Similarities and differences in spatial and non-spatial cognitive maps

Humans flexibly integrate social information despite interindividual differences in reward

There has been much progress in understanding human social learning, including recent studies integrating social information into the reinforcement learning framework. Yet previous studies often assume identical payoffs between observer and demonstrator, overlooking the diversity of social information in real-world interactions. We address this gap by introducing a socially correlated bandit task that accommodates payoff differences among participants, allowing for the study of social learning under more realistic conditions. Our Social Generalization (SG) model, tested through evolutionary simulations and two online experiments, outperforms existing models by incorporating social information into the generalization process, but treating it as noisier than individual observations. Our findings suggest that human social learning is more flexible than previously believed, with the SG model indicating a potential resource-rational trade-off where social learning partially replaces individual exploration. This research highlights the flexibility of humans' social learning, allowing us to integrate social information from others with different preferences, skills, or goals.

Testing the convergent validity, domain generality, and temporal stability of selected measures of people's tendency to explore

Given the ubiquity of exploration in everyday life, researchers from many disciplines have developed methods to measure exploratory behaviour. There are therefore many ways to quantify and measure exploration. However, it remains unclear whether the different measures (i) have convergent validity relative to one another, (ii) capture a domain general tendency, and (iii) capture a tendency that is stable across time. In a sample of 678 participants, we found very little evidence of convergent validity for the behavioural measures (Hypothesis 1); most of the behavioural measures lacked sufficient convergent validity with one another or with the self-reports. In psychometric modelling analyses, we could not identify a good fitting model with an assumed general tendency to explore (Hypothesis 2); the best fitting model suggested that the different behavioural measures capture behaviours that are specific to the tasks. In a subsample of 254 participants who completed the study a second time, we found that the measures had stability across an 1 month timespan (Hypothesis 3). Therefore, although there were stable individual differences in how people approached each task across time, there was no generalizability across tasks, and drawing broad conclusions about exploratory behaviour from studies using these tasks may be problematic. The Stage 1 protocol for this Registered Report was accepted in principle on 2nd December 2022 https://doi.org/10.6084/m9.figshare.21717407.v1 . The protocol, as accepted by the journal, can be found at https://doi.org/10.17605/OSF.IO/64QJU .

Shared structure facilitates working memory of multiple sequences

Daily experiences often involve the processing of multiple sequences, yet storing them challenges the limited capacity of working memory (WM). To achieve efficient memory storage, relational structures shared by sequences would be leveraged to reorganize and compress information. Here, participants memorized a sequence of items with different colors and spatial locations and later reproduced the full color and location sequences one after another. Crucially, we manipulated the consistency between location and color sequence trajectories. First, sequences with consistent trajectories demonstrate improved memory performance and a trajectory correlation between reproduced color and location sequences. Second, sequences with consistent trajectories show neural reactivation of common trajectories, and display spontaneous replay of color sequences when recalling locations. Finally, neural reactivation correlates with WM behavior. Our findings suggest that a shared common structure is leveraged for the storage of multiple sequences through compressed encoding and neural replay, together facilitating efficient information organization in WM.

Differential effects of bilateral hippocampal CA3 damage on the implicit learning and recognition of complex event sequences

Learning regularities in the environment is a fundament of human cognition, which is supported by a network of brain regions that include the hippocampus. In two experiments, we assessed the effects of selective bilateral damage to human hippocampal subregion CA3, which was associated with autobiographical episodic amnesia extending ~50 years prior to the damage, on the ability to recognize complex, deterministic event sequences presented either in a spatial or a non-spatial configuration. In contrast to findings from related paradigms, modalities, and homologue species, hippocampal damage did not preclude recognition memory for an event sequence studied and tested at four spatial locations, whereas recognition memory for an event sequence presented at a single location was at chance. In two additional experiments, recognition memory for novel single-items was intact, whereas the ability to recognize novel single-items in a different location from that presented at study was at chance. The results are at variance with a general role of the hippocampus in the learning and recognition of complex event sequences based on non-adjacent spatial and temporal dependencies. We discuss the impact of the results on established theoretical accounts of the hippocampal contributions to implicit sequence learning and episodic memory.

