ninfants themselves did not like). We found that infants who\nobserved two different experimenters liking different objects\nduring training later provided the experimenter with the food\nshe liked, even if it was something they disliked themselves.\nHowever, when infants observed two experimenters liking the\nsame objects, they later incorrectly shared the food that they\nthemselves liked with the experimenter. Experiment 2\ncontrolled for an alternative interpretation of these findings.\nOur results suggest that training allows infants to overturn an\ninitial theory in the domain of Theory of Mind for a more\nadvanced one.

This is a cross-cultural replication of Seiver, Gopnik & Goodman (2013) and compares the development of social causal attributions in Chinese and U.S. children. In this study, Chinese (n=110) 4-and 6-year-olds were directly compared to the U.S. children in Seiver, Gopnik & Goodman (2013). Children were shown covariation evidence that varied across conditions to imply that a person, situation, or neither was the cause of a person’s actions. Following observation, children were asked to explain why the person engaged in the actions. Findings indicate that U.S. children significantly increased the amount of person attributions they made with age, while Chinese 4-and 6-year-olds answered comparably. Children from both cultures were sensitive to covariation manipulations when they suggested a person was the cause of an action. Only U.S. children were sensitive to evidence when it favored the situation.

We explore the developmental trajectory and underlying mechanisms of relational reasoning. We describe a surprising developmental pattern: Younger learners are better than older ones at inferring abstract relations. Walker and Gopnik (2014) demonstrated that toddlers are able to infer the relations “same” and “different” in a causal system. However, these findings appear to contrast with the literature suggesting that older children have difficulty inferring these relations. Here we manipulate the data and children’s search procedure to assess the influence of these factors. In Experiment 1, we find that while younger children have no difficulty learning these relational concepts, older children fail to draw this abstract inference. In Experiment 2, we demonstrate that older children have learned the hypothesis that individual kinds of objects lead to effects. Finally, Experiment 3 indicates that including an explanation prompt during learning also improves performance. Findings are discussed in light of computational theories of learning.

Bayesian models have been applied to many areas of cognitive science including vision, language, and motor learning. We discuss the implications of this framework for cognitive development. We first present a brief introduction to the Bayesian framework. Bayesian models make assumptions about representation explicit, and provide a detailed account of learning. Furthermore, they can provide an account of developmental transitions and other phenomena in development, such as curiosity and exploration. Drawing on recent work bridging empirical developmental data and modeling, we show that these features of the Bayesian approach provide solutions to problems that elude traditional accounts of learning and raise new areas of investigation. WIREs Cogn Sci 2015, 6:75–86. doi: 10.1002/wcs.1330 This article is categorized under: Psychology > Development and Aging Psychology > Reasoning and Decision Making

Four experiments with preschool-aged children test the hypothesis that engaging in explanation promotes inductive reasoning on the basis of shared causal properties as opposed to salient (but superficial) perceptual properties. In Experiments 1a and 1b, 3- to 5-year-old children prompted to explain during a causal learning task were more likely to override a tendency to generalize according to perceptual similarity and instead extend an internal feature to an object that shared a causal property. Experiment 2 replicated this effect of explanation in a case of label extension (i.e., categorization). Experiment 3 demonstrated that explanation improves memory for clusters of causally relevant (non-perceptual) features, but impairs memory for superficial (perceptual) features, providing evidence that effects of explanation are selective in scope and apply to memory as well as inference. In sum, our data support the proposal that engaging in explanation influences children's reasoning by privileging inductively rich, causal properties.

Humans are capable of simply observing a correlation between cause and effect, and then producing a novel behavioural pattern in order to recreate the same outcome. However, it is unclear how the ability to create such causal interventions evolved. Here, we show that while 24-month-old children can produce an effective, novel action after observing a correlation, tool-making New Caledonian crows cannot. These results suggest that complex tool behaviours are not sufficient for the evolution of this ability, and that causal interventions can be cognitively and evolutionarily disassociated from other types of causal understanding.

Fiction presents a unique challenge to the developing child, in that children must learn when to generalize information from stories to the real world. This study examines how children acquire causal knowledge from storybooks, and whether children are sensitive to how closely the fictional world resembles reality. Preschoolers (N = 108) listened to stories in which a novel causal relation was embedded within realistic or fantastical contexts. Results indicate that by at least 3 years of age, children are sensitive to the underlying causal structure of the story: Children are more likely to generalize content if the fictional world is similar to reality. Additionally, children become better able at discriminating between realistic and fantastical story contexts between 3 and 5 years of age.

How do young children learn about causal structure in an uncertain and variable world? We tested whether they can use observed probabilistic information to solve causal learning problems. In two experiments, 24-month-olds observed an adult produce a probabilistic pattern of causal evidence. The toddlers then were given an opportunity to design their own intervention. In Experiment 1, toddlers saw one object bring about an effect with a higher probability than a second object. In Experiment 2, the frequency of the effect was held constant, though its probability differed. After observing the probabilistic evidence, toddlers in both experiments chose to act on the object that was more likely to produce the effect. The results demonstrate that toddlers can learn about cause and effect without trial-and-error or linguistic instruction on the task, simply by observing the probabilistic patterns of evidence resulting from the imperfect actions of other social agents. Such observational causal learning from probabilistic displays supports human children's rapid cultural learning.

Groups of objects are nearly everywhere we look. Adults can perceive and understand the 'gist' of multiple objects at once, engaging ensemble-coding mechanisms that summarize a group's overall appearance. Are these group-perception mechanisms in place early in childhood? Here, we provide the first evidence that 4-5-year-old children use ensemble coding to perceive the average size of a group of objects. Children viewed a pair of trees, with each containing a group of differently sized oranges. We found that, in order to determine which tree had the larger oranges overall, children integrated the sizes of multiple oranges into ensemble representations. This pooling occurred rapidly, and it occurred despite conflicting information from numerosity, continuous extent, density, and contrast. An ideal observer analysis showed that although children's integration mechanisms are sensitive, they are not yet as efficient as adults'. Overall, our results provide a new insight into the way children see and understand the environment, and they illustrate the fundamental nature of ensemble coding in visual perception.

Groups are nearly everywhere we look, but our ability to perceive them is anything but commonplace. Attention and memory have limited capacity, only allowing us to perceive and remember precise information about a few things at a time. To overcome these bottlenecks, the visual system engages a mechanism known as ensemble coding, in which noisy information about many items is condensed into a single representation that summarizes the group. This direct representation allows adults to appreciate the gist at little computational cost. Without it, the scope of our visual experience would be severely impoverished. Given the importance of ensemble coding, it is surprising how little is known about its development. While children can perceive basic properties of homogeneous groups, like numerosity, it is unclear if they possess the sophisticated ensemble algorithms necessary to summarize properties of heterogeneous groups, like average size. Here, we show that 4-5 year-old children do engage such summary mechanisms. Children viewed two groups of oranges or just a pair of oranges, all with a variety of sizes. When evaluating which group had the larger oranges overall, children integrated and averaged the sizes of multiple oranges, producing percepts of groups nearly as precise as those of single oranges. This pooling was independent of numerosity, continuous extent, density, and contrast. While sensitive, an ideal observer analysis showed that children integrated fewer oranges into their ensemble representations than adults. And with brief presentations, children engaged group perception with limited flexibility—those who excelled at ensemble coding were the worst at discriminating single oranges, and vice versa. Our findings provide a novel view of ensemble coding as it develops. More generally, they reveal new insights into the way children see and understand their environment, and they illustrate the fundamental nature of ensemble coding in visual perception. Meeting abstract presented at VSS 2014

Recent work has shown that young children can learn about preferences by observing the choices and emotional reactions of other people, but there is no unified account of how this learning occurs. We show that a rational model, built on ideas from economics and computer science, explains the behavior of children in several experiments, and offers new predictions as well. First, we demonstrate that when children use statistical information to learn about preferences, their inferences match the predictions of a simple econometric model. Next, we show that this same model can explain children's ability to learn that other people have preferences similar to or different from their own and use that knowledge to reason about the desirability of hidden objects. Finally, we use the model to explain a developmental shift in preference understanding.

How do young children learn about causal structure in an uncertain and variable world? We tested whether they can use observed probabilistic information to solve causal learning problems. In two experiments, 24-month-olds observed an adult produce a probabilistic pattern of causal evidence. The toddlers then were given an opportunity to design their own intervention. In Experiment 1, toddlers saw one object bring about an effect with a higher probability than a second object. In Experiment 2, the frequency of the effect was held constant, though its probability differed. After observing the probabilistic evidence, toddlers in both experiments chose to act on the object that was more likely to produce the effect. The results demonstrate that toddlers can learn about cause and effect without trial-and-error or linguistic instruction on the task, simply by observing the probabilistic patterns of evidence resulting from the imperfect actions of other social agents. Such observational causal learning from probabilistic displays supports human children’s rapid cultural learning. Research highlights • We examined toddlers ’ causal learning from proba-bilistic events using a two-choice action measure. • The methods isolate observational causal learning as the underlying mechanism and separate it from causal linguistic descriptions or spatial contact. • Toddlers learn the causal structure of probabilistic events from observing other people’s actions and use this to generate their own actions to bring about the same effect on the world.

A notion of free will requires an understanding that an agent who performed a particular action could have acted otherwise.Little is known about how young children reason about free will, and if this reasoning is affected by the surrounding cultural context.In this study, U.S. and Chinese 4-and 6-year-olds were asked if people could choose to inhibit or act against their desires.U.S. children attributed more choice to people than Chinese children for the inhibition question type but not the action question type.There was a significant age by ethnicity interaction, with ethnicity differences significant at age 6, but not at age 4. Children from both cultures did attribute more choice to other people than they did to themselves.These results suggest that folk intuitions of free will undergo change in early childhood and that this change is, at least in part, influenced by culturally variable factors.

Recent research suggests that the ability to infer the abstract, relational concepts "same" and "different" in a causal learning task is in place as early as 18 months (Walker & Gopnik, 2013).Here we address whether infants' developing linguistic abilities play a role in the acquisition of these concepts.Researchers have long proposed a close relationship between semantic and cognitive development, and a variety of connections have been found between specific linguistic and conceptual achievements.In the current study, we examine 14-and 15-month olds' relational inferences in a causal match-to-sample task, and assess performance as a function of relational word production.Infants' use of relational words such as "more" was correlated with their success on this task after controlling for general language abilities.Findings strengthen the claim that relational reasoning is early developing, and provide evidence that abstract concepts are reflected in and linked to specific linguistic representations.

The review by Lillard et al. (2013) highlighted the need for additional research to better clarify the nature of the relationship between pretend play and development. However, the authors did not provide a proposal for how to structure the direction of this future work. Here, we provide a possible framework for generating additional research. This theoretical proposal is based on recent computational approaches to cognition, in which counterfactual reasoning plays a central role in causal learning. We propose that pretend play initially emerges as a product of the cognitive mechanisms underlying human learning and then feeds back to become critical for enhancing the optimal functioning of these same processes. More specifically, we argue that pretending is in fact 1 of several forms of counterfactual reasoning, which is essential to causal cognition-and that the act of engaging in pretend scenarios may provide early opportunities to practice the skills that were initially responsible for its appearance. Here, we provide a brief overview of this theoretical framework, consider how these ideas may be integrated with the previous work covered in Lillard et al.'s (2013) review, and suggest some empirically testable questions to direct future directions.

