Jonathan D. Cohen’s research while affiliated with Princeton University and other places

Learning requires changing the brain. This typically occurs through experience, study, or instruction. We report an alternate route for humans to acquire visual knowledge, through the direct sculpting of activity patterns in the human brain that mirror those expected to arise through learning. We used neurofeedback from closed-loop real-time functional MRI to create new categories of visual objects in the brain, without the participants’ explicit awareness. After neural sculpting, participants exhibited behavioral and neural biases for the learned, but not for the control categories. The ability to sculpt new perceptual distinctions into the human brain offers a noninvasive research paradigm for causal testing of the link between neural representations and behavior. As such, beyond its current application to perception, our work potentially has broad relevance for advancing understanding in other domains of cognition such as decision-making, memory, and motor control.

Background Individuals with major depressive disorder (MDD) can experience reduced motivation and cognitive function, leading to challenges with goal-directed behavior. When selecting goals, people maximize ‘expected value’ by selecting actions that maximize potential reward while minimizing associated costs, including effort ‘costs’ and the opportunity cost of time. In MDD, differential weighing of costs and benefits are theorized mechanisms underlying changes in goal-directed cognition and may contribute to symptom heterogeneity. Methods We used the Effort Foraging Task to quantify cognitive and physical effort costs, and patch leaving thresholds in low effort conditions (reflecting perceived opportunity cost of time) and investigated their shared versus distinct relationships to clinical features in participants with MDD ( N = 52, 43 in-episode) and comparisons ( N = 27). Results Contrary to our predictions, none of the decision-making measures differed with MDD diagnosis. However, each of the measures was related to symptom severity, over and above effects of ability (i.e. performance). Greater anxiety symptoms were selectively associated with lower cognitive effort cost (i.e. greater willingness to exert effort). Anhedonia and behavioral apathy were associated with increased physical effort costs. Finally, greater overall depression was related to decreased patch leaving thresholds. Conclusions Markers of effort-based decision-making may inform understanding of MDD heterogeneity. Increased willingness to exert cognitive effort may contribute to anxiety symptoms such as worry. Decreased leaving threshold associations with symptom severity are consistent with reward rate-based accounts of reduced vigor in MDD. Future research should address subtypes of depression with or without anxiety, which may relate differentially to cognitive effort decisions.

Recent work has documented striking heterogeneity in the performance of state-of-the-art vision language models (VLMs), including both multimodal language models and text-to-image models. These models are able to describe and generate a diverse array of complex, naturalistic images, yet they exhibit surprising failures on basic multi-object reasoning tasks -- such as counting, localization, and simple forms of visual analogy -- that humans perform with near perfect accuracy. To better understand this puzzling pattern of successes and failures, we turn to theoretical accounts of the binding problem in cognitive science and neuroscience, a fundamental problem that arises when a shared set of representational resources must be used to represent distinct entities (e.g., to represent multiple objects in an image), necessitating the use of serial processing to avoid interference. We find that many of the puzzling failures of state-of-the-art VLMs can be explained as arising due to the binding problem, and that these failure modes are strikingly similar to the limitations exhibited by rapid, feedforward processing in the human brain.

Background: Brain reserve - the brain's resilience to age-related change or damage - provides protection against cognitive decline. The cerebellum is relatively unstudied as a contributor to brain reserve. This study investigates cerebellar brain reserve in the largest cohort to date. Methods: We used data from the Human Connectome Project (n=708, 36-100yrs), UK Biobank (n=45,013, 44-81yrs), and ADNI (n=1,423, 56-95yrs). ADNI participants were cognitively normal or had a diagnosis of mild cognitive impairment or Alzheimer's disease (AD) dementia. We examined associations between cerebellar tissue volume, age, Montreal Cognitive Assessment (MoCA) scores, global PET amyloid burden, and APOE genotype. Findings: HCP-Aging data revealed heterogenous aging-associated changes in cerebellar volume, with the greatest effects in posterior hemispheric regions (crus I) (Bonferroni-corrected, p<0.05). MoCA scores were associated with higher tissue density in the cerebellum (p<0.0001) to the same extent as neocortex, and MoCA scores coupled most strongly with posterior cerebellar cortex. Strikingly, greater volume in MoCA visuospatial-related cerebellar cortex protected against aging-related cognitive decline (p=0.0001). We replicated tissue aging results in the UK Biobank with the greatest aging-related effects in posterior cerebellum (p<0.0001), and an association of greater cerebellar volumes with less cognitive decline (Trails Making-B: p<0.00001; Digit Symbol Substitution: p=0.034). AD patients with low amyloid-beta burden (AB-) exhibited the strongest cerebellar association with MoCA (volume x group, AB- AD: p=0.0001). In AB- individuals, APOE e4/e4 carriers showed the greatest effect with MoCA (volume x APOE, e4/e4: p=0.017). Interpretation: Our large-scale study demonstrates a potentially strong role for the cerebellum in mitigating cognitive decline. The persistence of this protection in APOE e4 carriers reshapes our understanding of reserve and AD risk. Our findings open the cerebellum as a novel target for future clinical research on brain reserve in aging populations. Funding: National Science Foundation, National Academies of Sciences, Engineering and Medicine, National Institutes of Health.