Transcranial direct current stimulation suggests a causal role of the medial prefrontal cortex in learning social hierarchy

Social hierarchies can be inferred through observational learning of social relationships between individuals. Yet, little is known about the causal role of specific brain regions in learning hierarchies. Here, using transcranial direct current stimulation, we show a causal role of the medial prefrontal cortex (mPFC) in learning social versus non-social hierarchies. In a Training phase, participants acquired knowledge about social and non-social hierarchies by trial and error. During a Test phase, they were presented with two items from hierarchies that were never encountered together, requiring them to make transitive inferences. Anodal stimulation over mPFC impaired social compared with non-social hierarchy learning, and this modulation was influenced by the relative social rank of the members (higher or lower status). Anodal stimulation also impaired transitive inference making, but only during early blocks before learning was established. Together, these findings demonstrate a causal role of the mPFC in learning social ranks by observation.

GUANinE v1.0: Benchmark Datasets for Genomic AI Sequence-to-Function Models

Computational genomics increasingly relies on machine learning methods for genome interpretation, and the recent adoption of neural sequence-to-function models highlights the need for rigorous model specification and controlled evaluation, problems familiar to other fields of AI. Research strategies that have greatly benefited other fields - including benchmarking, auditing, and algorithmic fairness - are also needed to advance the field of genomic AI and to facilitate model development. Here we propose a genomic AI benchmark, GUANinE, for evaluating model generalization across a number of distinct genomic tasks. Compared to existing task formulations in computational genomics, GUANinE is large-scale, de-noised, and suitable for evaluating pretrained models. GUANinE v1.0 primarily focuses on functional genomics tasks such as functional element annotation and gene expression prediction, and it also draws upon connections to evolutionary biology through sequence conservation tasks. The current GUANinE tasks provide insight into the performance of existing genomic AI models and non-neural baselines, with opportunities to be refined, revisited, and broadened as the field matures. Finally, the GUANinE benchmark allows us to evaluate new self-supervised T5 models and explore the tradeoffs between tokenization and model performance, while showcasing the potential for self-supervision to complement existing pretraining procedures.

The Spread of Beliefs in Partially Modularized Communities

Many life-influencing social networks are characterized by considerable informational isolation. People within a community are far more likely to share beliefs than people who are part of different communities. The spread of useful information across communities is impeded by echo chambers (far greater connectivity within than between communities) and filter bubbles (more influence of beliefs by connected neighbors within than between communities). We apply the tools of network analysis to organize our understanding of the spread of beliefs across modularized communities and to predict the effect of individual and group parameters on the dynamics and distribution of beliefs. In our Spread of Beliefs in Modularized Communities (SBMC) framework, a stochastic block model generates social networks with variable degrees of modularity, beliefs have different observable utilities, individuals change their beliefs on the basis of summed or average evidence (or intermediate decision rules), and parameterized stochasticity introduces randomness into decisions. SBMC simulations show surprising patterns; for example, increasing out-group connectivity does not always improve group performance, adding randomness to decisions can promote performance, and decision rules that sum rather than average evidence can improve group performance, as measured by the average utility of beliefs that the agents adopt. Overall, the results suggest that intermediate degrees of belief exploration are beneficial for the spread of useful beliefs in a community, and so parameters that pull in opposite directions on an explore-exploit continuum are usefully paired.

The journey within: mental navigation as a novel framework for understanding psychotherapeutic transformation

BACKGROUND

Despite the demonstrated efficacy of psychotherapy, the precise mechanisms that drive therapeutic transformations have posed a challenge and still remain unresolved. Here, we suggest a potential solution to this problem by introducing a framework based on the concept of mental navigation. It refers to our ability to navigate our cognitive space of thoughts, ideas, concepts, and memories, similar to how we navigate physical space. We start by analyzing the neural, cognitive, and experiential constituents intrinsic to mental navigation. Subsequently, we posit that the metaphoric spatial language we employ to articulate introspective experiences (e.g., "unexplored territory" or "going in circles") serves as a robust marker of mental navigation.