Watch any crowded intersection, and you will see how adept people are at reading the subtle movements of one another. While adults can readily discriminate small differences in the direction of a moving person, it is unclear if this sensitivity is in place early in development. Here, we present evidence that 4-year-old children are sensitive to small differences in a person's direction of walking (∼7°) far beyond what has been previously shown. This sensitivity only occurred for perception of an upright walker, consistent with the recruitment of high-level visual areas. Even at 4 years of age, children's sensitivity approached that of adults'. This suggests that the sophisticated mechanisms adults use to perceive a person's direction of movement are in place and developing early in childhood. Although the neural mechanisms for perceiving biological motion develop slowly, they are refined enough by age 4 to support subtle perceptual judgments of heading. These judgments may be useful for predicting a person's future location or even their intentions and goals.

Causal knowledge is central to children’s understanding of the world, but there have been many different conceptions of how causal knowledge is represented and how it is learned. I outline four different approaches to causality that have influenced developmental research, stressing agency, mechanism, association, and probabilistic models, and give examples of theory and research in each area. I focus in more detail on the most recent probabilistic model accounts that stress the role of causation in inferences about possibilities. These include potential interventions to change the world and counterfactuals about alternate ways the world might be. These accounts also employ computational ideas about causal Bayes nets and Bayesian learning, and these ideas are outlined.

Young children spend a large portion of their time pretending about non-real situations. Why? We answer this question by using the framework of Bayesian causal models to argue that pretending and counterfactual reasoning engage the same component cognitive abilities: disengaging with current reality, making inferences about an alternative representation of reality, and keeping this representation separate from reality. In turn, according to causal models accounts, counterfactual reasoning is a crucial tool that children need to plan for the future and learn about the world. Both planning with causal models and learning about them require the ability to create false premises and generate conclusions from these premises. We argue that pretending allows children to practice these important cognitive skills. We also consider the prevalence of unrealistic scenarios in children's play and explain how they can be useful in learning, despite appearances to the contrary.

Children make inductive inferences about the causal properties of individual objects from a very young age. When can they infer higher-order relational properties? In three experiments, we examined 18- to 30-month-olds' relational inferences in a causal task. Results suggest that at this age, children are able to infer a higher-order relational causal principle from just a few observations and use this inference to guide their own subsequent actions and bring about a novel causal outcome. Moreover, the children passed a revised version of the relational match-to-sample task that has proven very difficult for nonhuman primates. The findings are considered in light of their implications for understanding the nature of relational and causal reasoning, and their evolutionary origins.

Many researchers have long assumed imaginative play critical to the healthy cognitive, social, and emotional development of children, which has important implications for early-education policy and practice. But, the authors find, a careful review of the existing literature highlights a need for a better theory to clarify the nature of the relationship between pretend play and childhood development. In particular, they ask why children spend so much time engaging in unreal scenarios at a time when they know relatively little about the real world? The authors review the idea that children pretend because it exercises their developing ability to reason counter-factually-an ability essential for causal reasoning and learning. The authors present a look at their study in progress aimed at assessing their theory. According to the model of play they outline, imaginative play serves as an engine of learning. Such play arises out of the human capacity for causal cognition and feeds back to help develop causal-reasoning skills. Keywords: Bayesian learning methods; causal learning; counterfactual reasoning; pretend play; probabilistic modelsAcross species, the activities typically involved in play are those that will become important in adulthood (Bekoff and Byers 1998). Play, then, is a type of exploratory learning in which the young animal engages in a variety of behaviors in a low-risk, low-cost context. Over the last ten years or so, a growing body of research has focused on a unique form of exploratory play in human chil- dren involving a process of informal experimentation on the world. This work demonstrates that children's early exploration of the external world during free play helps them learn the complexities of causal relationships (e.g., see Schulz [2012] for a review).However, human children also engage in a particularly distinctive kind of internal exploratory play-pretend play. During pretense, children not only play with objects and social partners in the actual world, they also construct (sometimes rather elaborate) unreal scenarios about possible worlds. The ideas we address in this article relate to a long-standing question in developmental psychology about the apparently paradoxical nature of this type of pretend play: Why do young children spend so much of their time and resources on unreality, given that they are still learning about the real world?Classically, researchers (Freud 1922; Piaget 1962) offered a rather unchari- table characterization of pretend play-attributing its prevalence in childhood to children's inability to differentiate between fantasy and reality. However, decades of research have shown that, to the contrary, children seem quite proficient at distinguishing between the two (Flavell, Flavell, and Green 1987; Morison and Gardner 1978; Skolnick and Bloom 2006; Taylor 1999; Walton 1990; Woolley and Cox 2007; Woolley and Ghossainy 2013). Many of these early deflationary theories of pretense in childhood also ignored some of the most prominent features of pretend play, features that suggest it is the result of cognitive compe- tence rather than cognitive limitations. Pretense is unique to human beings; it is often social in nature; and it becomes increasingly elaborate over the course of early childhood (Harris 2000). Indeed, one of the more compelling features of pretend play is its continuity throughout the lifespan: adults spend large por- tions of their lives engaged with fictional worlds.No wonder that developmental psychologists and educators have long sus- pected pretend play contributes to learning. Historically however, there have been surprisingly few empirical studies that provide strong support for this idea (see Lil- lard et al. 2013 for a recent review). We have argued elsewhere (Walker and Gopnik 2013a) that the lack of a unifying theory of pretend play has been detrimental to its empirical study because a coherent theory serves to generate a set of testable predictions to guide the direction of research. …

Author(s): Buchsbaum, Daphna; Walker, Caren; Gopnik, Alison; Charter, Nick; Banks, David; Lucas, Christopher; Kemp, Charles; Rafetseder, Eva; Perner, Josef

Children make inductive inferences about the causal properties of individual objects from a very young age.When can they infer higher-order relational properties -a task that has proven difficult for non-human primates?In two experiments, we examined 18-24-month-old infants' relational inferences using a causal version of a relational match-to-sample task.Results suggest that by 21-24 months of age, infants are able to infer a relational causal principle from just a few observations and use this inference to guide their own subsequent actions and bring about a novel causal outcome.Findings are considered in light of recent discussion about the nature of relational and causal reasoning, and their evolutionary origins.

Three experiments test the hypothesis that engaging in explanation prompts children to favor inductively rich properties when generalizing to novel cases.In Experiment 1, preschoolers prompted to explain during a causal learning task were more likely to override a tendency to generalize according to perceptual similarity and instead extend an internal feature to an object that shared a causal property.In Experiment 2, we replicated this effect of explanation in a case of label extension.Experiment 3 demonstrated that explanation improves memory for internal features and labels, but impairs memory for superficial features.We conclude that explaining can influence learning by prompting children to favor inductively rich properties over surface similarity.

How do infants and young children learn about the causal structure of the world around them? In 4 experiments we investigate whether young children initially give special weight to the outcomes of goal-directed interventions they see others perform and use this to distinguish correlations from genuine causal relations--observational causal learning. In a new 2-choice procedure, 2- to 4-year-old children saw 2 identical objects (potential causes). Activation of 1 but not the other triggered a spatially remote effect. Children systematically intervened on the causal object and predictively looked to the effect. Results fell to chance when the cause and effect were temporally reversed, so that the events were merely associated but not causally related. The youngest children (24- to 36-month-olds) were more likely to make causal inferences when covariations were the outcome of human interventions than when they were not. Observational causal learning may be a fundamental learning mechanism that enables infants to abstract the causal structure of the world.

We propose a new version of the "theory theory" grounded in the computational framework of probabilistic causal models and Bayesian learning. Probabilistic models allow a constructivist but rigorous and detailed approach to cognitive development. They also explain the learning of both more specific causal hypotheses and more abstract framework theories. We outline the new theoretical ideas, explain the computational framework in an intuitive and nontechnical way, and review an extensive but relatively recent body of empirical results that supports these ideas. These include new studies of the mechanisms of learning. Children infer causal structure from statistical information, through their own actions on the world and through observations of the actions of others. Studies demonstrate these learning mechanisms in children from 16 months to 4 years old and include research on causal statistical learning, informal experimentation through play, and imitation and informal pedagogy. They also include studies of the variability and progressive character of intuitive theory change, particularly theory of mind. These studies investigate both the physical and the psychological and social domains. We conclude with suggestions for further collaborative projects between developmental and computational cognitive scientists.

We argue for a theoretical link between the development of an extended period of immaturity in human evolution and the emergence of powerful and wide-ranging causal learning mechanisms, specifically the use of causal models and Bayesian learning. We suggest that exploratory childhood learning, childhood play in particular, and causal cognition are closely connected. We report an empirical study demonstrating one such connection--a link between pretend play and counterfactual causal reasoning. Preschool children given new information about a causal system made very similar inferences both when they considered counterfactuals about the system and when they engaged in pretend play about it. Counterfactual cognition and causally coherent pretence were also significantly correlated even when age, general cognitive development and executive function were controlled for. These findings link a distinctive human form of childhood play and an equally distinctive human form of causal inference. We speculate that, during human evolution, computations that were initially reserved for solving particularly important ecological problems came to be used much more widely and extensively during the long period of protected immaturity.

An intuitive theory is a system of abstract concepts and laws relating those concepts that together provide a framework for explaining some domain of phenomena. Constructing an intuitive theory based on observing the world, as in building a scientific theory from data, confronts learners with a “chicken-and-egg” problem: the laws can only be expressed in terms of the theory's core concepts, but these concepts are only meaningful in terms of the role they play in the theory's laws; how is a learner to discover appropriate concepts and laws simultaneously, knowing neither to begin with? Even knowing the number of categories in a theory does not resolve this problem: without knowing how individuals should be sorted (which categories each belongs to), a the causal relationships between categories cannot be resolved. We explore how children can solve this chicken-and-egg problem in the domain of magnetism, drawing on perspectives from history of science, computational modeling, and behavioral experiments. We present preschoolers with a simplified magnet learning task and show how our empirical results can be explained as rational inferences within a Bayesian computational framework.

Children rely on both evidence and prior knowledge to make physical causal inferences; this study explores whether they make attributions about others' behavior in the same manner. A total of one hundred and fifty-nine 4- and 6-year-olds saw 2 dolls interacting with 2 activities, and explained the dolls' actions. In the person condition, each doll acted consistently across activities, but differently from each other. In the situation condition, the two dolls acted differently for each activity, but both performed the same actions. Both age groups provided more "person" explanations (citing features of the doll) in the person condition than in the situation condition. In addition, 6-year-olds showed an overall bias toward "person" explanations. As in physical causal inference, social causal inference combines covariational evidence and prior knowledge.

New theoretical ideas and empirical research show that very young children's learning and thinking are strikingly similar to much learning and thinking in science. Preschoolers test hypotheses against data and make causal inferences; they learn from statistics and informal experimentation, and from watching and listening to others. The mathematical framework of probabilistic models and Bayesian inference can describe this learning in precise ways. These discoveries have implications for early childhood education and policy. In particular, they suggest both that early childhood experience is extremely important and that the trend toward more structured and academic early childhood programs is misguided.

Fictional information presents a unique challenge to the developing child. Children must learn when it is appropriate to transfer information from the fictional space to the real world and what contextual cues should be considered in this decision. The current research explores children’s causal inferences between fictional representations and reality by examining their developing sensitivity to the proximity of the fictional world to the real world, and the effect of this judgment on their subsequent generalization of novel causal properties. By 3-years of age, children are able to evaluate the data that they receive from fictional stories in order to inform their generalization of novel story content to the real world. Additionally, as children develop, they become better able to discriminate between close (realistic) and far (fantastical) fictional worlds when assessing which stories are likely to provide relevant causal knowledge.

