Rational Higher-Order Belief Revision in Young Children

Preschoolers use probabilistic evidence to flexibly change or maintain expectations on an active search task

This study investigated 3- to 5-year-olds' (N = 64, 37 girls, 62.5% White, data collected between 2021-2022) ability to use probabilistic information gleaned through active search to appropriately change or maintain expectations. In an online fishing game, children first learned that one of two ponds was good for catching fish. During a subsequent testing phase, children searched the ponds for fish. Half saw outcomes that were probabilistically consistent with training, and the other half saw outcomes that were probabilistically inconsistent. Children in the Inconsistent condition adapted their search strategies, showing evidence of changing their expectations. Those in the Consistent condition maintained their initial search strategy. Trial-by-trial analyses suggested that children used a combination of heuristic and information integration strategies to guide their search behavior.

Do reflection prompts promote children's conflict monitoring and revision of misconceptions?

We tested whether reflection prompts enhance conflict monitoring and facilitate the revision of misconceptions. German children (N = 97, Mage = 7.20, 56% female) were assigned to a prediction or a prediction with reflection condition that included reflection prompts. Children in the prediction with reflection condition (1) showed greater error-related response times and pupil dilation responses, indicating better conflict monitoring, and (2) performed closer to an optimal Bayesian learner, indicating better monitoring-based control. However, by the end of the study, all children had similar levels of misconception revision. Thus, reflection prompts can enhance learning from anomalous evidence by improving conflict monitoring, but they may need to be repeated often to sustain their beneficial effects.

Don't throw the associative baby out with the Bayesian bathwater: Children are more associative when reasoning retrospectively under information processing demands

Causal reasoning is a fundamental cognitive ability that enables individuals to learn about the complex interactions in the world around them. However, the mechanisms that underpin causal reasoning are not well understood. For example, it remains unresolved whether children's causal inferences are best explained by Bayesian inference or associative learning. The two experiments and computational models reported here were designed to examine whether 5- and 6-year-olds will retrospectively reevaluate objects-that is, adjust their beliefs about the causal status of some objects presented at an earlier point in time based on the observed causal status of other objects presented at a later point in time-when asked to reason about 3 and 4 objects and under varying degrees of information processing demands. Additionally, the experiments and models were designed to determine whether children's retrospective reevaluations were best explained by associative learning, Bayesian inference, or some combination of both. The results indicated that participants retrospectively reevaluated causal inferences under minimal information-processing demands (Experiment 1) but failed to do so under greater information processing demands (Experiment 2) and that their performance was better captured by an associative learning mechanism, with less support for descriptions that rely on Bayesian inference. RESEARCH HIGHLIGHTS: Five- and 6-year-old children engage in retrospective reevaluation under minimal information-processing demands (Experiment 1). Five- and 6-year-old children do not engage in retrospective reevaluation under more extensive information-processing demands (Experiment 2). Across both experiments, children's retrospective reevaluations were better explained by a simple associative learning model, with only minimal support for a simple Bayesian model. These data contribute to our understanding of the cognitive mechanisms by which children make causal judgements.

Testimony and observation of statistical evidence interact in adults' and children's category-based induction

Hearing generic or other kind-relevant claims can influence the use of information from direct observations in category learning. In the current study, we ask how both adults and children integrate their observations with testimony when learning about the causal property of a novel category. Participants were randomly assigned to hear one of four types of testimony: generic, quantified "all", specific, or only labels. In Study 1, adults (N = 1249) then observed that some proportion of objects (10%-100%) possessed a causal property. In Study 2, children (N = 123, Mage = 5.06 years, SD = 0.61 years, range 4.01-5.99 years) observed a sample where 30% of the objects had the causal property. Generic and quantified "all" claims led both adults and children to generalize the causal property beyond what was observed. Adults and children diverged, however, in their overall trust in testimony that could be verified by observations: adults were more skeptical of inaccurate quantified claims, whereas children were more accepting. Additional memory probes suggest that children's trust in unverified claims may have been due to misremembering what they saw in favor of what they heard. The current findings demonstrate that both child and adult learners integrate information from both sources, offering insights into the mechanisms by which language frames first-hand experience.

Great apes and human children rationally monitor their decisions

Several species can detect when they are uncertain about what decision to make-revealed by opting out of the choice, or by seeking more information before deciding. However, we do not know whether any nonhuman animals recognize when they need more information to make a decision because new evidence contradicts an already-formed belief. Here, we explore this ability in great apes and human children. First, we show that after great apes saw new evidence contradicting their belief about which of two rewards was greater, they stopped to recheck the evidence for their belief before deciding. This indicates the ability to keep track of the reasons for their decisions, or 'rational monitoring' of the decision-making process. Children did the same at 5 years of age, but not at 3 years. In a second study, participants formed a belief about a reward's location, but then a social partner contradicted them, by picking the opposite location. This time even 3-year-old children rechecked the evidence, while apes ignored the disagreement. While apes were sensitive only to the conflict in physical evidence, the youngest children were more sensitive to peer disagreement than conflicting physical evidence.

