Both humans and Large Language Models (LLMs) store a vast repository of semantic memories. In humans, efficient and strategic access to this memory store is a critical foundation for a variety of cognitive functions. Such access has long been a focus of psychology and the computational mechanisms behind it are now well characterized. Much of this understanding has been gleaned from a widely-used neuropsychological and cognitive science assessment called the Semantic Fluency Task (SFT), which requires the generation of as many semantically constrained concepts as possible. Our goal is to apply mechanistic interpretability techniques to bring greater rigor to the study of semantic memory foraging in LLMs. To this end, we present preliminary results examining SFT as a case study. A central focus is on convergent and divergent patterns of generative memory search, which in humans play complementary strategic roles in efficient memory foraging. We show that these same behavioral signatures, critical to human performance on the SFT, also emerge as identifiable patterns in LLMs across distinct layers. Potentially, this analysis provides new insights into how LLMs may be adapted into closer cognitive alignment with humans, or alternatively, guided toward productive cognitive disalignment to enhance complementary strengths in human-AI interaction.

Saltatory movement strategies in animal search are well known, but their role in pursuit and localization of targets is less so. There are multiple features of forager-target interactions that may generate saltatory forager movement during pursuit and/or localization of the target. Here, we explore how the reliability of the target’s signal and the dependence of the forager’s perception of that signal on the forager’s own motion can generate a variety of different continuous and saltatory pursuit patterns in evolved model foragers with stationary targets. We use dynamical analyses of evolved forager nervous systems to show the possible mechanisms through which saltatory movement was generated and how signals influence those movements in a foraging animal. Saltatory movement during pursuit independent of signal was routinely found when the forager’s perception depended on its movement. Saltatory movement patterns could also be generated in otherwise cruise-like behaviors when target signal either excited or inhibited ongoing forager movement. Additionally, the magnitude of the signal’s influence was found to depend on proximity to the target during pursuit. These models present hypotheses for future empirical research and emphasize the importance of exploring variation in animal movement during different phases of foraging. ### Competing Interest Statement The authors have declared no competing interest.

We test the hypothesis that wealth is associated with lower empathic concern because wealthier people experience fewer challenges and give help less often to people in need. As predicted, Study 1 (N = 513; cross-sectional) showed wealth was negatively associated with trait empathic concern through experiencing fewer challenges and providing less help. Study 2 (N = 915; longitudinal) replicated this while also revealing unexpected curvilinear associations. At low levels of helping (measured monthly across one year), wealth led to lower empathy through experiencing fewer challenges and less helping. However, more helping led to decreasing empathy at higher levels of helping. Moreover, person-centered analyses showed that people experienced higher state-level empathic concern in months in which they provided moderate amounts of help relative to when they helped less than their usual amount of helping. Results suggest people can cultivate empathy through helping, but only for those who help a moderate amount.

Despite the vital role of curiosity-driven exploration in learning, our understanding of how to enhance children's curiosity remains limited. Here, we tested whether hearing a strategic curiosity story with curiosity-promoting themes (e.g., strategically approaching uncertainty, adapting flexibly to new information) versus a control story with traditional pedagogical themes (e.g., following rules, learning from others) would influence children's strategic exploration across two cultures. Three- to 6-year-olds from the United States (N = 138) and Turkey (N = 88) were randomly assigned to hear one of these stories over Zoom, before playing a game in which they searched for sea creatures across five fish tanks. All tanks had the same number of hiding spots but varied in the number of creatures they contained. Time was limited and children could not return to prior tanks, pushing them to allocate search effort strategically. Results indicated that across both countries, children in the strategic curiosity condition explored the virtual "aquarium" more broadly; they moved through tanks more rapidly than children in the control condition and were more likely to explore all five tanks before time ran out. Children in the strategic curiosity condition also showed relatively more strategic search, adapting their search based on the likelihood of finding creatures in each tank. While further research is needed to pinpoint which elements of our stories produced differences in search behavior and whether they did so by enhancing or inhibiting children's strategic exploration, storybooks appear to be a promising method for shaping children's exploration across multiple countries. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Did the COVID-19 pandemic bring people together or push them apart? While infectious diseases tend to push people apart, crises can also bring people together through positive interdependence. We studied this question by asking an international sample (N = 1,006) about their inclinations to cooperate, perceptions of interdependence (i.e., shared fate), and perceived risk as well as local prevalence of COVID-19 infection across 14 time points from March to August, 2020. While perceived interdependence with others tended to increase during this time period, inclinations to cooperate decreased over time. At the within-person level, higher local prevalence of COVID-19 attenuated increases in perceived interdependence with others, and was associated with lower inclinations to cooperate. At the between-person level, people with high perceived interdependence with others reported more stable, or increasing, inclinations to cooperate over time than people with low perceived interdependence. Establishing a high sense of perceived interdependence with others may thus allow people to maintain cooperation during crises, even in the face of challenging circumstances such as those posed by a highly transmissible virus.