We rely on both our own observations and on others ’ testimony when making causal inferences. To integrate these sources of information we must consider an informant’s statements about the world, her expressed level of certainty, her previous accuracy, and perhaps her apparent self-knowledge – how accurately she conveys her own certainty. It can be difficult to tease apart the contributions of all these variables simply by observing people’s causal judgments. We present a computa-tional account of how these different cues contribute to a ratio-nal causal inference, and two experiments looking at adults’ inferences from causal demonstrations and informant testi-mony, focusing on cases where these sources conflict. We find that adults are able to combine social information with their own observations, and are sensitive to the reliability of each. Adults are also sensitive to the accuracy, certainty, and self-knowledge of the informant, a result confirmed by comparing predictions from models with and without these variables.

In two studies, we examine how prompting 5- and 6-year-olds to explain observed outcomes influences causal learning. In Study 1, children were presented with data consistent with two causal regularities. Explainers outperformed controls in generalizing the regularity that accounted for more observations. In Study 2, this regularity was pitted against an alternative that accounted for fewer observations but was consistent with prior knowledge. Explainers were less likely than controls to generalize the regularity that accounted for more observations. These findings suggest that explaining drives children to favor causal regularities that they expect to generalize, where current observations and prior knowledge both provide cues.

Preschoolers often switch a response on repeated questioning, even though no new evidence has been provided (Krahenbuhl, Blades, & Eiser, 2009). Though apparently irrational, this behavior may be understood as children making an induc-tive inferrence based on their beliefs about whether intial re-sponses were correct and the knowledgeability of the ques-tioner. We present a probabilistic model of how the question-ers ’ knowledge and biases to be positive should affect infer-ences. The model generates the qualitative prediction that an ideal learner should switch responses more often following a “neutral query ” from a knowledgeable questioner than follow-ing queries from an ignorant questioner. We test predictions of the model in an experiment. The results show that four-year-old children are sensitive to questioners ’ knowledge when re-sponding to a neutral query, demonstrating more switching be-havior when the query is provided by a knowledgeable ques-tioner. We conclude by discussing the practical and theoretical implications for cognitive development. When should a learner abandon their current hypothesis in favor of a new one? It is becoming clear that even preschool-aged children rationally update beliefs and generate new explanations following informative evidence (Gopnik, Gly-

Does contrastive access help preschoolers succeed on traditional false-belief tasks?Three-and four-year-olds were presented with a modified version of the change-of-location story in which two characters are the focus of interest.In the contrastive access condition preschoolers' observe that one character leaves the room while the other stays and witnesses the moving event; in the non-contrastive condition both characters leave the room and fail to observe the moving event.Despite having to track two different characters and their different knowledge states about the location of the toy, preschoolers were more likely to succeed on the task when the characters had contrasting access to the moving event.This result supports a previously unexplored qualitative prediction of the Goodman et al (2006) computational model of the false-belief task and also provides tentative support for the theory theory view of the false-belief transition.

Children’s Causal Learning from Fiction: Assessing the Proximity Between Real and Fictional Worlds Caren M. Walker 1 , Patricia A. Ganea 2 , & Alison Gopnik 1 (caren.walker@berkeley.edu, patricia.ganea@utoronto.ca, gopnik@berkeley.edu) Psychology Department, University of California, Berkeley, 3210 Tolman Hall, Berkeley, CA 94720 Human Development & Applied Psychology, University of Toronto, 56 Spadina Rd., Toronto, Canada M5R 2TI Abstract this perceived similarity by presenting information in a fictional world that seamlessly interweaves fantasy and reality (Woolley & Cox, 2007). Even in explicitly pedagogical scenarios, teachers often embed their intended curriculum within a fantasy context. This decision is based on the assumption that fictional worlds are more engaging to the young child, and may therefore encourage increased sustained attention and learning of novel material (Harris, 2000; Renninger & Wozniak, 1985). Previous research supports the proposal that fantasy contexts serve to improve children’s performance on certain types of cognitive tasks, such as deductive and syllogistic reasoning and theory of mind (e.g., Dias & Harris, 1988; Dias, Roazzi, & Harris, 2005; Hawkins, Pea, Glick & Scribner, 1984; Richards & Sanderson, 1999; Lillard & Sobel, 1999; Sobel & Lillard, 2001). For example, according to Dias et al. (2005), placing an unfamiliar premise in a fantastical context – particularly when the premise directly contradicts a currently-held theory – allows children to override their natural empirical orientation, or bias to reason in line with their past experiences. It is unknown, however, how learning and generalization of novel causal information (which does not require the suspension of existing knowledge) is affected by the fantastical contexts of the fictional stories in which this information is embedded. The ability to effectively process fictional information is dependent upon a variety of representational skills, including at least two major factors that are unique to learning from fictional material. The first includes the development of a mature concept about the boundary between the fictional and real world, as well as an understanding of what information is more appropriately quarantined to the fictional space. Second, it is necessary for children to develop an understanding of when it is appropriate to transfer information from the fictional to the real world, and what contextual cues should be considered in this decision. The current research explores the early development of each of these factors, and in particular, examines whether children’s sensitivity to contextual cues in fictional worlds changes over the course of development. Fictional information presents a unique challenge to the developing child. Children must learn when it is appropriate to transfer information from the fictional space to the real world and what contextual cues should be considered in this decision. The current research explores children’s causal inferences between fictional representations and reality by examining their developing sensitivity to the proximity of the fictional world to the real world, and the effect of this judgment on their subsequent generalization of novel causal properties. By 3-years of age, children are able to evaluate the data that they receive from fictional stories in order to inform their generalization of novel story content to the real world. Additionally, as children develop, they become better able to discriminate between close (realistic) and far (fantastical) fictional worlds when assessing which stories are likely to provide relevant causal knowledge. Keywords: causal inference; fiction; cognitive development; prior knowledge; representation The ‘Reader’s Dilemma’ Children’s growing knowledge about the world comes from a variety of sources, including their exposure to fictional material. In fact, much of the unfamiliar information that children encounter appears in the context of stories and fantastical representations of the world. Children, like adults, therefore often encounter the “reader’s dilemma”: the need to compartmentalize fictional information to insulate real world knowledge from false facts, and the simultaneous need to incorporate this information due to its potential application to a host of real world topics (Gerrig & Prentice, 1999; Potts, St. John, & Kirson, 1989). There is substantial evidence in developmental psychology that indicates that the ability to distinguish reality from fiction develops significantly during the preschool years (e.g., DeLoache, Pierroutakos, Uttal, Rosengren, & Gottlieb, 1998; Flavell, Flavell, & Green, 1989; Woolley & Cox, 2007; Woolley & Wellman, 1990; Woolley & Van Reet, 2006). However, very little research has explored children’s ability to learn causal information about the real world from their exposure to fictional material. Fictional information presents a unique challenge to the developing child. Research has shown that the transfer of knowledge is generally facilitated by similarity between the context in which the information is learned, and the context in which it is to be applied (Catranbone & Holyoak, 1989; Spencer & Weisberg, 1986). However, many of the learning contexts that are created for young children act to reduce Children’s Beliefs about Fictional Worlds There is a growing literature in developmental psychology regarding when and how children distinguish between fantasy and reality. Methods for testing this distinction vary

Splenectomy has been practised routinely, both in the emergency setting and as a therapeutic elective procedure. There is a correlation between asplenia/hyposplenia and the occurrence of fulminant sepsis, underlining the importance of developing surgical methods for preserving splenic function. Both clinical and experimental studies have shown at least partial morphological and functional regeneration of autogenic splenic implants (ASIs). Experimental studies investigating the immunoprotective effect of ASIs, based mostly on exposure of animals to various bacteria, have demonstrated that ASIs can increase the rate of bacterial clearance and decrease mortality from sepsis. Clinical studies have shown their ability to remove colloidal substances and altered erythrocyte corpuscular inclusions, such as Howell-Jolly, Heinz and Pappenheimer bodies, from the bloodstream. In this experimental study the functional and morphological regeneration of ASIs was studied over time in rats.

Abstract We need not propose, as Carey does, a radical discontinuity between core cognition, which is responsible for abstract structure, and language and “Quinian bootstrapping,” which are responsible for learning and conceptual change. From a probabilistic models view, conceptual structure and learning reflect the same principles, and they are both in place from the beginning.

Abstract My research program proposes that children have representations and learning mechanisms that can be characterized as causal models of the world – coherent, structured hypotheses with consistent relationships to probabilistic patterns of evidence. We also propose that Bayesian inference is one mechanism by which children learn these models from data. These proposals are straightforward psychological hypotheses and far from “Bayesian Fundamentalism.”

In laboratory experiments, infants are sensitive to patterns of visual features that co-occur (e.g., Fiser & Aslin, 2002). Once infants learn the statistical regularities, however, what do they do with that knowledge? Moreover, which patterns do infants learn in the cluttered world outside of the laboratory? Across 4 experiments, we show that 9-month-olds use this sensitivity to make inferences about object properties. In Experiment 1, 9-month-old infants expected co-occurring visual features to remain fused (i.e., infants looked longer when co-occurring features split apart than when they stayed together). Forming such expectations can help identify integral object parts for object individuation, recognition, and categorization. In Experiment 2, we increased the task difficulty by presenting the test stimuli simultaneously with a different spatial layout from the familiarization trials to provide a more ecologically valid condition. Infants did not make similar inferences in this more distracting test condition. However, Experiment 3 showed that a social cue did allow inferences in this more difficult test condition, and Experiment 4 showed that social cues helped infants choose patterns among distractor patterns during learning as well as during test. These findings suggest that infants can use feature co-occurrence to learn about objects and that social cues shape such foundational learning in distraction-filled environments.

People are adept at inferring novel causal relations, even from only a few observations. Prior knowledge about the probability of encountering causal relations of various types and the nature of the mechanisms relating causes and effects plays a crucial role in these inferences. We test a formal account of how this knowledge can be used and acquired, based on analyzing causal induction as Bayesian inference. Five studies explored the predictions of this account with adults and 4-year-olds, using tasks in which participants learned about the causal properties of a set of objects. The studies varied the two factors that our Bayesian approach predicted should be relevant to causal induction: the prior probability with which causal relations exist, and the assumption of a deterministic or a probabilistic relation between cause and effect. Adults' judgments (Experiments 1, 2, and 4) were in close correspondence with the quantitative predictions of the model, and children's judgments (Experiments 3 and 5) agreed qualitatively with this account.

Abstract N. S. Newcombe’s (2011) account of neoconstructivism opposes nativism to everything else—connectionist and dynamic systems theories, information processing theories, the “theory theory,” statistical learning, and Bayesian inference. But there is an alternative split that is equally important and that carves up the theoretical territory rather differently. This is a contrast between representational and nonrepresentational accounts of development. Whereas almost all nativist approaches are representational, among empiricists there is an important divide between those who, like nativists, embrace representation and those who deny it. “Probabilistic model” approaches, which include but are not limited to Bayesian learning, provide a developmental account that allows both representation and learning.

People are adept at inferring novel causal relations, even from only a few observations. Prior knowledge about the probability of encountering causal relations of various types and the nature of the mechanisms relating causes and effects plays a crucial role in these inferences. We test a formal account of how this knowledge can be used and acquired, based on analyzing causal induction as Bayesian inference. Five studies explored the predictions of this account with adults and 4-year-olds, using tasks in which participants learned about the causal properties of a set of objects. The studies varied the two factors that our Bayesian approach predicted should be relevant to causal induction: the prior probability with which causal relations exist, and the assumption of a deterministic or a probabilistic relation between cause and effect. Adults ’ judgments (Experiments 1, 2, and 4) were in close correspondence with the quantitative predictions of the model, and children’s judgments (Experiments 3 and 5) agreed qualitatively with this account.

