The Cognitive Reflection Test (CRT) is widely used in reasoning and decision making research, but it lacks a formal cognitive foundation. As a result, debates persist over why CRT scores correlate with mathematical ability, why some individuals solve CRT problems easily while others struggle, and which mental processes drive observed patterns in data. We use an ecological perspective combined with computational cognitive modeling to address these questions, focusing in particular on the bat-and-ball problem. First, we specify the learning environment by assembling a large dataset of grade school verbal math problems. Second, we specify a formal learning mechanism that selects the arithmetic operation most likely to be applied to a novel problem based on the linguistic association of the problem with learned exemplars. Our model generates the intuitive errors elicited by the bat-and-ball problem (as well as other CRT items) and explains why these errors can be seen as byproducts of adaptive cognition. It also makes new predictions about the effect of environmental structure and problem wording on strategy selection and downstream performance, and we validate these predictions in two new preregistered experiments. Overall, our work provides theoretical clarity and quantitative rigor to our understanding of intuitive judgment, and shows how such judgments can be understood as rational adaptations to the learning environment.
Disfluency in speech (e.g., English um and uh, also known as filled pauses) has long been argued to have social costs for the speaker. However, this conclusion is based on a limited set of measures obtained for scripted, lab-recorded speech. Here, using stimuli from spontaneous speech and an expanded set of social trait evaluations, we argue that phenomena typically described as “disfluency” are not always evaluated negatively. First, at the global person level, filled pauses did not affect the overall perceived competence, trustworthiness, or warmth of the speaker as a person (Experiment 1). However, at the local conversation level, filled pauses reduced perceived speaker trustworthiness within a conversation (Experiment 2). Furthermore, within a conversation, speakers who produced filled pauses were rated as less ready and less certain than speakers who did not (even though experts who used filled pauses were more likely to be considered careful than novices; Experiment 3). Finally, speakers using filled pauses were more likely to be avoided in future interactions (Experiment 4). Together, these data reveal flexible and detailed mappings of speech features such as disfluencies to social attributes and showcase the value of studying social meaning using speech generated beyond the lab.
Deciding what to do should typically involve considering what could have been done, but in fact was not done. We tested whether such unchosen alternative action paths leave traces in memory. Across two behavioural experiments, participants solved a modified Tower of London (ToL) task in which each problem afforded two alternative and equivalent optimal solution paths. They then undertook a recognition test in which they viewed configurations of the ToL drawn either from the chosen path, the unchosen path, or from new, unviewed configurations. Participants were consistently more likely to judge (falsely) that they had previously seen configurations from the unchosen but plausible path, compared to novel configurations. Signal detection analyses further showed reduced discriminability and a more liberal decision criterion for alternative-path configurations compared to control novel configurations. Unchosen action alternatives may be difficult to distinguish from actions that are actually executed, but may have a feeling of familiarity that leads to an “old” response. This effect replicated when novel items were matched to alternative items in visual similarity, ruling out explanations based on perceptual confusability. The results suggest that action planning generates representations of unchosen alternatives that can persist and bias later memory, even if not executed. Planning therefore leaves behind more than a route towards action goals: it also leaves mnemonic traces of paths not taken.
Humans tend to discount the value of delayed rewards. As a consequence, they forego larger long-term benefits (e.g., €29 in 91 days) in favor of smaller short-term gains (e.g., €20 now). Episodic future thinking (EFT) attenuates such delay discounting. That is, people are more likely to choose a larger but delayed reward after imagining an episode at the future time point (e.g., a café visit in 91 days). We here elucidate the underlying mechanism by examining the temporal specificity of this effect. Does EFT foster farsighted decisions in general by shifting the mental focus towards the future, or does it specifically influence decisions in favor of the imagined time point? In two experiments (n=84 total) we systematically varied the concordance between the imagined time point and the delay at which the larger reward could be received. When these time points matched, we replicated the beneficial effect of EFT (Experiments 1 and 2). Critically, when the imagined time point matched the sooner of two delayed choice options, the effect was reversed: Choices became more impulsive (Experiment 1). When the imagined delay was double the delay of the larger option, the effect was not significant and numerically lower than when the delays matched (Experiment 2). Notably, more vivid imaginations were associated with more farsighted choices at longer delays, irrespective of whether the imagined time point matched the delayed choice. Together, the data demonstrate that both temporally specific and unspecific mechanisms contribute to the impact of EFT on delay discounting.