METHODS

Using large text corpora, we compared the utilization of spatial language between transcripts of psychotherapy sessions (≈ 12 M. words), casual everyday conversations (≈ 12 M. words), and fictional dialogues in movies (≈ 14 M. words). We also examined 110 psychotherapy transcripts qualitatively to discern patterns and dynamics associated with mental navigation.

RESULTS

We found a notable increase in the utilization of spatial metaphors during psychotherapy compared to casual everyday dialogues (U = 192.0, p = .001, d = 0.549) and fictional conversations (U = 211, p < .001, d = 0.792). In turn, analyzing the usage of non-spatial metaphors, we did not find significant differences between the three datasets (H = 0.682, p = 0.710). The qualitative analysis highlighted specific examples of mental navigation at play.

CONCLUSION

Mental navigation might underlie the psychotherapy process and serve as a robust framework for understanding the transformative changes it brings about.

Hippocampal spatio-predictive cognitive maps adaptively guide reward generalization

The brain forms cognitive maps of relational knowledge-an organizing principle thought to underlie our ability to generalize and make inferences. However, how can a relevant map be selected in situations where a stimulus is embedded in multiple relational structures? Here, we find that both spatial and predictive cognitive maps influence generalization in a choice task, where spatial location determines reward magnitude. Mirroring behavior, the hippocampus not only builds a map of spatial relationships but also encodes the experienced transition structure. As the task progresses, participants' choices become more influenced by spatial relationships, reflected in a strengthening of the spatial map and a weakening of the predictive map. This change is driven by orbitofrontal cortex, which represents the degree to which an outcome is consistent with the spatial rather than the predictive map and updates hippocampal representations accordingly. Taken together, this demonstrates how hippocampal cognitive maps are used and updated flexibly for inference.

Exploration patterns shape cognitive map learning

Spontaneous, volitional spatial exploration is crucial for building up a cognitive map of the environment. However, decades of research have primarily measured the fidelity of cognitive maps after discrete, controlled learning episodes. We know little about how cognitive maps are formed during naturalistic free exploration. Here, we investigated whether exploration trajectories predicted cognitive map accuracy, and how these patterns were shaped by environmental structure. In two experiments, participants freely explored a previously unfamiliar virtual environment. We related their exploration trajectories to a measure of how long they spent in areas with high global environmental connectivity (integration, as assessed by space syntax). In both experiments, we found that participants who spent more time on paths that offered opportunities for integration formed more accurate cognitive maps. Interestingly, we found no support for our pre-registered hypothesis that self-reported trait differences in navigation ability would mediate this relationship. Our findings suggest that exploration patterns predict cognitive map accuracy, even for people who self-report low ability, and highlight the importance of considering both environmental structure and individual variability in formal theory- and model-building.

An instance-based model account of the benefits of varied practice in visuomotor skill

Exposing learners to variability during training has been demonstrated to improve performance in subsequent transfer testing. Such variability benefits are often accounted for by assuming that learners are developing some general task schema or structure. However much of this research has neglected to account for differences in similarity between varied and constant training conditions. In a between-groups manipulation, we trained participants on a simple projectile launching task, with either varied or constant conditions. We replicate previous findings showing a transfer advantage of varied over constant training. Furthermore, we show that a standard similarity model is insufficient to account for the benefits of variation, but, if the model is adjusted to assume that varied learners are tuned towards a broader generalization gradient, then a similarity-based model is sufficient to explain the observed benefits of variation. Our results therefore suggest that some variability benefits can be accommodated within instance-based models without positing the learning of some schemata or structure.

Time pressure changes how people explore and respond to uncertainty

How does time pressure influence exploration and decision-making? We investigated this question with several four-armed bandit tasks manipulating (within subjects) expected reward, uncertainty, and time pressure (limited vs. unlimited). With limited time, people have less opportunity to perform costly computations, thus shifting the cost-benefit balance of different exploration strategies. Through behavioral, reinforcement learning (RL), reaction time (RT), and evidence accumulation analyses, we show that time pressure changes how people explore and respond to uncertainty. Specifically, participants reduced their uncertainty-directed exploration under time pressure, were less value-directed, and repeated choices more often. Since our analyses relate uncertainty to slower responses and dampened evidence accumulation (i.e., drift rates), this demonstrates a resource-rational shift towards simpler, lower-cost strategies under time pressure. These results shed light on how people adapt their exploration and decision-making strategies to externally imposed cognitive constraints.