How do children use covariation information to infer causation?In this series of studies, we consider how the same 2 X 2 covariation pattern is interpreted by 4 and 6 year old children to explain a cause.In the first study, two dolls played on two activities where their behavior covaried by doll or activity.Children were asked to explain the cause of the dolls' behavior and predict future behavior.In the second study, we presented the same covariation data but instead framed it in terms of physical causality, i.e., the dolls were treated as physical objects that either stuck or did not stick to the activities.Preliminary results reveal that children think differently about physical and psychological causes, even with the same covariation pattern.Additionally, they appear to reason about probabilistic data much more naturally in the psychological than the physical domain.

People can apparently make surprisingly sophisticated inductive inferences, despite the fact that there are constraints on cognitive resources that would make performing exact Bayesian inference computationally intractable. What algorithms could they be using to make this possible? We show that a simple sequential algorithm, Win-Stay, Lose-Shift (WSLS), can be used to approximate Bayesian inference, and is consistent with human behavior on a causal learning task. This algorithm provides a new way to understand people’s judgments and a new efficient method for performing Bayesian inference.

A Simple Sequential Algorithm for Approximating Bayesian Inference Elizabeth Bonawitz, Stephanie Denison, Annie Chen, Alison Gopnik, & Thomas L. Griffiths {liz b, smdeniso, c annie, gopnik, tom griffiths}@berkeley.edu Department of Psychology, 5427 Tolman Hall Berkeley, CA 94720 USA Abstract People can apparently make surprisingly sophisticated in- ductive inferences, despite the fact that there are constraints on cognitive resources that would make performing exact Bayesian inference computationally intractable. What algo- rithms could they be using to make this possible? We show that a simple sequential algorithm, Win-Stay, Lose-Shift (WSLS), can be used to approximate Bayesian inference, and is consis- tent with human behavior on a causal learning task. This al- gorithm provides a new way to understand people’s judgments and a new efficient method for performing Bayesian inference. Keywords: Bayesian inference; algorithmic level; causal learning Introduction In the last five years a growing literature has demon- strated that people often act in ways consistent with optimal Bayesian models (e.g., Griffiths & Tenenbaum, 2005; Good- man, Tenenbaum, Feldman, & Griffiths, 2008). These ap- proaches have provided a precise framework for characteriz- ing intuitive theories and have provided an account of how a learner should update her beliefs as evidence is acquired (Gopnik & Schulz, 2004; Griffiths & Tenenbaum, 2009). The theory-based Bayesian approach has met with much success in describing the inferences made by adults (for a review see Tenenbaum, Griffiths, & Kemp, 2006) and research in cog- nitive development suggests that children can make similar inferences (Gopnik & Schulz, 2007; Gopnik et al., 2004; Gweon, Schulz, & Tenenbaum, 2010; Kushnir & Gopnik, 2007; Schulz, Bonawitz, & Griffiths, 2007). Taken together, this research strongly suggests that Bayesian statistics pro- vides a productive starting point for understanding human in- ductive inference. Theory-based Bayesian approaches have typically been used to give a “computational level” (Marr, 1982) analysis of the inferences people make when solving inductive prob- lems, focusing on the form of the computational problem and its ideal solution. However, it need not be the case that the algorithms people are using to solve these problems actually resemble exact Bayesian inference. Indeed, given the compu- tational complexity of exact Bayesian inference (Russell & Norvig, 2002) and the numerous findings that children and adults alike have difficulty with explicit hypothesis testing (e.g., Kuhn, 1989; Klahr, Fay, & Dunbar, 1993) and some- times only slowly progress from one belief to the next (Carey, 1991; Wellman, 1990), it becomes interesting to ask how learners might be behaving in a way that is consistent with Bayesian inference. Here we investigate the algorithms that learners might be using in solving a particular kind of inductive problem – causal learning. These algorithms need to approximate Bayesian inference, but also need to be computationally tractable. One strategy that has proven effective for approx- imating Bayesian inference in computer science and statis- tics is using sampling-based approximations, also known as Monte Carlo methods. We introduce a new sequential sam- pling algorithm based on the Win-Stay, Lose-Shift (WSLS) principle, in which a learner maintains a particular hypothe- sis until receiving evidence that is inconsistent with that hy- pothesis. We show that this WSLS algorithm approximates Bayesian inference, and can do so quite efficiently. Previous work in cognitive psychology has shown that peo- ple follow a WSLS strategy in concept learning tasks (Restle, 1962; Levine, 1975). We use this as the starting point for an investigation of whether human behavior that approximates Bayesian inference in causal learning might be explained in terms of a WSLS strategy. We compare the WSLS algo- rithm to simply sampling from the posterior distribution as an account of human behavior in a simple causal learning task. Both algorithms predict that people should behave in a way that is consistent with Bayesian inference, but WSLS also predicts that there should be a characteristic pattern of dependency between people’s successive responses. The plan of the paper is as follows. First, we introduce the causal learning task that will be the focus of our analysis, and summarize how Bayesian inference can be applied in this task. We then introduce the idea of sequential sampling algo- rithms, including our new WSLS algorithm. This is followed by a mathematical analysis of the WSLS algorithm, show- ing that it approximates Bayesian inference. The remainder of the paper focuses on an experiment in which we evaluate how well the WSLS algorithm captures people’s judgments in our causal learning task. Bayesian inference and causal learning While the algorithms that we present in this paper will apply to any inductive problem with a discrete hypothesis space, we will make our analysis concrete by focusing on a simple causal learning problem. In this problem, there are three cat- egories of objects: red, green, and blue blocks. Each of these kinds of blocks activate a machine with different probability when they are placed on the machine. The red blocks acti- vate the machine on five out of six trials, the green blocks on three out of six trials, and the blue blocks on just once out of six trials. A new block is then presented, which has lost its color, and needs to be classified as either a red, green, or blue block, based on some observations of what happens when it is placed on the machine over a series of trials. The question

Cross-domain variation in children’s causal reasoning Elizabeth Seiver UC Berkeley Alison Gopnik UC Berkeley Abstract: How do children use covariation information to infer causation? In this series of studies, we consider how the same 2 X 2 covariation pattern is interpreted by 4 and 6 year old children to explain a cause. In the first study, two dolls played on two activities where their behavior covaried by doll or activity. Children were asked to explain the cause of the dolls’ behavior and predict future behavior. In the second study, we presented the same covariation data but instead framed it in terms of physical causality, i.e., the dolls were treated as physical objects that either stuck or did not stick to the activities. Preliminary results reveal that children think differently about physical and psychological causes, even with the same covariation pattern. Additionally, they appear to reason about probabilistic data much more naturally in the psychological than the physical domain.

We used a new method to assess how people can infer unobserved causal structure from patterns of observed events. Participants were taught to draw causal graphs, and then shown a pattern of associations and interventions on a novel causal system. Given minimal training and no feedback, participants in Experiment 1 used causal graph notation to spontaneously draw structures containing one observed cause, one unobserved common cause, and two unobserved independent causes, depending on the pattern of associations and interventions they saw. We replicated these findings with less-informative training (Experiments 2 and 3) and a new apparatus (Experiment 3) to show that the pattern of data leads to hidden causal inferences across a range of prior constraints on causal knowledge.

Combining findings on evolution, maturation, socialization, and culture, Konner offers a broad-based explanation of how biology underlies "normally developing social behavior."

This paper has to do with Bruner’s work in Oxford and his contribution to developmental psycholinguistics. Three themes in Bruner’s work are identified – constructivism, functionalism, and an emphasis on the social context of knowledge. These three themes are exemplified by his stress on children’s active construction of linguistic structure, his development of models of developmental pragmatics, and his interest in the effects of mother-infant interaction on language acquisition. These three themes have also characterized his intellectual life, including his creative research programs, his engagement in social policy issues, and his influence as a teacher.

Ambiguous figures, such as the Necker Cube or ‘duck-rabbit,’ are a special class of images that can give rise to multiple interpretations. Traditionally, switching between the possible interpretations of an ambiguous figure, or reversing one's interpretation, has been attributed to either top-down or bottom-up processes (e.g., either attributed to having knowledge of the nature of the ambiguity or attributed to a form of neuronal fatigue). However, here we present evidence that is incompatible with both sorts of explanations. Five- to nine-year old observers participated in four tasks - an ambiguous figures reversal task, two ‘Theory of Mind’ tasks, and a Piagetian number conservation task (the last being used as a measure of general cognitive abilities). Going against purely top-down explanations, one third of the observers reversed the ambiguous figures when completely uninformed about the ambiguity. Further, going against purely bottom-up explanations, those children who made these ‘spontaneous’ reversals were more likely to succeed on a high-order theory of mind task, even when factoring out general cognitive abilities. These findings suggest that reversing between the possible percepts of a bi-stable image can occur spontaneously and raise the possibility of there being necessary, but not sufficient, cognitive conditions for reversals to occur.

Children learn causal relationships quickly, and make farreaching causal inferences on the basis of what they see. In knowledge to bear on their problems. This paper addresses children’s ability to acquire that knowledge. We present evidence that children can learn about the abstract properties of causal relationships using only a handful of events, and – consistent with a hierarchical Bayesian model of casual inference – children can be more sensitive to evidence than adults.

In laboratory experiments, infants can learn patterns of
\nfeatures that co-occur (e.g., Fiser & Aslin, 2002). This finding
\nleaves two questions unanswered: What do infants do with the
\nknowledge acquired from such statistical learning, and which
\npatterns do infants attend to in the noisy and cluttered world
\noutside of the laboratory? Here, we show that 9-month-old
\ninfants form expectations about co-occurring features
\nremaining fused, an essential skill for object individuation and
\nrecognition (e.g., Goldstone, 2000; Schyns & Rodet, 1997).
\nWe also show that though social cues may temporarily detract
\nattention away from learning events, they appear to stimulate
\ninfants to display learning better in complex situations than
\nwhen infants learn on their own without attention cues. These
\nfindings suggest that infants can use feature co-occurrence to
\nlearn about objects and that social cues shape such
\nfoundational learning in a noisy environment during infancy.

Researchers in both educational and developmental psychology have suggested that children are not particularly adept hypothesis testers, and that their behavior can often appear irrational.However, a growing body of research also suggests that people do engage in rational inference on a variety of tasks.Recently researchers have begun testing the idea that reasoners may be sampling hypotheses from an internal probability distribution when making inferences.If children are reasoning in this way, this might help to explain some seemingly irrational behavior seen in previous experiments.Forty 4-year-olds were tested on a probabilistic inference task that required them to make repeated guesses about which of two types of blocks had been randomly sampled from a population.Results suggest that children can sample from a probability distribution as evidenced by the fact that, as a group, they engaged in probability matching and that the dependency between successive guesses decreased over time.

Children are ubiquitous imitators, but how do they decide which actions to imitate?One possibility is that children might learn which actions are necessary to reproduce by observing the contingencies between action sequences and outcomes across repeated observations.We define a Bayesian model that predicts that children will decide whether to imitate part or all of a sequence based on the pattern of statistical evidence.To test this prediction, we conducted an experiment in which preschool children watched an experimenter repeatedly perform sequences of varying actions followed by an outcome.Children's imitation of sequences that produced the outcome increased, in some cases resulting in production of shorter sequences of actions that the children had never seen performed in isolation.This behavior is consistent with our model's predictions, and suggests that children attend to statistical evidence in deciding which actions to imitate, rather than obligately imitating successful actions.

Abstract Recent work in rational probabilistic modeling suggests that a kind of propositional reasoning is ubiquitous in cognition and especially in cognitive development. However, there is no reason to believe that this type of computation is necessarily conscious or resource-intensive.