This study investigated whether deaf readers of English predict semantic and phonological information during sentence reading in the same way as hearing readers. Using a self-paced moving-window paradigm, we tested four groups who were matched on reading ability and English vocabulary knowledge: deaf readers without cochlear implants (CIs), deaf readers with CIs, hearing native readers of English (HN), and hearing non-native readers of English (HNN). Participants read sentences ending in one of five target types: congruent highly predictable words, congruent less predictable words, incongruent words, pseudohomophones of the highly expected word, or orthographic-control pseudowords. Semantic prediction was assessed by comparing reading times for congruent endings in high- and low-cloze contexts, and semantic congruency was assessed by comparing congruent and incongruent endings. Phonological contextual facilitation was assessed by comparing pseudohomophones with orthographic controls. All groups showed sensitivity to semantic incongruence, although Deaf-CI readers showed weaker evidence of semantic prediction, suggesting weaker reliance on predictive mechanisms during reading. In contrast, phonological facilitation was observed only in the HN group, despite the groups being matched on reading ability, suggesting that phonological facilitation during sentence reading may depend on hearing-native-like phonological representations. These findings support the view that skilled reading can be achieved through different routes. While semantic prediction appears to be broadly available across reader groups, phonological facilitation from an expected word form may depend on native-like, automatically engaged spoken-language phonological representations.
Humans typically prioritize their closest social connections. Prior neuroimaging work found that approximately 80% of the differential brain processing across social network members favors individuals in one's innermost social circle. But is this neural pattern inherent, or influenced by individual differences in social characteristics? To explore this, we studied a unique group of people: extraordinary altruists, who had donated a vital organ to an anonymous recipient, and compared them with demographically closely-matched controls. We measured brain responses using functional MRI as participants processed individuals from various circles of their social networks. In control participants, 75% of social-circle-sensitive brain volume responded most strongly to the closest circle, replicating previous findings in young adults. In contrast, extraordinary altruists showed the opposite pattern: only 30% of this brain volume preferred the closest circle, while 70% showed stronger responses to more distant social circles. This difference was particularly pronounced in core social-cognition regions, especially the temporoparietal junction, which is typically tuned to close others. Within the temporoparietal junction, the closest vs. distant response contrast was significantly smaller in extraordinary altruists than in controls, indicating reduced selectivity for the innermost circle. Entropy analysis also suggested more uniform response profiles in extraordinary altruists in several regions, including the temporoparietal junction. Collectively, these findings suggest that extraordinary altruists process social networks uniquely at the neuroanatomical level, irrespective of demographic background, displaying a broader pattern of affiliation that may underlie their remarkable capacity to extend care beyond conventional social boundaries. SIGNIFICANCE STATEMENT: Extraordinary altruists, such as kidney donors to strangers, perform acts of care that defy conventional social boundaries. Can such rare behavior be traced to differences in brain processing? Using fMRI, we found that while most people's brains respond most strongly to their closest social ties, extraordinary altruists display a reversed pattern - allocating greater neural resources to more distant social circles. This shift is especially pronounced in key social-cognitive regions, including the temporoparietal junction. Such a broadened, less selective neural response may support care that transcends conventional social boundaries. These findings offer a neurobiological account of selfless behavior and reveal brain mechanisms underlying extraordinary altruism.
Non-contingent learning data can produce an apparent “illusory” or “false” perception of causality (“causal illusion”), especially if a potential cause and effect frequently co-occur — a phenomenon known as the “outcome-density bias.” In contrast to the prominent bias view, Bayesian models of causal induction explain these causal illusions as the result of a rational learning process in which observed data fail to fully override non-zero causal priors. Convincing evidence for this rational explanation requires an experimentally manipulated effect of causal priors — which has so far been lacking. We report four experiments (N=1860) supporting the Bayesian interpretation. We successfully manipulated participants’ causal priors either through visually conveyed causal mechanism information or through statistical base-rate information, and found that both manipulations influenced the degree to which participants reported a causal relation after having processed non-contingent learning data. The results were in line with Bayesian updating: lower causal priors yielded lower post-learning beliefs in a causal connection, effectively reducing the “causal illusion.” The results strengthen computational models implementing a rational Bayesian view of causal induction.