Aim Perceived uncontrollable mortality risk (PUMR) refers to people's beliefs regarding their risk of death due to factors outside of their control. Previous theoretical models and empirical studies provide evidence that those with greater PUMR are less motivated to invest in preventative health behaviors, but little is known about how accurately people estimate PUMR compared to objective measures of risk exposure, an important consideration for interventions designed to address the link between PUMR and health behavior. Here, we explore how objective risk indices and personal characteristics relate to PUMR. Subject and methods We performed a series of pre-registered analyses on a US-representative longitudinal study ( N = 915), connecting these results to external data from the Global Burden of Diseases, Injuries, and Risk Factors Study. Results We show that (Study 1) PUMR is associated with objective measures of risk exposure, and that (Study 2) perceptions of risk due to disease drive PUMR, and more educated individuals report less perceived risk. Additionally, we find that (Study 3) estimates of PUMR are relatively stable over a 4-month period ( R = 0.7), indicating that behaviors influenced by PUMR are likely to persist over time. Finally, we show that (Study 4) those who believe they are at greater risk of dying due to factors outside of their control (i.e., greater PUMR) are less likely to engage in general health behaviors. Conclusion By assessing the determinants of PUMR, we can create data-driven policy solutions that lead individuals to more accurate mortality risk assessments and improved health behavior.

The COVID-19 pandemic drastically changed the ways in which social interactions happened. This is particularly true for valued social relationships such as friendships. Friendships provide many physical and mental health benefits, including buffering against loneliness during the pandemic, so contact and interactions with friends may have become more important during the pandemic. Additionally, an additional benefit of friendships is that they may have provided individuals with the ability to reduce their uncertainty and lower their risks during the pandemic. But, if individuals or their friends are not being careful about their risky behavior during this time, this could lead to conflict within these valued relationships. In this study, we investigated how COVID-19, demographics, and preferences for risk management influenced participants’ perceptions of their friendship during the pandemic. We found that participants who were more concerned about contracting COVID-19 and more likely to use risk retention reported having more conflict with their friends during the pandemic. These results suggest that there may be individual differences in risk tolerance of contracting COVID-19 and these differences fundamentally influence social relationships, loneliness, or social connectedness during the pandemic.

Social search has stably evolved across various species and is often used by humans to search for resources (such as food, information, social partners). In turn, these resources frequently come distributed in patches or clusters. In the current work, we use an ecologically inspired agent-based model to investigate whether social search and clustering are stable outcomes of the dynamical mutual interactions between the two. While previous research has studied unidirectional influences of social search on resource clustering and vice versa, the current work investigates the consequential patterns emerging from their two-way interactions over time. In our model, consumers evolved search strategies (ranging from competitive to social) as adaptations to their environmental resource structures, and resources varied in distributions (ranging from random to clustered) that were shaped by agents' consumption patterns. Across four experiments, we systematically analyzed the patterns of influence that search strategies and environment structure have on each other to identify stable attractor states of both. In Experiment 1, we fixed resource clustering at various levels and observed its influence on social search, and in Experiment 2, we observed the influence of social search on resource distribution. In both these experiments we found that increasing levels of one variable produced increases in the other; however, at very high levels of the manipulated variable, the dependent variable tended to fall. Finally in Experiments 3 and 4, we studied the dynamics that arose when resource clustering and social search could both change and mutually influence each other, finding that low levels of social search and clustering were stable attractor states. Our simple 2D model yielded results that qualitatively resemble those across a wide range of search domains (from physical search for food to abstract search for information), highlighting some stable outcomes of mutually interacting consumer/resource systems.

