Learning and Representation of Hierarchical Concepts in Hippocampus and Prefrontal Cortex

Task imprinting: Another mechanism of representational change?

Research from several areas suggests that mental representations adapt to the specific tasks we carry out in our environment. In this study, we propose a mechanism of adaptive representational change, task imprinting. Thereby, we introduce a computational model, which portrays task imprinting as an adaptation to specific task goals via selective storage of helpful representations in long-term memory. We test the main qualitative prediction of the model in four behavioral experiments using healthy young adults as participants. In each experiment, we assess participants' baseline representations in the beginning of the experiment, then expose participants to one of two tasks intended to shape representations differently according to our model, and finally assess any potential change in representations. Crucially, the tasks used to measure representations differ in the amount that strategic, judgmental processes play a role. The results of Experiments 1 and 2 allow us to exclude the option that representations used in more perceptual tasks become biased categorically. The results of Experiment 4 make it likely that people strategically decide given the specific task context whether they use categorical information or not. One signature of representational change was however observed: category learning practice increased the perceptual sensitivity over and above mere exposure to the same stimuli.

Learning feature relationships in CNN model via relational embedding convolution layer

Establishing the relationships among hierarchical visual attributes of objects in the visual world is crucial for human cognition. The classic convolution neural network (CNN) can successfully extract hierarchical features but ignore the relationships among features, resulting in shortcomings compared to humans in areas like interpretability and domain generalization. Recently, algorithms have introduced feature relationships by external prior knowledge and special auxiliary modules, which have been proven to bring multiple improvements in many computer vision tasks. However, prior knowledge is often difficult to obtain, and auxiliary modules bring additional consumption of computing and storage resources, which limits the flexibility and practicality of the algorithm. In this paper, we aim to drive the CNN model to learn the relationships among hierarchical deep features without prior knowledge and consumption increasing, while enhancing the fundamental performance of some aspects. Firstly, the task of learning the relationships among hierarchical features in CNN is defined and three key problems related to this task are pointed out, including the quantitative metric of connection intensity, the threshold of useless connections, and the updating strategy of relation graph. Secondly, Relational Embedding Convolution (RE-Conv) layer is proposed for the representation of feature relationships in convolution layer, followed by a scheme called use & disuse strategy which aims to address the three problems of feature relation learning. Finally, the improvements brought by the proposed feature relation learning scheme have been demonstrated through numerous experiments, including interpretability, domain generalization, noise robustness, and inference efficiency. In particular, the proposed scheme outperforms many state-of-the-art methods in the domain generalization community and can be seamlessly integrated with existing methods for further improvement. Meanwhile, it maintains comparable precision to the original CNN model while reducing floating point operations (FLOPs) by approximately 50%.

Reconciling category exceptions through representational shifts

Real-world categories often contain exceptions that disobey the perceptual regularities followed by other members. Prominent psychological and neurobiological theories indicate that exception learning relies on the flexible modulation of object representations, but the specific representational shifts key to learning remain poorly understood. Here, we leveraged behavioral and computational approaches to elucidate the representational dynamics during the acquisition of exceptions that violate established regularity knowledge. In our study, participants (n = 42) learned novel categories in which regular and exceptional items were introduced successively; we then fitted a computational model to individuals' categorization performance to infer latent stimulus representations before and after exception learning. We found that in the representational space, exception learning not only drove confusable exceptions to be differentiated from regular items, but also led exceptions within the same category to be integrated based on shared characteristics. These shifts resulted in distinct representational clusters of regular items and exceptions that constituted hierarchically structured category representations, and the distinct clustering of exceptions from regular items was associated with a high ability to generalize and reconcile knowledge of regularities and exceptions. Moreover, by having a second group of participants (n = 42) to judge stimuli's similarity before and after exception learning, we revealed misalignment between representational similarity and behavioral similarity judgments, which further highlights the hierarchical layouts of categories with regularities and exceptions. Altogether, our findings elucidate the representational dynamics giving rise to generalizable category structures that reconcile perceptually inconsistent category members, thereby advancing the understanding of knowledge formation.