The SNARC effect (Spatial-Numerical Association of Response Codes) is the most extensively studied spatial-numerical association, referring to faster left- and right-sided responses to small and large numbers, respectively. However, evidence for a relation between the SNARC effect and mathematical performance remains inconclusive. Several factors may explain these inconsistencies. First, they may arise from how the SNARC effect is typically computed. Conventional analyses use a single regression slope linking numerical magnitude and spatial response, thereby conflating two dimensions: the directionality (left-right polarity) and the size of the association, which may differentially relate to mathematical performance and should be examined separately. Moreover, intra-individual variability in SNARC estimates introduces imprecision that can obscure meaningful relations. Second, different mathematical skills may engage spatial-numerical processes to varying degrees. Accounting for this is essential to clarify the nature of the SNARC-math relation, particularly in younger children when early mathematical skills are first emerging. To address these issues, we examined the SNARC effect in 135 kindergarteners using a magnitude judgment task (i.e., a speeded task in which participants judge whether centrally presented Arabic digits are smaller or larger than a reference value), deriving three indices characterising the SNARC slope: absolute size, directionality, and intra-individual variability. Early numerical skills were assessed through counting, numerosity comparison, magnitude judgment and ordinal judgment tasks. Traditional SNARC slopes did not correlate with mathematical performance, but regression analyses separating size, directionality, and variability revealed distinct effects: intra-individual variability predicted magnitude judgment, whereas directionality predicted ordinal judgment. These findings indicate that decomposing the SNARC effect and accounting for intra-individual variability reveals meaningful, task-specific associations, offering a potential explanation for previous inconsistencies.
Early deception poses a representational puzzle: young children can strategically deny and conceal transgressions before they reliably succeed on standard measures of false-belief understanding. Standard interpretations often force a choice between two unsatisfying extremes: either early deceptive behavior is reduced to a routine punishment-avoidance response, or it is taken to imply a surprisingly rich capacity for belief representation. This paper develops an intermediate alternative. Early deception is argued to be better explained by an access-based epistemic policy that tracks another agent’s epistemic standing, specifically whether that agent is in a position to know, on the basis of perceptual access, evidential availability, and blocking conditions. This proposal is developed as a policy framework in which actions such as denial, concealment, and withholding are selected under uncertainty as functions of graded epistemic standing. The developmental literature is treated not as the paper’s main payload, but as a constraint on representational format. The resulting pattern is asymmetric: young children show flexible sensitivity to witness access, audience knowledge, and opportunities for concealment, yet remain brittle under follow-up questioning, semantic leakage, and evidence-coordinated cover-story demands. This pattern is best understood as evidence for a factive, access-based form of epistemic mindreading that precedes robust belief-based deception.
Statistical learning (SL) is widely recognized as a critical contributor to the acquisition of a wide range of cognitive skills, particularly language. Overall, SL refers to the ability to extract regularities in how stimuli co-occur across space and time, in different domains (verbal or non-verbal) and modalities (auditory or visual). Mathematics is often referred to as a language of symbols and rules. However, the potential role of SL in mathematical cognition remains largely underexamined. Given the multidimensional nature of mathematics (e.g., arithmetic, geometry), we hypothesized that the acquisition of different math abilities relies on the extraction of verbal and/or visuospatial regularities. To test this hypothesis, the current study represents the first investigation of the relationship between SL in different domains and modalities of learning input and several math skills. To this end, 151 adult participants were evaluated on auditory-verbal and visual/visuospatial SL abilities, as well as on two basic math skills (order processing and geometry intuition) and on general math performance. The results of a path analysis showed that SL is related to basic math skills, which in turn are (indirectly) related to general math performance. Importantly, domain/modality-specific trends were observed, such that auditory-verbal and visuospatial SL were related to order processing and geometry intuition, respectively. These findings indicate a consistent, albeit small, association between SL and mathematical cognition, and underline the importance of considering different domains and modalities of SL in relation to a multidimensional skill like mathematics.
Human creativity is often studied in relation to idea-generation and problem-solving; yet a growing body of evidence suggests that creativity-related traits may also shape how we perceive and understand the world around us. In this study, we examined whether individual differences in creative disposition are associated with differences in crossmodal associations triggered by musical stimuli. A total of 97 healthy adult participants completed a crossmodal association task wherein they rated the extent to which short musical melodies could be associated with illustrations of everyday objects, followed by questionnaires assessing creative personal identity, creative self-efficacy, openness to experience, and creative activities. Using a Gaussian Mixture model-based clustering approach, we identified two participant profiles characterised by consistently high versus low creativity-related scores. By means of a Bayesian Zero-Inflated Beta-distributed Mixed Model, credible evidence was obtained that individuals from the former cluster, characterised by a broadly developed creative profile, gave higher association scores. Despite some uncertainty regarding its practical relevance, the effect showed a 95.82% probability of existence, suggesting that creativity-related traits may influence the crossmodal associations between musical and visual percepts.