Human sociality is governed by two types of social norms: injunctive norms, which prescribe what people ought to do, and descriptive norms, which reflect what people actually do. The process by which these norms emerge and their causal influences on cooperative behavior over time are not well understood. Here, we study these questions through social norms influencing mask wearing during the COVID-19 pandemic. Leveraging 2 years of data from the United States (18 time points; n = 915), we tracked mask wearing and perceived injunctive and descriptive mask wearing norms as the pandemic unfolded. Longitudinal trends suggested that norms and behavior were tightly coupled, changing quickly in response to public health recommendations. In addition, longitudinal modeling revealed that descriptive norms caused future increases in mask wearing across multiple waves of data collection. These cross-lagged causal effects of descriptive norms were large, even after controlling for non-social beliefs and demographic variables. Injunctive norms, by contrast, had less frequent and generally weaker causal effects on future mask wearing. During uncertain times, cooperative behavior is more strongly driven by what others are actually doing, rather than what others think ought to be done.

Given the importance of friendships during challenging times and the mixed associations between personality traits and disease-related behaviors, we investigated the correlations between personality traits and perceptions of friendships during the COVID-19 pandemic. Data were collected as part of a longitudinal investigation of the correlations between the pandemic and various cooperative relationships. In this investigation, we found that agreeableness and neuroticism predicted participants being more concerned about COVID-19 and bothered by friends' risky behavior, and extraversion predicted enjoying helping friends during the pandemic. Our results suggest that personality differences are associated with how individuals cope with friends' risky behaviors during the COVID-19 pandemic.

Food sharing is central to the human experience, serving biological and sociocultural functions. Historically, food sharing has allowed people to manage food shortages, creating positive interdependence among those who pool calories and other sources of risk. However, sharing food may lead to negative outcomes when food is scarce, or when there is a threat of disease. We found that sharing food (compared with sharing pencils) led to reduced cooperation with an experiment partner (Study 1) and that perceived scarcity partially mediated a negative association between zero-sum orientation and perceived interdependence with people involved in a recently shared meal (Studies 2-3). Disgust was also associated with lower perceived interdependence toward people involved in a shared meal (Study 3; N-Total = 1126). Our results suggest that scarcity and disgust can interfere with the positive feelings people might otherwise experience when eating together, warning against lay beliefs that 'breaking bread' necessarily brings people together.

Sometimes people help one another expecting to be repaid, while at other times people help without an expectation of repayment. What might underlie this difference in expectations of repayment? We investigate this question in a nationally representative sample of US adults (N = 915), and find that people are more likely to expect repayment when needs are perceived to be more predictable. We then replicate these findings in a new sample of US adults (N = 417), and show that people have higher expectations of repayment when needs are perceived to be more predictable because people assign greater responsibility to others for experiencing such predictable needs (e.g., needing money for utilities). This is consistent with previous work based on smaller-scale societies, which shows that the predictability of needs influences expectations of repayment. Our results also add to this previous work by (1) showing that the positive relationship between predictability of needs and expectations of repayment previously found in smaller-scale communities is generalizable to the US population, and (2) showing that attributions of responsibility partially mediate this relationship. This work shows that the predictability of needs and attributions of responsibility for that need are important factors underlying the psychology of helping in times of need.

