Shared functional specialization in transformer-based language models and the human brain.

When processing language, the brain is thought to deploy specialized computations to construct meaning from complex linguistic structures. Recently, artificial neural networks based on the Transformer architecture have revolutionized the field of natural language processing. Transformers integrate contextual information across words via structured circuit computations. Prior work has focused on the internal representations ("embeddings") generated by these circuits. In this paper, we instead analyze the circuit computations directly: we deconstruct these computations into the functionally-specialized "transformations" that integrate contextual information across words. Using functional MRI data acquired while participants listened to naturalistic stories, we first verify that the transformations account for considerable variance in brain activity across the cortical language network. We then demonstrate that the emergent computations performed by individual, functionally-specialized "attention heads" differentially predict brain activity in specific cortical regions. These heads fall along gradients corresponding to different layers and context lengths in a low-dimensional cortical space.

Word Forms Reflect Trade-Offs Between Speaker Effort and Robust Listener Recognition.

How do cognitive pressures shape the lexicons of natural languages? Here, we reframe George Kingsley Zipf's proposed "law of abbreviation" within a more general framework that relates it to cognitive pressures that affect speakers and listeners. In this new framework, speakers' drive to reduce effort (Zipf's proposal) is counteracted by the need for low-frequency words to have word forms that are sufficiently distinctive to allow for accurate recognition by listeners. To support this framework, we replicate and extend recent work using the prevalence of subword phonemic sequences (phonotactic probability) to measure speakers' production effort in place of Zipf's measure of length. Across languages and corpora, phonotactic probability is more strongly correlated with word frequency than word length. We also show this measure of ease of speech production (phonotactic probability) is strongly correlated with a measure of perceptual difficulty that indexes the degree of competition from alternative interpretations in word recognition. This is consistent with the claim that there must be trade-offs between these two factors, and is inconsistent with a recent proposal that phonotactic probability facilitates both perception and production. To our knowledge, this is the first work to offer an explanation why long, phonotactically improbable word forms remain in the lexicons of natural languages.

Using games to understand the mind.

Board, card or video games have been played by virtually every individual in the world. Games are popular because they are intuitive and fun. These distinctive qualities of games also make them ideal for studying the mind. By being intuitive, games provide a unique vantage point for understanding the inductive biases that support behaviour in more complex, ecological settings than traditional laboratory 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 the advantages and drawbacks of using games relative to standard laboratory-based experiments and lay out a set of recommendations on how to gain the most from using games to study cognition. We hope this Perspective will lead to a wider use of games as experimental paradigms, elevating the ecological validity, scale and robustness of research on the mind.

Identifying resource-rational heuristics for risky choice.

Perfectly rational decision making is almost always out of reach for people because their computational resources are limited. Instead, people may rely on computationally frugal heuristics that usually yield good outcomes. Although previous research has identified many such heuristics, discovering good heuristics and predicting when they will be used remains challenging. Here, we present a theoretical framework that allows us to use methods from machine learning to automatically derive the best heuristic to use in any given situation by considering how to make the best use of limited cognitive resources. To demonstrate the generalizability and accuracy of our method, we compare the heuristics it discovers against those used by people across a wide range of multi-attribute risky choice environments in a behavioral experiment that is an order of magnitude larger than any previous experiments of its type. Our method rediscovered known heuristics, identifying them as rational strategies for specific environments, and discovered novel heuristics that had been previously overlooked. Our results show that people adapt their decision strategies to the structure of the environment and generally make good use of their limited cognitive resources, although their strategy choices do not always fully exploit the structure of the environment. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

The universal law of generalization holds for naturalistic stimuli.

Shepard's universal law of generalization is a remarkable hypothesis about how intelligent organisms should perceive similarity. In its broadest form, the universal law states that the level of perceived similarity between a pair of stimuli should decay as a concave function of their distance when embedded in an appropriate psychological space. While extensively studied, evidence in support of the universal law has relied on low-dimensional stimuli and small stimulus sets that are very different from their real-world counterparts. This is largely because pairwise comparisons-as required for similarity judgments-scale quadratically in the number of stimuli. We provide strong evidence for the universal law in a naturalistic high-dimensional regime by analyzing an existing data set of 214,200 human similarity judgments and a newly collected data set of 390,819 human generalization judgments (N = 2,406 U.S. participants) across three sets of natural images. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Replies to commentaries on beyond playing 20 questions with nature.