Memories tend to form networks that reflect relationships between experiences. These networks, or cognitive graphs, have been used to study how structural features of memory can bias decisions — analogous to how reward shapes choices by prioritizing reward-associated items in memory. However, how reward interacts with the structure of existing cognitive graphs to influence subsequent value-based decisions remains unclear. Across multiple experiments, participants learned a community-structured cognitive graph and made value-based decisions, with some experiments including graph reconstruction and or/reward learning. Results indicated a limited retroactive spread of reward, contingent on the presence of a retrievable representation of either the entire cognitive graph or reward-item associations. Reward also shaped memory representations by preserving the relational structure of directly associated items to guide value-based decisions. These findings suggest that reward acts as an organizing force within memory representations — selectively preserving relevant associations to support adaptive, value-based decision making, and most prominently when these associations are retrievable.
The most important objects - from predators to projectiles - are often those that are moving. Accordingly, visual systems are specialized for object tracking, and many models of such processing involve continually re-identifying surface features across space and time (as you might follow a tiger by keeping track of its orange stripes). Here, in contrast, we show across four experiments - and in phenomenologically compelling demonstrations - that people also spontaneously perceive and track change-defined objects, with no enduring surface properties from moment to moment. Observers viewed regular grids filled with hundreds of small elements (e.g. randomly oriented crosses). Change-defined objects were implemented by having an element change (e.g., from one random orientation to another random orientation), with these changes propagating through space and time. Observers still detected the 'second-order' motion in such displays - despite the lack of persisting features, and without being able to identify the object in any static frame. And beyond motion detection, observers also spontaneously perceived and tracked persisting objects - and were even able to perform multiple-object tracking. We also generalized this phenomenon in several ways, showing that it occurs with other types of changes (e.g., to brightness or shape), and even when different types of changes are haphazardly interleaved (e.g. with a random orientation change to one element followed by a random brightness change to its neighbor, etc.). In essence, these stimuli show how observers can track a tiger that is always fully hidden while moving through tall grass, by tracking variable local changes to the grass itself.
Judgment and decision making research has typically not explicitly considered whether there are individual differences in overconfidence. When correlations among individual measures are reported, the strength of associations tends to be modest. Despite this, we show that when using the methods of individual differences research to analyze measures of overconfidence, there is substantial evidence for reliable, trait-level overconfidence-which we call 'core overconfidence'-that is stable over both short and long time periods. Moreover, three forms of overconfidence have primarily been studied: overestimation, overplacement, and overprecision. We find individual differences in each form, and that the different forms share a common core, supporting the conceptualization of overconfidence as a stable, multifaceted trait. Differences in both self-views and in reflective cognitive processes appear to contribute to this stability: A range of individual difference measures (e.g., narcissism, actively open-minded thinking) significantly correlate with core overconfidence. These relationships are generally explained distinctly by either a relationship with confidence or accuracy. In additional studies, we find that individual differences in overconfidence are robust even when minimizing the correlations in accuracy across tasks, demonstrating that they are not specific to self-views in a narrow domain. Overall, by studying the same respondents' behaviors across domains, forms of overconfidence, and time periods, we find strong evidence that overconfidence is a stable, multifaceted trait that can be explained by common motivated and cognitive processes.
Event schemas are important for event cognition because they support the activation of causal antecedents and causal consequences. However, an open question is whether there are biases toward the activation of antecedents versus consequences. This study investigated the activation of antecedents and consequences within event schemas. In two experiments, participants read a sentence describing an action within a schema and were then asked to judge if a second action was related to the former action. The directionality (antecedent or consequence) and temporal distance of the second action were manipulated. Experiment 1 explicitly instantiated the schema by providing a sentence describing the event but Experiment 2 did not. Both experiments revealed a robust consequence bias such that participants verified consequences faster and more accurately than antecedents. A cross-experiment analysis indicated that both antecedents and consequences were activated when the schema was instantiated relative to when it was not, albeit there was still a consequence bias. The implications of these findings for theories of event perception and text comprehension are discussed.