Human learning of hierarchical graphs

Humans are exposed to sequences of events in the environment, and the interevent transition probabilities in these sequences can be modeled as a graph or network. Many real-world networks are organized hierarchically and while much is known about how humans learn basic transition graph topology, whether and to what degree humans can learn hierarchical structures in such graphs remains unknown. We probe the mental estimates of transition probabilities via the surprisal effect phenomenon: humans react more slowly to less expected transitions. Using mean-field predictions and numerical simulations, we show that surprisal effects are stronger for finer-level than coarser-level hierarchical transitions, and that surprisal effects at coarser levels are difficult to detect for limited learning times or in small samples. Using a serial response experiment with human participants (n=100), we replicate our predictions by detecting a surprisal effect at the finer level of the hierarchy but not at the coarser level of the hierarchy. We then evaluate the presence of a trade-off in learning, whereby humans who learned the finer level of the hierarchy better also tended to learn the coarser level worse, and vice versa. This study elucidates the processes by which humans learn sequential events in hierarchical contexts. More broadly, our work charts a road map for future investigation of the neural underpinnings and behavioral manifestations of graph learning.

Event Integration and Temporal Differentiation: How Hierarchical Knowledge Emerges in Hippocampal Subfields through Learning

Everyday life is composed of events organized by changes in contexts, with each event containing an unfolding sequence of occurrences. A major challenge facing our memory systems is how to integrate sequential occurrences within events while also maintaining their details and avoiding over-integration across different contexts. We asked if and how distinct hippocampal subfields come to hierarchically and, in parallel, represent both event context and subevent occurrences with learning. Female and male human participants viewed sequential events defined as sequences of objects superimposed on shared color frames while undergoing high-resolution fMRI. Importantly, these events were repeated to induce learning. Event segmentation, as indexed by increased reaction times at event boundaries, was observed in all repetitions. Temporal memory decisions were quicker for items from the same event compared to across different events, indicating that events shaped memory. With learning, hippocampal CA3 multivoxel activation patterns clustered to reflect the event context, with more clustering correlated with behavioral facilitation during event transitions. In contrast, in the dentate gyrus (DG), temporally proximal items that belonged to the same event became associated with more differentiated neural patterns. A computational model explained these results by dynamic inhibition in the DG. Additional similarity measures support the notion that CA3 clustered representations reflect shared voxel populations, while DG's distinct item representations reflect different voxel populations. These findings suggest an interplay between temporal differentiation in the DG and attractor dynamics in CA3. They advance our understanding of how knowledge is structured through integration and separation across time and context.

Distinct hippocampal mechanisms support concept formation and updating

Learning systems must constantly decide whether to create new representations or update existing ones. For example, a child learning that a bat is a mammal and not a bird would be best served by creating a new representation, whereas updating may be best when encountering a second similar bat. Characterizing the neural dynamics that underlie these complementary memory operations requires identifying the exact moments when each operation occurs. We address this challenge by interrogating fMRI brain activation with a computational learning model that predicts trial-by-trial when memories are created versus updated. We found distinct neural engagement in anterior hippocampus and ventral striatum for model-predicted memory create and update events during early learning. Notably, the degree of this effect in hippocampus, but not ventral striatum, significantly related to learning outcome. Hippocampus additionally showed distinct patterns of functional coactivation with ventromedial prefrontal cortex and angular gyrus during memory creation and premotor cortex during memory updating. These findings suggest that complementary memory functions, as formalized in computational learning models, underlie the rapid formation of novel conceptual knowledge, with the hippocampus and its interactions with frontoparietal circuits playing a crucial role in successful learning.

Significance statement

How do we reconcile new experiences with existing knowledge? Prominent theories suggest that novel information is either captured by creating new memories or leveraged to update existing memories, yet empirical support of how these distinct memory operations unfold during learning is limited. Here, we combine computational modeling of human learning behaviour with functional neuroimaging to identify moments of memory formation and updating and characterize their neural signatures. We find that both hippocampus and ventral striatum are distinctly engaged when memories are created versus updated; however, it is only hippocampus activation that is associated with learning outcomes. Our findings motivate a key theoretical revision that positions hippocampus is a key player in building organized memories from the earliest moments of learning.