The central hypothesis of the work is that the dimension of control-no control plays an important role in motion sickness. Although it is generally agreed that having control over a moving vehicle greatly reduces the likelihood of motion sickness, few studies have addressed this issue directly, and the theoretical explanation for this phenomenon is not completely clear. In this study, we equated groups differing in controllability for head movement, vision, activity, and predictability, which have often been suggested in the literature as explanations for the driver's immunity to motion sickness. Twenty-two pairs of yoked subjects were exposed to nauseogenic rotation. One subject of each pair had control over the rotation and head movements, while the other was exposed passively to the same motion stimulus. Subjects who had control reported significantly fewer motion sickness symptoms and less of a decrement in their well-being, as compared to the yoked subject without control. The results are discussed in relation to Reason's sensory rearrangement theory and the concept of feed-forward mechanisms in motion perception.

Philosophers and Buddhist scholars have noted the affinities between David Hume's empiricism and the Buddhist philosophical tradition. I show that it was possible for Hume to have had contact with Buddhist philosophical views. The link to Buddhism comes through the Jesuit scholars at the Royal College of La Flèche. Charles François Dolu was a Jesuit missionary who lived at the Royal College from 1723-1740, overlapping with Hume's stay. He had extensive knowledge both of other religions and cultures and of scientific ideas. Dolu had had first-hand experience with Theravada Buddhism as part of the second French embassy to Siam in 1687-1688. In 1727, Dolu also had talked with Ippolito Desideri, a Jesuit missionary who visited Tibet and made an extensive study of Tibetan Buddhism from 1716-1721. It is at least possible that Hume heard about Buddhist ideas through Dolu.

Philosophers and Buddhist scholars have noted the affinities between David Hume's empiricism and the Buddhist philosophical tradition. I show that it was possible for Hume to have had contact with Buddhist philosophical views. The link to Buddhism comes through the Jesuit scholars at the Royal College of La Fleche. Charles Francois Dolu was a Jesuit missionary who lived at the Royal College from 1723-1740, overlapping with Hume's stay. He had extensive knowledge both of other religions and cultures and of scientific ideas. Dolu had had first-hand experience with Theravada Buddhism as part of the second French embassy to Siam in 1687-1688. In 1727, Dolu also had talked with Ippolito Desideri, a Jesuit missionary who visited Tibet and made an extensive study of Tibetan Buddhism from 1716-1721. It is at least possible that Hume heard about Buddhist ideas through Dolu.

Abstract Block argues for a method and a substantive thesis – that consciousness overflows accessibility. The method can help answer the question of what it is like to be a baby. Substantively, infant consciousness may be accessible in some ways but not others. But development itself can also add important methodological tools and substantive insights to the study of consciousness.

This study examines preschoolers' causal assumptions about spatial contiguity and how these assumptions interact with new evidence in the form of conditional probabilities. Preschoolers saw a toy that activated in the presence of certain objects. Children were shown evidence for the toy's activation rule in the form of patterns of probability: The toy was more likely to activate either when objects made contact with its surface (on condition) or when objects were several inches above its surface (over condition). In Experiment 1, 61 three-year-olds saw a deterministic activation rule. In Experiments 2 and 3, 48 four-year-olds saw an activation rule that was probabilistic. In Experiment 4, 30 four-year-olds saw a screening-off pattern of activation. In all 4 experiments, children used new evidence in the form of patterns of probability to make accurate causal inferences, even in the face of conflicting prior beliefs about spatial contiguity. However, children were more likely to make correct inferences when causes were spatially contiguous, particularly when faced with ambiguous evidence.

Four experiments examined children's inferences about the relation between objects' internal parts and their causal properties. In Experiment 1, 4-year-olds recognized that objects with different internal parts had different causal properties, and those causal properties transferred if the internal part moved to another object. In Experiment 2, 4-year-olds made inferences from an object's internal parts to its causal properties without being given verbal labels for objects or being shown that insides and causal properties covaried. Experiment 3 found that 4-year-olds chose an object with the same internal part over one with the same external property when asked which object had the same causal property as the target (which had both the internal part and external property). Finally, Experiment 4 demonstrated that 4-year-olds made similar inferences from causal properties to internal parts, but 3-year-olds relied more on objects' external perceptual appearance. These results suggest that by the age of 4, children have developed an understanding of a relation between an artifact's internal parts and its causal properties.

Abstract This chapter starts from the premise that much of children's knowledge takes the form of abstract, coherent, causal claims that are learned from, and defeasible by, evidence. This view is consistent with an interventionist view of causal knowledge, formalized in computational models using causal Bayes net representations. The chapter reviews empirical studies suggesting that, consistent with this account, preschoolers use patterns of evidence to: a) create novel, effective interventions; b) infer the structure of causal relationships, including relationships involving unobserved causes; c) accurately predict distinct outcomes from observed evidence and evidence generated by interventions; d) integrate novel evidence with prior beliefs; and e) distinguish informative interventions from confounded ones.

Abstract This chapter presents a dialogue that is a distillation of a real series of conversations which took place at that most platonic of academies — the Center for Advanced Studies in the Behavioral Sciences — between the first and third authors. The second author had determinist leanings to begin with and acted as an intermediary.

The conditional intervention principle is a formal principle that relates patterns of interventions and outcomes to causal structure. It is a central assumption of experimental design and the causal Bayes net formalism. Two studies suggest that preschoolers can use the conditional intervention principle to distinguish causal chains, common cause and interactive causal structures even in the absence of differential spatiotemporal cues and specific mechanism knowledge. Children were also able to use knowledge of causal structure to predict the patterns of evidence that would result from interventions. A third study suggests that children's spontaneous play can generate evidence that would support such accurate causal learning.

Ambiguous figures are a special class of images that can give rise to multiple interpretations. Traditionally, switching between the possible interpretations of an ambiguous figure, or reversing one's interpretation, has been attributed either to top-down or to bottom-up processes (e.g. attributed to having knowledge of the nature of the ambiguity, or to a form of neuronal fatigue). Here we present evidence that is incompatible with both forms of explanations. Observers aged 5-9 years can reverse ambiguous figures when uninformed about the ambiguity, negating purely top-down explanations. Further, those children who make these 'spontaneous' reversals are more likely to succeed on a high-order theory-of-mind task, negating purely bottom-up explanations.

Abstract We outline a new computational account of learning in children using the causal Bayes net formalism. We also present evidence that children as young as 2 years old use something like causal Bayes net learning mechanisms to infer the causal structure of the world around them. This kind of learning may play an important role in the development of intuitive theories. Finally, we contrast causal Bayes net and neural net learning mechanisms.

The present study investigates whether five-to-six-year-old children with Williams syndrome (N = 8) can form new object categories based on naming information alone, and compares them with five groups of typically developing children aged 2;0 to 6;0 (N = 34 children). Children were presented with triads of dissimilar objects; all objects in a triad were labelled, two of them with the same pseudoname. Name-based categorization was evaluated through object selection. Performance was above chance level for all groups. Performance reached a ceiling at about 4;0 for the typically developing children. For the children with Williams Syndrome, performance remained below chronological age level. The present results are discussed in light of previous findings of a failure to perform name-based categorization in younger children with Williams syndrome and the persistent asynchrony between cognitive and lexical development in this disorder.

We examine the interaction of two cues that children use to make judgments about cause-effect relations: probabilities and interventions. Children were shown a "detector" that lit up and played music when a block was placed on its surface. We varied the probabilistic effectiveness of the block, as well as whether the experimenter or the child was performing the interventions. In Experiment 1, we found that children can use probabilistic evidence to make inferences about causal strength. However, when the results of their own interventions are in conflict with the overall frequencies, preschoolers favor the results of their own interventions. In Experiment 2, children used probabilistic evidence to infer a hidden causal mechanism. Though they again gave preference to their own interventions, they did not do so when their interventions were explicitly confounded by an alternative cause.

Researchers in early social‐cognition have found that the ability to reverse an ambiguous figure is correlated with success on theory of mind tasks (e.g. ). The present experiment examined children with autistic spectrum disorders (ASD) without mental delay to see whether a similar relationship existed. demonstrated that children with ASD with mental delay were impaired on theory of mind tasks, but were as likely as mentally delayed controls to generate both interpretations of an ambiguous figure when informed of its ambiguity. The present study replicated this finding on children with ASD without mental delay. However, overall perception of ambiguous figures was different. These children were less likely to spontaneously generate both interpretations of the figure, and more likely to perseverate on a single interpretation than the comparison children. Like Ropar et al. , we found no correlation between theory of mind and informed reversals, but spontaneous reversals were correlated with performance on an advanced theory of mind task. These data suggest that reversals of ambiguous figures are linked to higher‐level representational abilities, which might also be involved in social functioning, and impaired in children with ASD.

We propose a causal Bayesian model of false belief reasoning in children. This model realizes theory of mind as the rational use of intuitive theories and supports causal prediction, explanation, and theory revision. The model undergoes an experience-driven false belief transition. We investigate the relationship between prediction, explanation, and surprise; this is used to interpret an empirical study of children’s explanations in an extension of the false belief task. Our study includes the standard outcome, surprising to younger children, and a novel “Psychic Sally ” condition that challenges older children with an unexpected outcome. In everyday life, humans constantly attribute unobservable mental states to one another, and use them to

Three studies investigated children's belief in causal determinism.If children are causal determinists, they should infer the existence of unobserved causes whenever effects occur stochastically.In Experiment 1, preschoolers saw a stochastic generative cause and inferred the existence of an unobserved inhibitory cause.In Experiment 2, preschoolers appropriately traded-off inferences about the presence of unobserved inhibitory causes and the absence of generative causes.Experiment 3 suggested that children resist inferring stochastic causation even though they accept non-causal stochastic relationships.Children's belief in determinism seems to support inferences about causal relations not obtainable from other cues

Five studies investigated (a) children's ability to use the dependent and independent probabilities of events to make causal inferences and (b) the interaction between such inferences and domain-specific knowledge. In Experiment 1, preschoolers used patterns of dependence and independence to make accurate causal inferences in the domains of biology and psychology. Experiment 2 replicated the results in the domain of biology with a more complex pattern of conditional dependencies. In Experiment 3, children used evidence about patterns of dependence and independence to craft novel interventions across domains. In Experiments 4 and 5, children's sensitivity to patterns of dependence was pitted against their domain-specific knowledge. Children used conditional probabilities to make accurate causal inferences even when asked to violate domain boundaries.

The authors outline a cognitive and computational account of causal learning in children. They propose that children use specialized cognitive systems that allow them to recover an accurate "causal map" of the world: an abstract, coherent, learned representation of the causal relations among events. This kind of knowledge can be perspicuously understood in terms of the formalism of directed graphical causal models, or Bayes nets. Children's causal learning and inference may involve computations similar to those for learning causal Bayes nets and for predicting with them. Experimental results suggest that 2- to 4-year-old children construct new causal maps and that their learning is consistent with the Bayes net formalism.