Humans may retrieve words from memory by exploring and exploiting in "semantic space" similar to how nonhuman animals forage for resources in physical space. This has been studied using the verbal fluency test (VFT), in which participants generate words belonging to a semantic or phonetic category in a limited time. People produce bursts of related items during VFT, referred to as "clustering" and "switching." The strategic foraging model posits that cognitive search behavior is guided by a monitoring process which detects relevant declines in performance and then triggers the searcher to seek a new patch or cluster in memory after the current patch has been depleted. An alternative body of research proposes that this behavior can be explained by an undirected rather than strategic search process, such as random walks with or without random jumps to new parts of semantic space. This study contributes to this theoretical debate by testing for neural evidence of strategically timed switches during memory search. Thirty participants performed category and letter VFT during functional MRI. Responses were classified as cluster or switch events based on computational metrics of similarity and participant evaluations. Results showed greater hippocampal and posterior cerebellar activation during switching than clustering, even while controlling for interresponse times and linguistic distance. Furthermore, these regions exhibited ramping activity which increased during within-patch search leading up to switches. Findings support the strategic foraging model, clarifying how neural switch processes may guide memory search in a manner akin to foraging in patchy spatial environments.

Vaccine hesitancy is variable across individuals and contexts. Theoretical work suggests that group membership should differentially affect attitudes and behavior related to COVID-19 vaccines, as individuals draw on their identities and experiences relevant to their social groups to deal with uncertainty concerning the vaccines. The present work uses longitudinal survey data to explore how identity predicted vaccination attitudes in the U.S. before COVID-19 vaccines were widely available and which role-based or contextual variables influenced early vaccination decisions. We replicate these patterns in hesitancy with a larger and more racially diverse sample and identify whether encouragement from people with shared identities was important when participants made the decision to be vaccinated. Before the vaccines were widely available, higher SES predicted less hesitancy toward the COVID-19 vaccines, while being Black and/or having more conservative political orientation predicted more hesitancy. More conservative political orientation was a predictor of vaccine hesitancy across racial identities. While rationale for mistrust may vary by identity, vaccine trust is a significant predictor of early vaccination, and it is influenced by local norms surrounding the COVID-19 vaccines. Encouragement from shared group members was more important for women, liberal individuals, and younger individuals in making the decision to be vaccinated.

Information-seeking is usually conceived of as gathering information to make better decisions by observing and sampling from the external world. But for humans and many other intelligent agents, much of that information, once gathered, is also stored to guide future decisions, necessitating mechanisms for seeking information in some form of inner space. Here we survey various types of evidence suggesting that strategies adapted for search in external spatial environments are also used to seek information internally from memory. These include foraging strategies such as area-restricted search, which adaptively balances exploitation of locally clustered resources with exploration for resources more widely dispersed. We also describe how internal search satisfies the predictions of external foraging theory via the Marginal Value Theorem and show how these predictions can be used to investigate individual differences in memory search such as those caused by aging and cognitive impairment. Finally, we consider evidence that the structure of inner space may be a result of the very processes we use to search it.

Environmentally sustainable food consumption is one component of addressing climate change. Previous research has largely approached sustainable food consumption by investigating individual behaviors, without a broader conceptualization of what motivates food consumers to act sustainably. Using a representative sample of Indiana consumers, we explore sustainability across a range of food behaviors through latent class analysis, controlling for environmental attitudes, spatial access to food, and consumer demographics. This approach allows us to go beyond consumer segmentation analysis to explore how consumers conceptualize sustainable food behavior. The largest class of consumers (44% of the sample) appear either unwilling or unable to pay more for sustainability but are more likely to engage in sustainable behaviors that intersect with self-oriented attributes such as health benefits and lower cost. A second class (34%) consists of consumers who seem to be primarily motivated by the single issue of buying organic, are on average higher income, more educated, have better access to food, and are not opposed to paying for sustainability. Consumers in the smallest and most highly motivated group (9%) in terms of sustainability attitudes and self-perceived sustainability focus on local food production and are generally rural dwelling with less income. Only 13% of consumers engage in few to no sustainable behaviors, and these people notably exhibit the least sustainable attitudes. These findings illustrate the ways in which food sustainability is more nuanced than often characterized—much of it is driven by convenience and self-interest rather than reputation with respect to sustainability or conviction about environmental outcomes. This work also highlights how a combination of social, psychological, and spatial barriers exists and shape how different consumer groups conceptualize sustainable food consumption.