Multitask learning over shared subspaces

This paper uses constructs from machine learning to define pairs of learning tasks that either shared or did not share a common subspace. Human subjects then learnt these tasks using a feedback-based approach and we hypothesised that learning would be boosted for shared subspaces. Our findings broadly supported this hypothesis with either better performance on the second task if it shared the same subspace as the first, or positive correlations over task performance for shared subspaces. These empirical findings were compared to the behaviour of a Neural Network model trained using sequential Bayesian learning and human performance was found to be consistent with a minimal capacity variant of this model. Networks with an increased representational capacity, and networks without Bayesian learning, did not show these transfer effects. We propose that the concept of shared subspaces provides a useful framework for the experimental study of human multitask and transfer learning.

How does knowledge gained from previous experience affect learning of new tasks? This question of “Transfer Learning” has been addressed by teachers, psychologists, and more recently by researchers in the fields of neural networks and machine learning. Leveraging constructs from machine learning, we designed pairs of learning tasks that either shared or did not share a common subspace. We compared the dynamics of transfer learning in humans with those of a multitask neural network model, finding that human performance was consistent with a minimal capacity variant of the model. Learning was boosted in the second task if the same subspace was shared between tasks. Additionally, accuracy between tasks was positively correlated but only when they shared the same subspace. Our results highlight the roles of subspaces, showing how they could act as a learning boost if shared, and be detrimental if not.

Human Representation Learning

The central theme of this review is the dynamic interaction between information selection and learning. We pose a fundamental question about this interaction: How do we learn what features of our experiences are worth learning about? In humans, this process depends on attention and memory, two cognitive functions that together constrain representations of the world to features that are relevant for goal attainment. Recent evidence suggests that the representations shaped by attention and memory are themselves inferred from experience with each task. We review this evidence and place it in the context of work that has explicitly characterized representation learning as statistical inference. We discuss how inference can be scaled to real-world decisions by approximating beliefs based on a small number of experiences. Finally, we highlight some implications of this inference process for human decision-making in social environments.

Development of directed and random exploration in children

Are young children just random explorers who learn serendipitously? Or are even young children guided by uncertainty-directed sampling, seeking to explore in a systematic fashion? We study how children between the ages of 4 and 9 search in an explore-exploit task with spatially correlated rewards, where exhaustive exploration is infeasible and not all options can be experienced. By combining behavioral data with a computational model that decomposes search into similarity-based generalization, uncertainty-directed exploration, and random exploration, we map out developmental trajectories of generalization and exploration. The behavioral data show strong developmental differences in children's capability to exploit environmental structure, with performance and adaptiveness of sampling decisions increasing with age. Through model-based analyses, we disentangle different forms of exploration, finding signature of both uncertainty-directed and random exploration. The amount of random exploration strongly decreases as children get older, supporting the notion of a developmental "cooling off" process that modulates the randomness in sampling. However, even at the youngest age range, children do not solely rely on random exploration. Even as random exploration begins to taper off, children are actively seeking out options with high uncertainty in a goal-directed fashion, and using inductive inferences to generalize their experience to novel options. Our findings provide critical insights into the behavioral and computational principles underlying the developmental trajectory of learning and exploration.

From Topological Analyses to Functional Modeling: The Case of Hippocampus

Topological data analyses are widely used for describing and conceptualizing large volumes of neurobiological data, e.g., for quantifying spiking outputs of large neuronal ensembles and thus understanding the functions of the corresponding networks. Below we discuss an approach in which convergent topological analyses produce insights into how information may be processed in mammalian hippocampus—a brain part that plays a key role in learning and memory. The resulting functional model provides a unifying framework for integrating spiking data at different timescales and following the course of spatial learning at different levels of spatiotemporal granularity. This approach allows accounting for contributions from various physiological phenomena into spatial cognition—the neuronal spiking statistics, the effects of spiking synchronization by different brain waves, the roles played by synaptic efficacies and so forth. In particular, it is possible to demonstrate that networks with plastic and transient synaptic architectures can encode stable cognitive maps, revealing the characteristic timescales of memory processing.