Popper had a fateful meeting with the philosopher of language Ludwig Wittgenstein at the Cambridge Philosophy Club. In a talk to the Club, with Wittgenstein in the audience, Popper described several “philosophical problems”– important, dif1⁄2cult questions that he thought would one day be answered. Here Popper was issuing a direct challenge to Wittgenstein, who had argued that philosophy could only analyze linguistic puzzles–not solve any real problems. The visit has become most famous for the subsequent controversy among eyewitnesses over whether or not Wittgenstein’s response to this challenge was to angrily brandish a 1⁄2replace poker at Popper. But there is a more interesting aspect to the story. One of the problems Popper described was the problem of causal induction: How is it possible for us to correctly infer the causal structure of the world from our limited and fragmentary experience? Popper claimed that this problem would one day be solved, and he turned out to be right. Surprisingly, at least part of the solution to the problem comes from a source about as far removed from the chilly Cambridge seminar room of 1⁄2fty years ago as possible–it comes from babies and young children. The past thirty years have been a golden age for the study of cognitive development. We’ve learned more about what babies and young children know, and when they know it, than we did in the preceding two thousand years. And this new science has completely overturned traditional ideas about what children are like. The conventional wisdom, from Locke to Freud and Piaget, had been that babies and young children are irrational, egocentric, pre-causal, and solipsistic, governed by sensation rather than reason, and impulse rather than intention. In contrast, the last thirty years of research have taught us that even the youngest infants–literally newborns –already know a great deal about a wide range of subjects. Moreover, we have been able to chart consistent changes

Abstract Previous research suggests that children can infer causal relations from patterns of events. However, what appear to be cases of causal inference may simply reduce to children recognizing relevant associations among events, and responding based on those associations. To examine this claim, in Experiments 1 and 2, children were introduced to a “blicket detector,” a machine that lit up and played music when certain objects were placed upon it. Children observed patterns of contingency between objects and the machine's activation that required them to use indirect evidence to make causal inferences. Critically, associative models either made no predictions, or made incorrect predictions about these inferences. In general, children were able to make these inferences, but some developmental differences between 3‐ and 4‐year‐olds were found. We suggest that children's causal inferences are not based on recognizing associations, but rather that children develop a mechanism for Bayesian structure learning. Experiment 3 explicitly tests a prediction of this account. Children were asked to make an inference about ambiguous data based on the base rate of certain events occurring. Four‐year‐olds, but not 3‐year‐olds were able to make this inference.

Alternative splicing plays critical roles in many disease processes and splicing dysregulation is a hallmark of cancer. The different splicing isoforms may have significantly different effects on the malignant progression of cancer. Checkpoint kinase 1 (CHK1) is a serine/threonine kinase and regulates DNA damage response. In this study, we measured the expression of an alternative CHK1 transcript (CHK1-S, excluded exon 3) in hepatocellular carcinoma (HCC) tissues. Our results showed that CHK1-S was significantly upregulated in HCC tissues compared with paired adjacent noncancerous hepatic tissues. The levels of full-length CHK1(CHK1-L), CHK1-S and the ratio of CHK1-S/L in tumor tissue were associated with relapse free survival (RFS) of postoperative HCC patients, respectively, but not the levels of CHK1-L, CHK1-S and the ratio of CHK1-S/L in adjacent normal tissue. To further demonstrate the role of CHK1-S in HCC, CCK-8 assays, EdU incorporation assays and colony formation assays were used. The results showed that overexpression of CHK1-S significantly accelerated HCC cell proliferation, compared with CHK1-L. In addition, we found that serine-arginine protein kinase 1 (SRPK1), as an upstream regulator kinase of splicing factor, could upregulate the expression of CHK1-S and its expression level was significantly higher in HCC tumors than the paired normal tissues and was associated with the levels of CHK1-S (P=0.016). In conclusion, our study demonstrated that CHK1-S, acts as an oncogene, which was upregulated and associated with RFS in HCC patients. SRPK1 may mediate its mRNA splicing in HCC. All these data indicated that the expression of CHK1-S would have potential prognostic values and splicing kinase SRPK1 might be developed as therapeutic target in HCC.

One of the deepest and most ancient problems in philosophy is what we might call the problem of knowledge. There seems to be an unbridgeable gap between our abstract, complex, highly structured knowledge of the world, and the concrete, limited and confused information provided by our senses. Since the Meno, there have been two basic ways of approaching this problem, rationalism and empiricism. Each era seems to have its matched pair of advocates of each view, making their way through the centuries like couples in some eternal philosophical gavotte, Plato and Aristotle, Descartes and Locke, Leibniz and Berkeley, Kant and Mill. The rationalist approach says that although it looks as if we learn about the world from our experience, we don’t really. Actually, we knew about it all along. The most important things we know were there to begin with, planted innately in our minds by God or evolution (or chance). The empiricist approach says that although it looks as if our knowledge is far removed from our experience, it isn’t really. If we rearrange the elements of our experience in particular ways, by associating ideas, or putting together stimuli and responses, we’ll end up with our knowledge of the world. There is both a tension and a kind of complementarity between these two ideas,

Abstract For some 25 years, the prevailing theories of categorization in philosophy have invoked the idea of “kinds” (Kripke, 1972; Putnam, 1975). When we look at how adults use words to refer to categories of things, we find that they only rarely categorize objects on the basis of their common properties. Instead, adults seem to categorize objects together when they believe that they belong to the same “kind,” that is, that they share some common, abstract “essence.” Psychological investigations of adults have largely confirmed these philosophical intuitions; adults do seem to group objects together based on “kinds” rather than properties (Murphy & Medin, 1985; Rips, 1989). Several investigators, particularly Gelman and her colleagues (Gelman & Coley, 1990; Gelman & Markman, 1986), have argued that children as young as 2 years old also categorize and name objects based on kinds (see also Mandler & McDonough, 1993; Mervis & Bertrand, 1994; Soja, Carey, & Spelke, 1991). Other investigators, in contrast, have suggested that young children categorize and name objects on the basis of perceptual properties such as shape (Imai, Gentner, & Uchida, 1994; Landau, Smith, & Jones, 1988). However, the question of whether children use names to refer to kinds has been bedeviled by the problem that the philosophical notion is not easily translatable into empirical predictions. In the philosophical literature, the concept of kinds refers to the idea that members of a category share an “essence,” a common, abstract, ontological property, and that names refer to this essence (Putnam, 1975). Obviously, it is difficult to ask 3-year-olds if they conceive of objects or use names in this way.

One of the important aspects of human causal reasoning is that from the time we are young children we reason about unobserved causes.How can we learn about unobserved causes from information about observed events?Causal Bayes nets provide a formal account of how causal structure is learned from a combination of associations and interventions.This formalism makes specific predictions about the conditions under which learners postulate hidden causes.In this study adult learners were shown a pattern of associations and interventions on a novel causal system.We found that they were able to infer hidden causes as predicted by the Bayes net formalism, and were able to distinguish between one hidden common cause and two hidden independent causes of the observed events.

序章 人はいつから、「他者の心」「外の世界」「言葉」を知るのか 第１章 子どもが「人」について学ぶこと 第２章 子どもが「もの」について学ぶこと 第３章 子どもが言葉について学ぶこと 第４章 科学者が子どもの心について学んだこと 第５章 科学者が子どもの脳について学んだこと 終章 子どもたちの未来

The Cognitive Basis of Science concerns the question 'What makes science possible?' Specifically, what features of the human mind and of human culture and cognitive development permit and facilitate the conduct of science? The essays in this volume address these questions, which are inherently interdisciplinary, requiring co-operation between philosophers, psychologists, and others in the social and cognitive sciences. They concern the cognitive, social, and motivational underpinnings of scientific reasoning in children and lay persons as well as in professional scientists. The editors' introduction lays out the background to the debates, and the volume includes a consolidated bibliography that will be a valuable reference resource for all those interested in this area. The volume will be of great importance to all researchers and students interested in the philosophy or psychology of scientific reasoning, as well as those, more generally, who are interested in the nature of the human mind.

In this chapter, we outline a more precise cognitive and computational account of the 'theory theory' of cognitive development. Theories and theory-formation processes are cognitive systems that allow us to recover an accurate 'causal map' of the world: an abstract, coherent representation of the causal relations among events. This kind of knowledge can be perspicuously represented by the formalism of directed graphical causal models, or 'Bayes nets'. Human theory formation may involve similar computations.

The relation between source monitoring and suggestibility was examined among preschool children. Thirty-two 3- to 5-year-olds were simultaneously presented with a brief story in two different modalities, as a silent video vignette and a spoken narrative. Each modality presented unique information about the story, but the information in the two versions was mutually compatible. The children were then asked a series of questions, including questions about the source (modality) of story details, and leading questions about story details (to assess suggestibility). Performance on the source-monitoring questions was highly correlated with the ability to resist suggestion. In addition, children who were asked source-monitoring questions prior to leading questioning were less susceptible to suggestion than were those who were asked the leading questions first. This study provides evidence that source monitoring can play a causal role in reducing the suggestibility of preschool children.

If there is one clear conclusion to be drawn from this volume it is that, after decades of obloquy, Benjamin Whorf is back. Of course, Whorf never really went away in the popular imagination or in the wilder reaches of the post-modern humanities. In serious cognitive psychology and cognitive science, though, that very fact reinforced the sense that Whorfian ideas were disreputable, not to say crackpot. In contrast to this scornful tradition, many of the chapters in this volume, both empirically and conceptually, seriously explore the possibility that the language we hear can have strong effects on the ways that we understand the world.

Three studies investigated whether young children make accurate causal inferences on the basis of patterns of variation and covariation. Children were presented with a new causal relation by means of a machine called the "blicket detector." Some objects, but not others, made the machine light up and play music. In the first 2 experiments, children were told that "blickets make the machine go" and were then asked to identify which objects were "blickets." Two-, 3-, and 4-year-old children were shown various patterns of variation and covariation between two different objects and the activation of the machine. All 3 age groups took this information into account in their causal judgments about which objects were blickets. In a 3rd experiment, 3- and 4-year-old children used the information when they were asked to make the machine stop. These results are related to Bayes-net causal graphical models of causal learning.

In two studies we investigated 3‐ to 5‐year‐old children’s ability to reverse ambiguous figures and the relation between this ability and the ability to understand multiple representations. Children never reversed before they were informed of the two alternative interpretations of the figures. Even when they were informed of the alternatives and understood that both were possible, 3‐year‐olds and most 4‐year‐olds did not reverse. In contrast, a majority of 5‐year‐olds did reverse. In general, children only reversed if they also passed a standard false belief task. However, there was a closer correlation between reversals and a ‘droodle’ task that involved an understanding of the ambiguity of perceptual representations. These results suggest that the immediate experience of reversal may depend on a more abstract understanding of ambiguous representations.

The Work of the Imagination . Paul L. Harris. Blackwell, Oxford, 2000. 234 pp. $59.95, £50. ISBN 0-631-21885-8. Paper, $26.95, £14.99. ISBN 0-631-21886-6. Refuting long-held views that the fantasies of children reflect primitive and irrational thought, Harris explores how imagination contributes to cognitive and emotional development.

The present study explores the ability of 3.5‐ and 4.5‐year‐olds to use a causal property (making a machine light up and play music) to build categories of objects, and attach a name to them. First, this use is assessed in the presence or absence of simple perceptual information (color and shape) leading to a conflicting categorization. Second, the role of language is evaluated by varying how the experimenter describes the actions of the objects on the machine. Results show that although children of both ages use causal properties to categorize objects in the absence of conflicting perceptual categorization, older children are more likely than younger children to favor the causal over the perceptual categorization when they conflict. An effect of language was also found with the older children, with explicit causal descriptions of the events enhancing causal categorizations. Finally, a memory probe showed that younger children were likely to misremember causal information when it conflicted with perceptual information. These results suggest that perceptual, linguistic and causal information are all correlated for the younger children, whereas these cues are more independent for the older children.

Three studies explored whether and when children could categorize objects on the basis of a novel underlying causal power. To test this we constructed a "blicket detector," a machine that lit up and played music when certain objects were placed on it. First, 2-, 3- and 4-year-old children saw that an object labeled as a "blicket" would set off the machine. In a categorization task, other objects were demonstrated on the machine. Some set it off and some did not. Children were asked to say which objects were "blickets." In an induction task, other objects were or were not labeled as "blickets." Children had to predict which objects would have the causal power to set off the machine. The causal power could conflict with perceptual properties of the object, such as color and shape. In an association task the object was associated with the machine's lighting up but did not cause it to light up. Even the youngest children sometimes used the causal power to determine the object's name rather than using its perceptual properties and sometimes used the object's name rather than its perceptual properties to predict the object's causal powers. Children rarely categorized the object on the basis of the associated event. Young children also sometimes made interesting memory errors-they incorrectly reported that objects with the same perceptual features had had the same causal power. These studies demonstrate that even very young children will easily and swiftly learn about a new causal power of an object and spontaneously use that information in classifying and naming the object.