Commentaries on the target article offer diverse perspectives on integrative experiment design. Our responses engage three themes: (1) Disputes of our characterization of the problem, (2) skepticism toward our proposed solution, and (3) endorsement of the solution, with accompanying discussions of its implementation in existing work and its potential for other domains. Collectively, the commentaries enhance our confidence in the promise and viability of integrative experiment design, while highlighting important considerations about how it is used.

Reconciling categorization and memory via environmental statistics.

How people represent categories and how those representations change over time is a basic question about human cognition. Previous research has demonstrated that people categorize objects by comparing them to category prototypes in early stages of learning but consider the individual exemplars within each category in later stages. However, these results do not seem consistent with findings in the memory literature showing that it becomes increasingly easier to access representations of general knowledge than representations of specific items over time. Why would one rely more on exemplar-based representations in later stages of categorization when it is more difficult to access these exemplars in memory? To reconcile these incongruities, our study proposed that previous findings on categorization are a result of human participants adapting to a specific experimental environment, in which the probability of encountering an object stays uniform over time. In a more realistic environment, however, one would be less likely to encounter the same object if a long time has passed. Confirming our hypothesis, we demonstrated that under environmental statistics identical to typical categorization experiments the advantage of exemplar-based categorization over prototype-based categorization increases over time, replicating previous research in categorization. In contrast, under realistic environmental statistics simulated by our experiments the advantage of exemplar-based categorization over prototype-based categorization decreases over time. A second set of experiments replicated our results, while additionally demonstrating that human categorization is sensitive to the category structure presented to the participants. These results provide converging evidence that human categorization adapts appropriately to environmental statistics.

Improving Memory Search Through Model-Based Cue Selection.

We often use cues from our environment when we get stuck searching our memories, but prior research has failed to show benefits of cuing with other, randomly selected list items during memory search. What accounts for this discrepancy? We proposed that cues' content critically determines their effectiveness and sought to select the right cues by building a computational model of how cues affect memory search. Participants (N = 195 young adults from the United States) recalled significantly more items when receiving our model's best (vs. worst) cue. Our model provides an account of why some cues better aid recall: Effective cues activate contexts most similar to the remaining items' contexts, facilitating recall in an unsearched area of memory. We discuss our contributions in relation to prominent theories about the effect of external cues.

Optimal metacognitive control of memory recall.

Most of us have experienced moments when we could not recall some piece of information but felt that it was just out of reach. Research in metamemory has established that such judgments are often accurate; but what adaptive purpose do they serve? Here, we present an optimal model of how metacognitive monitoring (feeling of knowing) could dynamically inform metacognitive control of memory (the direction of retrieval efforts). In two experiments, we find that, consistent with the optimal model, people report having a stronger memory for targets they are likely to recall and direct their search efforts accordingly, cutting off the search when it is unlikely to succeed and prioritizing the search for stronger memories. Our results suggest that metamemory is indeed adaptive and motivate the development of process-level theories that account for the dynamic interplay between monitoring and control. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Reconciling truthfulness and relevance as epistemic and decision-theoretic utility.

People use language to influence others' beliefs and actions. Yet models of communication have diverged along these lines, formalizing the speaker's objective in terms of either the listener's beliefs or actions. We argue that this divergence lies at the root of a longstanding controversy over the Gricean maxims of truthfulness and relevance. We first bridge the divide by introducing a speaker model which considers both the listener's beliefs (epistemic utility) and their actions (decision-theoretic utility). We show that formalizing truthfulness as an epistemic utility and relevance as a decision-theoretic utility reconciles the tension between them, readily explaining puzzles such as context-dependent standards of truthfulness. We then test a set of novel predictions generated by our model. We introduce a new signaling game which decouples utterances' truthfulness and relevance, then use it to conduct a pair of experiments. Our first experiment demonstrates that participants jointly maximize epistemic and decision-theoretic utility, rather than either alone. Our second experiment shows that when the two conflict, participants make a graded tradeoff rather than prioritizing one over the other. These results demonstrate that human communication cannot be reduced to influencing beliefs or actions alone. Taken together, our work provides a new foundation for grounding rational communication not only in what we believe, but in what those beliefs lead us to do. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Machine culture.