Human Learning of Hierarchical Graphs

Humans are constantly exposed to sequences of events in the environment. Those sequences frequently evince statistical regularities, such as the probabilities with which one event transitions to another. Collectively, inter-event transition probabilities can be modeled as a graph or network. Many real-world networks are organized hierarchically and understanding how these networks are learned by humans is an ongoing aim of current investigations. While much is known about how humans learn basic transition graph topology, whether and to what degree humans can learn hierarchical structures in such graphs remains unknown. Here, we investigate how humans learn hierarchical graphs of the Sierpiński family using computer simulations and behavioral laboratory experiments. We probe the mental estimates of transition probabilities via the surprisal effect: a phenomenon in which humans react more slowly to less expected transitions, such as those between communities or modules in the network. Using mean-field predictions and numerical simulations, we show that surprisal effects are stronger for finer-level than coarser-level hierarchical transitions. Notably, surprisal effects at coarser levels of the hierarchy are difficult to detect for limited learning times or in small samples. Using a serial response experiment with human participants (n=100), we replicate our predictions by detecting a surprisal effect at the finer-level of the hierarchy but not at the coarser-level of the hierarchy. To further explain our findings, we evaluate the presence of a trade-off in learning, whereby humans who learned the finer-level of the hierarchy better tended to learn the coarser-level worse, and vice versa. Taken together, our computational and experimental studies elucidate the processes by which humans learn sequential events in hierarchical contexts. More broadly, our work charts a road map for future investigation of the neural underpinnings and behavioral manifestations of graph learning.

Neural knowledge assembly in humans and neural networks

Human understanding of the world can change rapidly when new information comes to light, such as when a plot twist occurs in a work of fiction. This flexible "knowledge assembly" requires few-shot reorganization of neural codes for relations among objects and events. However, existing computational theories are largely silent about how this could occur. Here, participants learned a transitive ordering among novel objects within two distinct contexts before exposure to new knowledge that revealed how they were linked. Blood-oxygen-level-dependent (BOLD) signals in dorsal frontoparietal cortical areas revealed that objects were rapidly and dramatically rearranged on the neural manifold after minimal exposure to linking information. We then adapt online stochastic gradient descent to permit similar rapid knowledge assembly in a neural network model.

Hierarchical categorization learning is associated with representational changes in the dorsal striatum and posterior frontal and parietal cortex

Learning and recognition can be improved by sorting novel items into categories and subcategories. Such hierarchical categorization is easy when it can be performed according to learned rules (e.g., "if car, then automatic or stick shift" or "if boat, then motor or sail"). Here, we present results showing that human participants acquire categorization rules for new visual hierarchies rapidly, and that, as they do, corresponding hierarchical representations of the categorized stimuli emerge in patterns of neural activation in the dorsal striatum and in posterior frontal and parietal cortex. Participants learned to categorize novel visual objects into a hierarchy with superordinate and subordinate levels based on the objects' shape features, without having been told the categorization rules for doing so. On each trial, participants were asked to report the category and subcategory of the object, after which they received feedback about the correctness of their categorization responses. Participants trained over the course of a one-hour-long session while their brain activation was measured using functional magnetic resonance imaging. Over the course of training, significant hierarchy learning took place as participants discovered the nested categorization rules, as evidenced by the occurrence of a learning trial, after which performance suddenly increased. This learning was associated with increased representational strength of the newly acquired hierarchical rules in a corticostriatal network including the posterior frontal and parietal cortex and the dorsal striatum. We also found evidence suggesting that reinforcement learning in the dorsal striatum contributed to hierarchical rule learning.