Learning begins in the first days of life. Scientists are now discovering how young children develop emotionally and intellectually, and are beginning to realize that from birth babies already know a staggering amount about the world around them. In the first book of its kind for a popular audience, three leading US scientists draw on twenty-five years of research in philosophy, psychology, computer science, linguistics and neuroscience to reveal what babies know and how they learn it.

Field-flow fractionation is an analytical technique that allows the separation of particles over a size range, from a few nanometers to several microns in diameter. The separation takes place under mild conditions and is suited for the analysis of neutral or charged particles. A single measurement yields the size and concentration of each component of a mixture. However, developing a suitable fractionation method can be tedious and time-consuming. In this chapter, we present asymmetrical flow field-flow fractionation (AF4) conditions that have proven their reliability for the analysis of quantum dots and other nanoparticles in the 5-50 nm size range. Common pitfalls are emphasized together with strategies to overcome them.

Abstract In the first edition of this book we presented a hypothesis about the early development of children’s understanding of the mind (Meltzoff and Gopnik 1993).We suggested that normally-developing children are born linking their own minds and the minds of others. This ability is most clearly evidenced in the innate capacity to imitate facial expressions. This primitive link means that, later on in development, further discoveries about other people’s minds are easily and naturally extended to knowledge of our own minds, and vice versa. We suggested that this initial link between the self and others may be missing in at least some cases of autism.

This book combines two worlds -- children and science -- in an entirely unique way that yields exciting discoveries about both. authors show that by the time children are three, they've solved problems that stumped Socrates with an agility computers still can't match. The Scientist in the Crib explains just how, and how much, babies and young children know and learn, and how much parents naturally teach them. In fact, The Scientist in the Crib argues that evolution designed us to both teach and learn. Nurture is our nature, and the drive to learn is our most important instinct. new science of children also reveals insights about our adult capacities, helping to solve some ancient questions: How do we know there really is a world out there? How do we know that other people have minds like ours? It turns out that we find solutions to these problems when we are very small. But these astonishing capabilities don't disappear in later life, as the authors show in their engaging discussion of humans' potential for learning. In fact, they argue that even very young children -- as well as adults use some of the same methods that allow scientists to learn so much about the world. Written by three top scientists -- themselves parents -- who conducted much of the pioneering research in this field, The Scientist in the Crib is vivid, lucid, and often funny. Filled with surprises at every turn, it gives us a new view of the inner life of children and the mysteries of the mind.

I suggest several ways that externalism could influence psychological theorizing. Externalism could just capture our everyday intuitions about concepts and meanings. More profoundly, it could enter into psychology through evolutionary theory, guide our hypotheses about conceptual abilities, and, most significantly, it could influence our accounts of learning and conceptual change.

Abstract The articles by Lillard and by Joseph are an example of how apparently divergent empirical results may turn out to reflect interesting differences between children and adults.

Although there is substantial evidence that 30-month-old children can reason about other people's desires, little is known about the developmental antecedents of this ability. A food-request procedure was devised to explore this understanding in 14- and 18-month-olds. Children observed an experimenter expressing disgust as she tasted 1 type of food and happiness as she lasted another type of food. They were then required to predict which food the experimenter would subsequently desire. The 14-month-olds responded egocentrically, offering whichever food they themselves preferred. However, 18-month-olds correctly inferred that the experimenter wanted the food associated with her prior positive affect. They were able to make this inference even when the experimenter's desires differed from their own. These data constitute the first empirical evidence that 18-month-olds are able to engage in some form of desire reasoning. Children not only inferred that another person held a desire, but also recognized how desires are related to emotions and understood something about the subjectivity of these desires.

The other Socratic method: Socrates's problem Augustine's problem a road map. Part 1 The theory theory: the scientist as child - but surely it can't really be a theory? a cognitive view of science, naturalistic epistemology and development - an evolutionary speculation, science as horticulture, objections - sociology, objections - timing and convergence, objections - magic, empirical advances, what is theory? structural features of theories, functional features of theories, dynamic features of theories, theories in childhood, theories as representations theories, modules, and empirical generalizations - modules, theories and development, modules and development, modularity in peripheral and central processing, empirical generalization - scripts, narratives, and nets, interactions among theories, modules, and empirical generalizations, nonconceptual development - information processing and social construction. Part 2 Evidence for the theory theory: the child's theory of appearances - the adult theory, the initial theory, the paradox of invisible objects, an alternative - a theory-change account, the nine-month-old's theory, the A-not-B error as an auxiliary hypothesis, the 18-month-old's theory, other evidence for the theory theory, semantic developments - from objects permanence to perspective taking, later semantic developments - gone and see, conclusion the child's theory of action - the adult theory, the initial theory, the nine-month-old's theory, the 18-month-old's theory, other evidence for the theory theory, semantic development - no, uh-oh and there, later developments - from actions to desires, later semantic developments - want, conclusion the child's theory of kinds - the adult theory, categories and kinds, the initial theory, the nine-month-old's theory, the 18-month-old's theory , other evidence for the theory theory, semantic development - the naming Spurt, later developments, conclusion. Part 3 Theories and language: language and thought - prerequisites, interactions, a theory-theory view, methodological issues - specificity and correlation, developmental relations between language and cognition, theories and constraints, crosslinguistic studies, individual-difference studies, conclusion the Darwinian conclusion - who's afraid of semantic holism? a developmental cognitive science, computational and neurological mechanisms, after Piaget, sailing in Neurath's boat.

Words, Thoughts, and Theories articulates and defends the "theory theory" of cognitive and semantic development, the idea that infants and young children, like scientists, learn about the world by forming and revising theories, a view of the origins of knowledge and meaning that has broad implications for cognitive science. Gopnik and Meltzoff interweave philosophical arguments and empirical data from their own and other's research. Both the philosophy and the psychology, the arguments and the data, address the same fundamental epistemological question: How do we come to understand the world around us? Recently, the theory theory has led to much interesting research. However, this is the first book to look at the theory in extensive detail and to systematically contrast it with other theories. It is also the first to apply the theory to infancy and early childhood, to use the theory to provide a framework for understanding semantic development, and to demonstrate that language acquisition influences theory change in children.The authors show that children just beginning to talk are engaged in profound restructurings of several domains of knowledge. These restructurings are similar to theory changes in science, and they influence children's early semantic development, since children's cognitive concerns shape and motivate their use of very early words. But, in addition, children pay attention to the language they hear around them and this too reshapes their cognition, and causes them to reorganize their theories. Bradford Books imprint

I and others have made a case for the similarities between scientific knowledge and children's knowledge, and between scientific change and cognitive development (Gopnik and Wellman, 1992, 1994; Gopnik, 1988; Carey, 1986, 1988; Keil, 1989; Wellman, 1990). In the rest of this chapter I will call this general view of cognitive development the 'theory-formation theory'.

This paper argues that there are powerful similarities between cognitive development in children and scientific theory change. These similarities are best explained by postulating an underlying abstract set of rules and representations that underwrite both types of cognitive abilities. In fact, science may be successful largely because it exploits powerful and flexible cognitive devices that were designed by evolution to facilitate learning in young children. Both science and cognitive development involve abstract, coherent systems of entities and rules, theories. In both cases, theories provide predictions, explanations, and interpretations. In both, theories change in characteristic ways in response to counterevidence. These ideas are illustrated by an account of children's developing understanding of the mind.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

Intuitive theories are defined as coherently interrelated systems of concepts that generate explanations and predictions in a particular domain of experience. 2 studies tested whether the child's theory of mind can be characterized as an intuitive theory. In Study 1, 3-year-old children who did not pass a false belief pretest were trained in 2 groups: (1) on the concept of belief, or (2) on the related concepts of desire and perception. Training took place over 2 weeks, with children given mental state tasks and receiving feedback according to their performance. Both training groups showed improved false belief performance on the posttest, compared to a control group trained on number conservation. This result is interpreted as demonstrating coherence in the child's theory of mind. In Study 2, these findings were repeated and expanded: training on belief as well as training on desire and perception resulted in improved performance on a variety of standard theory of mind posttests. Results are discussed with respect to competing theories of children's intuitive psychological knowledge.

Abstract In this response various possible objections to the critique of first-person-authority in Gopnik (1993), are considered. The heart of all three objections is that “beliefs” and other notions of commonsense psychology should not be construed as attempts to describe the functional character of our minds. Rather, they are 1) an evolutionarily determined technique for dealing with our conspecifics (Bogdan) or 2) the result of interpretation (Van Brakel) or 3) self-constitutive entities which exist only when we have the concept of their existence (Greve and Buchner). While the word belief might be construed in all these ways, they do not correspond either to the construals of our commonsense psychology or of a scientific psychology.

This cross-linguistic study investigates children's early lexical development in English and Korean, and compares caregivers' linguistic input in the two languages. In Study 1, the lexical development of nine Korean children was followed from 1;2 to 1;10 by monthly visits and maternal reports. These Korean data were compared to previously collected English longitudinal data. We find that: (1) Korean children as young as 1;3 use verbs productively with appropriate inflections. (2) Seven of the nine children show a verb spurt at around 1;7; for six of these children the verb spurt occurs before the noun spurt. No such early verb spurt is found in the English data. Unlike in English, both verbs and nouns in Korean are dominant categories from the single-word stage. (3) Korean children express language-specific distinctions of locative actions with verbs. Study 2, a crosslinguistic study of caregivers' input in English and Korean, shows that Korean mothers provide more action verbs but fewer object nouns than American mothers. Also, Korean mothers engage in activity-oriented discourse significantly more than American mothers. Our study suggests that verbs are accessible to children from the beginning, and that they may be acquired early in children who are encouraged to do so by their language-specific grammar and input.

In the 1946 film noir Lady in the Lake, the hero, Philip Marlowe, is represented by the camera's eye, and by a voice-over narration. We hear his thoughts and see his vision of others: the heroine's face coming into close-up for a kiss, the villain's scowl, his own feet upended as he falls to the ground from a Mickey Finn, his unshaven face briefly in a mirror after he regains consciousness.

This is a book about domain-specific cognition – the proposal that at least some human conceptual abilities are specialized for some types of contents and not for others. In this chapter we address the development of a single domain: everyday understanding of the mind. We suggest that this development is best understood as the formulation of a succession of naive theories. Moreover, this “theory theory” can help to characterize cognitive domains more generally and to explain domain-specific development. Our chapter, therefore, joins company with a number of recent discussions drawing parallels between theory change in science and cognitive development (Carey, 1985, 1988; Gopnik, 1984, 1988; Karmiloff-Smith & Inhelder, 1974; Keil, 1987; Wellman, 1985, 1990).

Most adult observers tend not to reverse ambiguous figures if they are not informed about the ambiguity of the figure. The question was asked whether very young children who have rarely, if ever, seen any kinds of ambiguous figures will reverse if uninformed, and also how they will behave if they are informed. It was found that 3 and 4 year olds never reverse when presented with two different kinds of ambiguous figures when uninformed, and only some do even when informed; further, those that do reverse do so only once or twice over a 60 s inspection period. These results are interpreted as further confirmation of an earlier finding with adults--that reversal is not simply a matter of prolonged inspection of an ambiguous figure leading automatically to neural satiation. Instead, cognitive factors such as utilization of memory and intention are implicated.