The ability of humans to create and disseminate culture is often credited as the single most important factor of our success as a species. In this Perspective, we explore the notion of 'machine culture', culture mediated or generated by machines. We argue that intelligent machines simultaneously transform the cultural evolutionary processes of variation, transmission and selection. Recommender algorithms are altering social learning dynamics. Chatbots are forming a new mode of cultural transmission, serving as cultural models. Furthermore, intelligent machines are evolving as contributors in generating cultural traits-from game strategies and visual art to scientific results. We provide a conceptual framework for studying the present and anticipated future impact of machines on cultural evolution, and present a research agenda for the study of machine culture.

Optimal nudging for cognitively bounded agents: A framework for modeling, predicting, and controlling the effects of choice architectures.

People's decisions often deviate from classical notions of rationality, incurring costs to themselves and society. One way to reduce the costs of poor decisions is to redesign the decision problems people face to encourage better choices. While often subtle, these nudges can have dramatic effects on behavior and are increasingly popular in public policy, health care, and marketing. Although nudges are often designed with psychological theories in mind, they are typically not formalized in computational terms and their effects can be hard to predict. As a result, designing nudges can be difficult and time-consuming. To address this challenge, we propose a computational framework for understanding and predicting the effects of nudges. Our approach builds on recent work modeling human decision making as adaptive use of limited cognitive resources, an approach called resource-rational analysis. In our framework, nudges change the metalevel problem the agent faces-that is, the problem of how to make a decision. This changes the optimal sequence of cognitive operations an agent should execute, which in turn influences their behavior. We show that models based on this framework can account for known effects of nudges based on default options, suggested alternatives, and information highlighting. In each case, we validate the model's predictions in an experimental process-tracing paradigm. We then show how the framework can be used to automatically construct optimal nudges, and demonstrate that these nudges improve people's decisions more than intuitive heuristic approaches. Overall, our results show that resource-rational analysis is a promising framework for formally characterizing and constructing nudges. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Rational Simplification and Rigidity in Human Planning.

Planning underpins the impressive flexibility of goal-directed behavior. However, even when planning, people can display surprising rigidity in how they think about problems (e.g., "functional fixedness") that lead them astray. How can our capacity for behavioral flexibility be reconciled with our susceptibility to conceptual inflexibility? We propose that these tendencies reflect avoidance of two cognitive costs: the cost of representing task details and the cost of switching between representations. To test this hypothesis, we developed a novel paradigm that affords participants opportunities to choose different families of simplified representations to plan. In two preregistered, online studies (Ns = 377 and 294 adults), we found that participants' optimal behavior, suboptimal behavior, and reaction time were explained by a computational model that formalized people's avoidance of representational complexity and switching. These results demonstrate how the selection of simplified, rigid representations leads to the otherwise puzzling combination of flexibility and inflexibility observed in problem solving.

Resampling reduces bias amplification in experimental social networks.

Large-scale social networks are thought to contribute to polarization by amplifying people's biases. However, the complexity of these technologies makes it difficult to identify the mechanisms responsible and evaluate mitigation strategies. Here we show under controlled laboratory conditions that transmission through social networks amplifies motivational biases on a simple artificial decision-making task. Participants in a large behavioural experiment showed increased rates of biased decision-making when part of a social network relative to asocial participants in 40 independently evolving populations. Drawing on ideas from Bayesian statistics, we identify a simple adjustment to content-selection algorithms that is predicted to mitigate bias amplification by generating samples of perspectives from within an individual's network that are more representative of the wider population. In two large experiments, this strategy was effective at reducing bias amplification while maintaining the benefits of information sharing. Simulations show that this algorithm can also be effective in more complex networks.

On the hazards of relating representations and inductive biases.

The success of models of human behavior based on Bayesian inference over logical formulas or programs is taken as evidence that people employ a "language-of-thought" that has similarly discrete and compositional structure. We argue that this conclusion problematically crosses levels of analysis, identifying representations at the algorithmic level based on inductive biases at the computational level.

Disentangling Abstraction from Statistical Pattern Matching in Human and Machine Learning.