Is love an abstract concept? A view of concepts from an interaction-based perspective

Research concerning concepts in the cognitive sciences has been dominated by the information-processing approach, which has resulted in a certain narrowing of the range of questions and methods of investigation. Recent trends have sought to broaden the scope of such research, but they have not yet been integrated within a theoretical framework that would allow us to reconcile new perspectives with the insights already obtained. In this paper, we focus on the processes involved in early concept acquisition and demonstrate that certain aspects of these processes remain largely understudied. These aspects include the primacy of movement and coordination with others within a structured social environment as well as the importance of first-person experiences pertaining to perception and action. We argue that alternative approaches to cognition, such as ecological psychology, enactivism and interactivism, are helpful for foregrounding these understudied areas. These approaches can complement the extant research concerning concepts to help us obtain a more comprehensive view of knowledge structures, thus providing us with a new perspective on recurring problems, suggesting novel questions and enriching our methodological toolbox. This article is part of the theme issue 'Concepts in interaction: social engagement and inner experiences'.

Predictive motor activation: Modulated by expectancy or predictability?

Predicting actions is a fundamental ability that helps us to comprehend what is happening in our environment and to interact with others. The motor system was previously identified as source of action predictions. Yet, which aspect of the statistical likelihood of upcoming actions the motor system is sensitive to remains an open question. This EEG study investigated how regularities in observed actions are reflected in the motor system and utilized to predict upcoming actions. Prior to measuring EEG, participants watched videos of action sequences with different transitional probabilities. After training, participants' brain activity over motor areas was measured using EEG while watching videos of action sequences with the same statistical structure. Focusing on the mu and beta frequency bands we tested whether activity of the motor system reflects the statistical likelihood of upcoming actions. We also explored two distinct aspects of the statistical structure that capture different prediction processes, expectancy and predictability. Expectancy describes participants' expectation of the most likely action, whereas predictability represents all possible actions and their relative probabilities. Results revealed that mu and beta oscillations play different roles during action prediction. While the mu rhythm reflected anticipatory activity without any link to the statistical structure, the beta rhythm was related to the expectancy of an action. Our findings support theories proposing that the motor system underlies action prediction, and they extend such theories by showing that multiple forms of statistical information are extracted when observing action sequences. This information is integrated in the prediction generated by the neural motor system of which action is going to happen next.

What Does the Frontopolar Cortex Contribute to Goal-Directed Cognition and Action?

Understanding the unique functions of different subregions of primate prefrontal cortex has been a longstanding goal in cognitive neuroscience. Yet, the anatomy and function of one of its largest subregions (the frontopolar cortex) remain enigmatic and underspecified. Our Society for Neuroscience minisymposium Primate Frontopolar Cortex: From Circuits to Complex Behaviors will comprise a range of new anatomic and functional approaches that have helped to clarify the basic circuit anatomy of the frontal pole, its functional involvement during performance of cognitively demanding behavioral paradigms in monkeys and humans, and its clinical potential as a target for noninvasive brain stimulation in patients with brain disorders. This review consolidates knowledge about the anatomy and connectivity of frontopolar cortex and provides an integrative summary of its function in primates. We aim to answer the question: what, if anything, does frontopolar cortex contribute to goal-directed cognition and action?

The Role of Aberrant Neural Oscillations in the Hippocampal-Medial Prefrontal Cortex Circuit in Neurodevelopmental and Neurological Disorders

The hippocampus (HPC) and medial prefrontal cortex (mPFC) have well-established roles in cognition, emotion, and sensory processing. In recent years, interests have shifted towards developing a deeper understanding of the mechanisms underlying interactions between the HPC and mPFC in achieving these functions. Considerable research supports the idea that synchronized activity between the HPC and the mPFC is a general mechanism by which brain functions are regulated. In this review, we summarize current knowledge on the hippocampal-medial prefrontal cortex (HPC-mPFC) circuit in normal brain function with a focus on oscillations and highlight several neurodevelopmental and neurological disorders associated with aberrant HPC-mPFC circuitry. We further discuss oscillatory dynamics across the HPC-mPFC circuit as potentially useful biomarkers to assess interventions for neurodevelopmental and neurological disorders. Finally, advancements in brain stimulation, gene therapy and pharmacotherapy are explored as promising therapies for disorders with aberrant HPC-mPFC circuit dynamics.