Abstract As adults we believe that our knowledge of our own psychological states is substantially different from our knowledge of the psychological states of others: First-person knowledge comes directly from experience, but third-person knowledge involves inference. Developmental evidence suggests otherwise. Many 3-year-old children are consistently wrong in reporting some of their own immediately past psychological states and show similar difficulties reporting the psychological states of others. At about age 4 there is an important developmental shift to a representational model of the mind. This affects children's understanding of their own minds as well as the minds of others. Our sense that our perception of our own minds is direct may be analogous to many cases where expertise provides an illusion of direct perception. These empirical findings have important implications for debates about the foundations of cognitive science.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

2 studies investigate whether 18-month-old children spontaneously sort objects into basic-level categories, and how this ability is related to naming. In Study 1, 18-month-old children were given spontaneous sorting tasks, involving both identical objects and objects with basic-level intracategory variation. Children were scored as having passed the tasks if they produced “exhaustive grouping,” that is, physically grouped all the objects of one kind into one location and the objects of the other kind into a different location. The children also received means-ends and object-permanence tasks. Children's parents received a checklist of early names. Children who produced exhaustive grouping used significantly more names than those who did not, in both identical and basic-level cases. There was no such relation between object-permanence and naming or between means-ends performance and naming. In Study 2, children received arrays of the same objects, with either identical objects or objects with basic-level variation in each group. No significant differences were found between the identical and basic-level tasks. However, as in the previous task, performance on both types of categorization was related to naming. Children who produced exhaustive grouping were reported to produce more names than those who did not. There appears to be a close relation between object categorization and naming in young children. The theoretical implications of this empirical association are discussed.

Peer Reviewed

Young children's ability to understand which experiences led to a belief was investigated in 3 studies. Three-, 4-, and 5-year-olds learned about the contents of a toy tunnel in three different ways : they saw the contents, were told about them, or felt them. Immediately after each trial, children were asked to state what was in the tunnel and also how they knew about the contents of the tunnel.

3‐year‐old children have difficulty reporting their past false beliefs. We investigated their ability to remember and report other types of past mental state, in particular, pretenses, images, perceptions, desires, and intentions. In a series of tasks, children were placed in one mental state, that state was changed, and they were asked to report the initial state. 4‐year‐olds were generally able to report all their past mental states, including beliefs. 3‐year‐olds were able to report past pretenses, images, and perceptions extremely well. They had great difficulty reporting past beliefs. Reporting past desires and intentions was more difficult than reporting pretenses, images, and perceptions, but slightly less difficult than reporting beliefs. The evidence suggests that 3‐year‐olds have difficulty understanding the nature of representation.

This is a state‐of‐the‐art review of children's understanding of the mind. It examines both empirical evidence and explanatory theories. Considering evidence, the key issue of children's understanding of false belief is first discussed. Then other evidence for children's folk psychological understanding is summarized, from late infancy to the end of the preschool period. Considering explanations, the dominant idea in the field, the ‘heory theory’, is discussed first. Then five other perspectives are considered: enculturation, nativism, folk psychology based on procedural knowledge, on experience, or on scripts. The positions, and the empirical predictions each would make, are compared. The paper concludes with a synthesis of the different views and a brief consideration of where the field is headed.

At least since Augustine, philosophers have constructed developmental just-so stories about the origins of certain concepts. In these just-so stories, philosophers tell us how children must develop these concepts. However, philosophers have by and large neglected the empirical data about how children actually do develop their ideas about the world. At best they have used information about children in an anecdotal and unsystematic, though often illuminating, way (see, for example, Matthews, 1980).

Semantic and cognitive development were studied in eight Korean and French-speaking children, and these results were compared with results for 12 English-speakers. The children received object- permanence, means-ends and categorization tasks and their develop ment of related linguistic forms, disappearance and success/failure words and a naming spurt, was recorded. There were close and specific relations between these semantic and cognitive developments. However, non-English-speakers used very different forms from English-speakers to encode disappearance, success and failure; in particular, Korean-speakers used verbs. This suggests that children may have been motivated to acquire these words because of their cognitive significance. Moreover, both categorization and naming developed later in the Korean-speakers than in English-speakers and French-speakers. This may be due to the fact that verbs are more perceptually salient in Korean than nouns. This difference between the language groups also suggests that linguistic development may influence and motivate cognitive development. Thus, there appears to be a thorough-going two-way interaction between language and cognition in this period.

This research concerns the development of children's understanding of representational change and its relation to other cognitive developments. Children were shown deceptive objects, and the true nature of the objects was then revealed. Children were then asked what they thought the object was when they first saw it, testing their understanding of representational change; what another child would think the object was, testing their understanding of false belief; and what the object looked like and really was, testing their understanding of the appearance-reality distinction. Most 3-year-olds answered the representational change question incorrectly. Most 5-year-olds did not make this error. Children's performance on the representational change question was poorer than their performance on the false-belief question. There were correlations between performance on all 3 tasks. Apparently children begin to be able to consider alternative representations of the same object at about age 4.

The early words that 27 children produced between the ages of 12 and 20 months were analysed. Social words, such as that used to point out objects or no used to refuse suggestions, were consistently among the first words to appear. A small number of names also appeared in the very early period. Cognitive-relational words, such as gone used to indicate disappearance or no used to indicate failure consistently appeared after some social words had been acquired. The cognitive development of 18 of these children was also recorded. Social words and names, though not cognitive-relational words, appeared before children solved 'early stage 6' cognitive problems. These results suggest that children move from using language communicatively to using it to sort out their cognitive problems early in the one-word stage and that this shift is related to 'stage 6' cognitive developments.

Alison Gopnik, Janet W. Astington, Children's Understanding of Representational Change and Its Relation to the Understanding of False Belief and the Appearance-Reality Distinction, Child Development, Vol. 59, No. 1 (Feb., 1988), pp. 26-37

GOPNIK, ALISON, and GRAF, PETER. Knowing How You Know: Young Children's Ability to Identify and Remember the Sources of Their Beliefs. CHILD DEVELOPMENT, 1988, 59, 1366-1371. Young children's understanding of the sources of their beliefs was investigated. 3, 4, and 5-year-olds learned about the contents of a drawer in 3 different ways: they saw the contents, were told about them, or inferred their identity from a clue. Children were then asked, immediately and after a brief delay, how they knew about the contents of the drawer. 3-year-olds had difficulty identifying the sources of their knowledge, while 5-year-olds did not. Moreover, even 3-year-olds who could correctly identify the source immediately had difficulty remembering the source after a delay. Explicit training in identifying sources did not improve the 3-year-olds' performance. These results support the hypothesis that children learn about the causal relation between the world and the mind between 3 and 5 years of age.

Alison Gopnik, Andrew Meltzoff, The Development of Categorization in the Second Year and Its Relation to Other Cognitive and Linguistic Developments, Child Development, Vol. 58, No. 6 (Dec., 1987), pp. 1523-1531

Alison Gopnik, Andrew Meltzoff, The Development of Categorization in the Second Year and Its Relation to Other Cognitive and Linguistic Developments, Child Development, Vol. 58, No. 6 (Dec., 1987), pp. 1523-1531

GOPNIK, ALISON, and MELTZOFF, ANDREW N. Relations between Semantic and Cognitive Development in the One-Word Stage: The Specificity Hypothesis. CHILD DEVELOPMENT, 1986, 57, 1040-1053. We develop a which we call the specificity hypothesis, concerning the interrelation between early linguistic and cognitive development. This hypothesis states that there are specific relations between particular linguistic and particular cognitive developments in the 1-word period. It suggests that the acquisition of words encoding disappearance is related to the development of the object concept, while the acquisition of words encoding success and failure is related to the development of means-ends understanding. The results of a longitudinal study of 19 children confirm this specificity hypothesis. The results show that children acquired disappearance words within a few weeks of their solution of a complex object-permanence task involving invisible displacements and acquired success/failure words within a few weeks of their solution of complex means-end tasks. The cross-relations did not hold; that is, there were no similarly close relations between disappearance words and the means-ends tasks or between success/failure words and the object-permanence tasks. These findings show that there are at least 2 specific relations between semantic and conceptual development in the 1-word stage.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

ABSTRACT The relationship between the acquisition of gone and the development of the object concept is investigated. Children acquire gone at the same time that they solve serial invisible displacement tasks, in a transitional period in the development of the object concept. According to a ‘neo-piagetian’ account of the object concept, children move from using specific rules to predict the reappearance of objects, to using a general theory of object movements during this period. The children's uses of gone suggest that gone encodes the fact that the child does not perceive an existing object. This concept is an essential component of the new theory of objects developed in this period. Children seem to acquire gone when they are in the midst of developing the concept that gone encodes, which may explain why they acquire the word in this period. In this case, conceptual development does not precede semantic development; rather, conceptual and semantic development appear to proceed simultaneously.

ABSTRACT Recent studies suggest a specific relationship between the acquisition of the word ‘gone’ and the development of the object concept. A longitudinal study investigated whether there is a similar relationship between the acquisition of words that encode success and failure, such as no, uh-oh and there , and the development of means–ends skills. A cross-sectional study investigated both these relationships. The results suggest that children acquire the word gone after they solve simple invisible displacement tasks but before they solve serial invisible displacement tasks. They also show that children acquire success/failure words after they use insight to solve the problem of the string but before they use insight to solve more difficult means–ends problems. These sequences cannot be explained in terms of age, general cognitive development or general linguistic development. It is argued that gone encodes a concept related to the concepts that underlie success on invisible displacement tasks, and that success/failure words encode concepts related to the concepts that underlie the development of insight. Children seem to acquire words that encode concepts they have just developed or are in the process of developing. These results suggest that there are strong and specific links between the acquisition of particular types of meanings and particular cognitive achievements. Several alternative explanations for these early links between semantic and conceptual development are proposed.

ABSTRACT Many early words encode aspects of plans, that is, actions performed in order to bring about events. Gone and down encode the fact that an action is intended to bring about a certain type of event. There and oh dear encode the success or failure of plans, no encodes the rejection of a plan. More encodes the repetition of a plan. It is argued that the concepts encoded by gone and down are like the concepts that underlie trial- and-error problem-solving and are developed after 12 months. The more abstract concepts encoded by there, oh dear, no and more are associated with the development of insight after 18 months. These early words are acquired because they are relevant to the child's cognitive problems.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

An abstract is not available for this content so a preview has been provided. Please use the Get access link above for information on how to access this content.

The meaning of expressions other than names in the language of 9 12-24-month olds was studied. The children were audiotaped or videotaped regularly from about 15 months to about 24 months. The meaning of each early word was discovered. by looking at the contexts in which the children used that word. Changes in the meaning of each word were recorded. 11That 11 , "there", "no", "more", "oh dear", "gone", "down", "up", "in", "out", "on" and "off" were studied particularly closely. "That" and/or "there" encodes the fact that the child perceives an object, that he successfully completes a plan, or that an object changes its location. "No" encodes the fact that the child does not successfully implement a plan, "more" that he intends to repeat a plan, and "oh dear" that his plan fails. "Gone" encodes the fact that the child does .not perceive an object, "down" and "up" encode downward and upward movement, "in" and "out" encode movement in to or out of a container, and "on" and "off" encode attachment and detachment. Children initially used these expressions in egocentric, action-oriented ways and later used them in more objective ways. The cognitive literature provides the basis for an account of the development of the concepts encoded by these expressions, particularly, the concepts of abstract aspects of plans, of disappearance, balance, containment and attachment. These concepts appear to be developed at around 18 months, that is,at about the same time that the words that encode them first appear. This suggests that early nonnominal expressions encode new, problematic, concepts, rather than well-established concepts. It is argued that the development of early non-nominal expressions could facilitate the development of the abstract concepts they encode. This argument may also be applied to early names.