Investigating mediodorsal thalamus representations underlying human cognitive flexibility

PROJECT SUMMARY The human brain's ability to learn and execute behaviors tailored to environmental contingencies is crucial for adaptive cognition. Central to this capacity are neural representations of context, which organize the associations between sensory features and behavior utility. Despite their significance, the neural mechanisms underlying the encoding and updating of context representations remain poorly understood. This gap in knowledge is particularly relevant for understanding cognitive deficits in psychiatric disorders such as schizophrenia and ADHD. Our research proposes to test the hypothesis that the human mediodorsal (MD) thalamus is critical for encoding, updating, and generalizing context representations. This hypothesis is grounded in empirical and theoretical contributions from our team, suggesting the MD thalamus's essential role in context-guided cognitive control and flexibility. Our previous studies have shown that damage to the MD thalamus impairs task switching and working memory in humans, consistent with findings from non-human animal models. Additionally, single- unit recordings in animals reveal that MD neurons rapidly encode task context, a process dependent on the convergence of prefrontal afferents. Preliminary fMRI data from our labs indicate that the human MD thalamus tracks task context and its updates following switches. However, a critical gap exists in interpreting these results due to the lack of quantitative models and advanced neuroimaging approaches to delineate the specific representations and computations carried out by the human MD. To address this gap, we will use a computational cognitive neuroscience approach, integrating computational models with high-resolution 7T MRI and advanced neuroimaging analyses. Our research has three specific Aims. In Aim 1, we will ask whether the MD thalamus encodes context representations that organizes how working memory guides task selection. In Aim 2, we will ask whether the MD encodes context prediction errors that switch prefrontal task representations flexibly. In Aim 3, we will ask whether MD context representations enable generalization to novel stimulus- response contingencies during decision-making. For all aims, we will develop computational models that specify both cognitive processes and neural implementations to predict behavior and guide neuroimaging data analyses. This collaborative effort between the MPIs aims to establish a new conceptual and empirical framework for understanding thalamic computation in humans, with significant implications for theories of cognitive control, adaptive human behavior, and cognitive dysfunction.