Are impulsive behaviors an adaptive response to living in harsh or unpredictable environments? Formal models help address this question by providing cost-benefit analyses across a broad range of environmental conditions, but their various results have not been systematically integrated. Here, we survey models from diverse disciplines including psychology, biology, economics, and management to develop a conceptual framework of impulsivity. Using this framework, we integrated results from 30 models to review whether impulsivity is adaptive across a range of environmental conditions. We focus on information impulsivity, that is, acting without considering consequences, and temporal impulsivity, that is, the tendency to pick sooner outcomes over later ones. Results show that both types are adaptive when individuals are close to a critical threshold (e.g., bankruptcy), resources are predictable, or interruptions are common. When resources are scarce, impulsivity can be adaptive or maladaptive, depending on the type and degree of scarcity. Information impulsivity is also adaptive when environments do not change over time or change very often (but maladaptive in between), or if local resource patches have similar properties, reducing the need to gather further information. Temporal impulsivity is adaptive when environments do not change over time and when local resource patches differ. Our review shows theoreticians how ideas from different disciplines are connected, affords formal modelers to see similarities and differences between their own models and those of others, and informs researchers about which empirical predictions generalize across a broad range of environmental conditions and which ones do not. To end, we provide concrete recommendations for future empirical studies.

Individuals can hold contrasting views about distinct times: for example, dread over tomorrow's appointment and excitement about next summer's vacation. Yet, psychological measures of optimism often assess only one time point or ask participants to generalize about their future. Here, we address these limitations by developing the optimism curve, a measure of societal optimism that compares positivity toward different future times that was inspired by the Treasury bond yield curve. By performing sentiment analysis on over 3.5 million tweets that reference 23 future time points (2 days to 30 years), we measured how positivity differs across short-, medium-, and longer-term future references. We found a consistent negative association between positivity and the distance into the future referenced: From August 2017 to February 2020, the long-term future was discussed less positively than the short-term future. During the COVID-19 pandemic, this relationship inverted, indicating declining near-future- but stable distant-future-optimism. Our results demonstrate that individuals hold differentiated attitudes toward the near and distant future that shift in aggregate over time in response to external events. The optimism curve uniquely captures these shifting attitudes and may serve as a useful tool that can expand existing psychometric measures of optimism.

The last common ancestor shared by humans and other vertebrates lived over half a billion years ago. In the time since that ancestral line diverged, evolution by natural selection has produced an impressive diversity—from fish to birds to elephants—of vertebrate morphology; yet despite the great species-level differences that otherwise exist across the brains of many animals, the neural circuitry that underlies motor control features a functional architecture that is virtually unchanged in every living species of vertebrate. In this article, we review how that circuitry facilitates motor control, trial-and-error-based procedural learning, and habit formation; we then develop a model that describes how this circuitry (embodied in an agent) works to build and refine sequences of goal-directed actions that are molded to fit the structure of the environment (in which the agent is embedded). We subsequently review evidence suggesting that this same functional circuitry became further adapted to regulate cognitive control in humans as well as motor control; then, using examples of heuristic decision-making from the ecological rationality tradition, we show how the model can be used to understand how that circuitry operates analogously in both cognitive and motor domains. We conclude with a discussion of how the model encourages a shift in perspective regarding ecological rationality’s “adaptive toolbox”—namely, to one that views heuristic processes and other forms of goal-directed cognition as likely being implemented by the same neural circuitry (and in the same fashion) as goal-directed action in the motor domain—and how this change of perspective can be useful.