Sleep consolidation effects are well established in the cognitive domains of memory and language learning, however very little is known regarding sleep effects in social contexts. Across three experiments, we investigated whether sleep enhances the accuracy of socially formed memories and communicative efficiency, including the use of partner-specific common ground between interlocutors in conversation. Participants completed a computerised (Experiment 1) or face-to-face (Experiment 2) matcher-director task with two other people before and after 12 h of sleep/wake. Both experiments showed more accurate item recognition and source memory after sleep vs. wake. The Sleep group also used fewer words to describe old vs. new items in Experiment 2, suggesting an influence of sleep on communicative behaviour based on prior discourse, but this effect was not partner specific. In Experiment 3, the retention of item/source memory and common ground was measured across a two-hour, polysomnographically recorded nap. Results revealed no significant relationships between consolidation-linked sleep electrophysiology (e.g., slow oscillation-spindle coupling) and memory/common ground retention. Together, these findings indicate that sleep plays a key role in stabilising the social-cognitive foundations of communication but cannot elucidate the mechanism(s) underpinning this support. Regardless, by preserving conversational memories and the efficiency of shared understanding, sleep may help sustain smooth coordination in everyday interaction and social relationships.
Decades of research have examined the consequences of disagreement, both negative (harm to relationships) and positive (fostering learning opportunities). Yet the psychological mechanisms underlying disagreement judgments themselves are poorly understood. Much research assumes that disagreement tracks divergence: the difference between two individuals' beliefs with respect to a proposition, where this can be understood in binary terms (individuals either agree or disagree) or as a continuous difference (for instance, in subjective probability). We test divergence as an account of interpersonal disagreement judgments using predictive modeling (N = 238). Our results indicate that while judgments of disagreement track divergence, other properties of beliefs-such as their extremity-play a significant role. Moreover, these additional factors are necessary for predicting key social consequences of disagreement, including inferences of bias, coldness, and incompetence. These results suggest that the assumption that disagreement judgments merely track differences in belief is empirically unjustified.
Tracking the hierarchical structure of connected speech is a central component of speech perception, yet the timescales at which motor systems contribute to this process remain poorly understood, especially given strong individual differences in auditory-motor coupling strength. Using a modified speech-to-speech synchronization (SSS) paradigm with hierarchically structured speech, we recorded participants' articulatory vocalization, eye movements, and cardiac activity while they synchronized their speech to the syllable rhythm and then continued the rhythm after stimulus offset. Syllable-rate synchronization performance exhibited the characteristic bimodal distribution that separated high and low synchronizers, allowing us to examine how individual differences in auditory-motor coupling shape the contributions of multiple motor systems to speech tracking. We found that vocalization showed spontaneous word-level tracking and timing shifts consistent with temporal prediction based on high-level structures, and these signatures were markedly stronger in high synchronizers. Meanwhile, covert oculomotor activity exhibited spontaneous syllable-rate tracking, occurring in high synchronizers only, with temporal dynamics that diverged from overt production. Finally, high synchronizers showed subtle structure-dependent modulation of autonomic cardiac activity. Together, these findings demonstrate that articulatory and oculomotor activity provide complementary contributions to speech tracking, and that the presence and organization of these contributions vary systematically across individuals. This multisystem approach offers behavioral and physiological markers for the embodied mechanisms of hierarchical speech processing.
Across two preregistered studies (N = 199), we investigate whether children's developing theory of mind incorporates assumptions that people revise their beliefs rationally. Specifically, we focus on the role of prior belief strength (magnitude of and degree of certainty in the belief) in shaping children's judgments and predictions about others' belief revision. We presented four- to eight-year-old children with pairs of characters who hold the same belief, but differ in the strength and certainty of their belief. In Study 1, we found that children judge it possible to change both strong and weakly held beliefs-but by age five, children judge strongly held beliefs to be both less possible to change, and more difficult to change than weakly held ones. Additionally, children's open-ended explanations predominantly cited evidence as the reason for belief change. In Study 2, we tested whether children use prior belief strength to predict responses to counterevidence. After both characters encountered identical disconfirming testimony from a group of peers, children by age five predicted that characters with weak priors would revise their beliefs more readily than those with strong priors. In both studies, we found these effects of prior belief across diverse belief domains. Together, these studies offer evidence that children's developing theory of mind-in particular, their understanding of belief change-incorporates an assumption that beliefs are graded states whose revision is constrained by prior strength.