Developmental Trajectories in Diagnostic Reasoning: Understanding Data Are Confounded Develops Independently of Choosing Informative Interventions to Resolve Confounded Data

Two facets of diagnostic reasoning related to scientific thinking are recognizing the difference between confounded and unconfounded evidence and selecting appropriate interventions that could provide learners the evidence necessary to make an appropriate causal conclusion (i.e., the control-of-variables strategy). The present study investigates both these abilities in 3- to 6-year-old children (N = 57). We found both competence and developmental progress in the capacity to recognize that evidence is confounded. Similarly, children performed above chance in some tasks testing for the selection of a controlled test of a hypothesis. However, these capacities were unrelated, suggesting that preschoolers' nascent understanding of the control-of-variables strategy may not be driven by a metacognitive understanding that confounded evidence does not support a unique causal conclusion, and requires further investigation.

Children's developing capacity to calibrate the verbal testimony of others with observed evidence when inferring causal relations

Across two studies (N = 120), we investigated the development of children's ability to calibrate the certainty of verbal testimony with observable data that varied in the degree of predictive causal accuracy. In Study 1, 4- and 5-year-olds heard a certain explanation or an uncertain explanation about deterministic causal relations. The 5-year-olds made more accurate causal inferences when the informant provided a certain and more calibrated explanation. In Study 2, children heard similar explanations about probabilistic relations, making the uncertain informant more calibrated. The 5-year-olds were more likely to infer the correct causal relations when the informant was uncertain, but only when the explanation was attuned to the stochasticity of the individual causal events (or outcomes that sometimes occur). These findings imply that the capacity to integrate, and make efficient inferences, from distinct sources of knowledge emerges during the preschool years.

Tackling Scientific Misconceptions: The Element of Surprise

Misconceptions about scientific concepts often prevail even if learners are confronted with conflicting evidence. This study tested the facilitative role of surprise in children's revision of misconceptions regarding water displacement in a sample of German children (N = 94, aged 6-9 years, 46% female). Surprise was measured via the pupil dilation response. It was induced by letting children generate predictions before presenting them with outcomes that conflicted with their misconception. Compared to a control condition, generating predictions boosted children's surprise and led to a greater revision of misconceptions (d = 0.56). Surprise further predicted successful belief revision during the learning phase. These results suggest that surprise increases the salience of a cognitive conflict, thereby facilitating the revision of misconceptions.

Developing an Understanding of Science

Young children are adept at several types of scientific reasoning, yet older children and adults have difficulty mastering formal scientific ideas and practices. Why do “little scientists” often become scientifically illiterate adults? We address this question by examining the role of intuition in learning science, both as a body of knowledge and as a method of inquiry. Intuition supports children's understanding of everyday phenomena but conflicts with their ability to learn physical and biological concepts that defy firsthand observation, such as molecules, forces, genes, and germs. Likewise, intuition supports children's causal learning but provides little guidance on how to navigate higher-order constraints on scientific induction, such as the control of variables or the coordination of theory and data. We characterize the foundations of children's intuitive understanding of the natural world, as well as the conceptual scaffolds needed to bridge these intuitions with formal science.

Retrospective inferences in selective trust

Young children learn selectively from others based on the speakers' prior accuracy. This indicates that they recognize the models' (in)competence and use it to predict who will provide the most accurate and useful information in the future. Here, we investigated whether 5-year-old children are also able to use speaker reliability retrospectively, once they have more information regarding their competence. They first experienced two previously unknown speakers who provided conflicting information about the referent of a novel label, with each speaker using the same novel label to refer exclusively to a different novel object. Following this, children learned about the speakers' differing labelling accuracy. Subsequently, children selectively endorsed the object-label link initially provided by the speaker who turned out to be reliable significantly above chance. Crucially, more than half of these children justified their object selection with reference to speaker reliability, indicating the ability to explicitly reason about their selective trust in others based on the informants' individual competences. Findings further corroborate the notion that young children are able to use advanced, metacognitive strategies (trait reasoning) to learn selectively. By contrast, since learning preceded reliability exposure and gaze data showed no preferential looking toward the more reliable speaker, findings cannot be accounted for by attentional bias accounts of selective social learning.