The ability to acquire abstract knowledge is a hallmark of human intelligence and is believed by many to be one of the core differences between humans and neural network models. Agents can be endowed with an inductive bias towards abstraction through meta-learning, where they are trained on a distribution of tasks that share some abstract structure that can be learned and applied. However, because neural networks are hard to interpret, it can be difficult to tell whether agents have learned the underlying abstraction, or alternatively statistical patterns that are characteristic of that abstraction. In this work, we compare the performance of humans and agents in a meta-reinforcement learning paradigm in which tasks are generated from abstract rules. We define a novel methodology for building "task metamers" that closely match the statistics of the abstract tasks but use a different underlying generative process, and evaluate performance on both abstract and metamer tasks. We find that humans perform better at abstract tasks than metamer tasks whereas common neural network architectures typically perform worse on the abstract tasks than the matched metamers. This work provides a foundation for characterizing differences between humans and machine learning that can be used in future work towards developing machines with more human-like behavior.

The Computational Challenges of Means Selection Problems: Network Structure of Goal Systems Predicts Human Performance.

We study human performance in two classical NP-hard optimization problems: Set Cover and Maximum Coverage. We suggest that Set Cover and Max Coverage are related to means selection problems that arise in human problem-solving and in pursuing multiple goals: The relationship between goals and means is expressed as a bipartite graph where edges between means and goals indicate which means can be used to achieve which goals. While these problems are believed to be computationally intractable in general, they become more tractable when the structure of the network resembles a tree. Thus, our main prediction is that people should perform better with goal systems that are more tree-like. We report three behavioral experiments which confirm this prediction. Our results suggest that combinatorial parameters that are instrumental to algorithm design can also be useful for understanding when and why people struggle to choose between multiple means to achieve multiple goals.

Humans decompose tasks by trading off utility and computational cost.

Human behavior emerges from planning over elaborate decompositions of tasks into goals, subgoals, and low-level actions. How are these decompositions created and used? Here, we propose and evaluate a normative framework for task decomposition based on the simple idea that people decompose tasks to reduce the overall cost of planning while maintaining task performance. Analyzing 11,117 distinct graph-structured planning tasks, we find that our framework justifies several existing heuristics for task decomposition and makes predictions that can be distinguished from two alternative normative accounts. We report a behavioral study of task decomposition (N = 806) that uses 30 randomly sampled graphs, a larger and more diverse set than that of any previous behavioral study on this topic. We find that human responses are more consistent with our framework for task decomposition than alternative normative accounts and are most consistent with a heuristic-betweenness centrality-that is justified by our approach. Taken together, our results suggest the computational cost of planning is a key principle guiding the intelligent structuring of goal-directed behavior.

Learning to Learn Functions.

Humans can learn complex functional relationships between variables from small amounts of data. In doing so, they draw on prior expectations about the form of these relationships. In three experiments, we show that people learn to adjust these expectations through experience, learning about the likely forms of the functions they will encounter. Previous work has used Gaussian processes-a statistical framework that extends Bayesian nonparametric approaches to regression-to model human function learning. We build on this work, modeling the process of learning to learn functions as a form of hierarchical Bayesian inference about the Gaussian process hyperparameters.

Stress, intertemporal choice, and mitigation behavior during the COVID-19 pandemic.

Delayed gratification is an important focus of research, given its potential relationship to forms of behavior, such as savings, susceptibility to addiction, and pro-social behaviors. The COVID-19 pandemic may be one of the most consequential recent examples of this phenomenon, with people's willingness to delay gratification affecting their willingness to socially distance themselves. COVID-19 also provides a naturalistic context by which to evaluate the ecological validity of delayed gratification. This article outlines four large-scale online experiments (total N = 12, 906) where we ask participants to perform Money Earlier or Later (MEL) decisions (e.g., $5 today vs. $10 tomorrow) and to also report stress measures and pandemic mitigation behaviors. We found that stress increases impulsivity and that less stressed and more patient individuals socially distanced more throughout the pandemic. These results help resolve longstanding theoretical debates in the MEL literature as well as provide policymakers with scientific evidence that can help inform response strategies in the future. (PsycInfo Database Record (c) 2023 APA, all rights reserved).