Despite perinatal damage to the cerebellum being one of the highest risk factors for later being diagnosed with autism spectrum disorder (ASD), it is not yet clear how the cerebellum might influence the development of cerebral cortex and whether this co-developmental process is distinct between neurotypical and ASD children. Leveraging a large structural brain MRI dataset of neurotypical children and those diagnosed with ASD, we examined whether structural variation in cerebellar tissue across individuals was correlated with neocortical variation during development, including the thalamus as a coupling factor. We found that the thalamus plays a distinct role in moderating cerebro-cerebellar structural coordination in ASD. Notably, structural coupling between cerebellum, thalamus, and neocortex was strongest in younger childhood and waned by early adolescence, mirroring a previously undescribed trajectory of behavioral development between ASD and neurotypical children. Complementary functional connectivity analyses likewise revealed atypical connectivity between cerebellum and neocortex in ASD. This relationship was particularly prominent in a model of cerebellar structure predicting functional connectivity, where ASD and neurotypical children showed divergent patterns. Interestingly, these functional-structural relationships became more prominent with age, while structural effects were most prominent earlier in childhood, and showed significant lateralization. This pattern may suggest a developmental sequence where early uncoordinated structural growth amongst regions is followed by increasingly atypical functional synchronization. These findings provide multimodal evidence in the living brain for a cerebellar diaschisis model of autism, where both increased cerebellar-cerebral structural coupling and altered functional connectivity in cerebro-cerebellar pathways contribute to the ontogeny of this neurodevelopmental disorder.

The ability to switch between different tasks is a critical component of adaptive cognitive functioning, but a mechanistic understanding of this capacity has remained elusive. Longstanding debates over whether task switching requires active preparation remain hotly contested, in large part due to the difficulty of inferring task preparation from behavior alone. We make progress on this debate by quantifying neural task representations through high-dimensional linear dynamical systems fit to human electroencephalographic recordings. We find that these dynamical systems are highly predictive of macroscopic neural activity, and reveal neural signatures of active preparation that are shared with task-optimized neural networks. These findings help inform a classic debate about how we control our cognition, and offer a promising new paradigm for neuroimaging analysis.

Current models frame the allocation of cognitive control as a process of expected utility maximization. The benefits of a candidate control signal are weighed against its costs (e.g., opportunity costs). Recent theorizing has found that, despite promoting the counterintuitive behavior of longer deliberation, which is less rewarding in the short term, it is nevertheless normative to account for the value of learning when determining control allocation. Here, we sought to test this proposal by examining whether people were willing to allocate greater control and thereby expend greater effort (e.g., deliberate for longer) when they perceived a task to be learnable compared to when they did not. We found that participants' proficiency and learning rate in the first block of a simple perceptual dot-motion task were able to predict their willingness to deliberate in a second block. These findings support the hypothesis that agents consider learnability when allocating cognitive control, and comply with a formal model of control allocation that considers the future discounted value of learning on reward.

Deep neural networks have made tremendous gains in emulating human-like intelligence, and have been used increasingly as ways of understanding how the brain may solve the complex computational problems on which this relies. However, these still fall short of, and therefore fail to provide insight into how the brain supports strong forms of generalization of which humans are capable. One such case is out-of-distribution (OOD) generalization – successful performance on test examples that lie outside the distribution of the training set. Here, we identify properties of processing in the brain that may contribute to this ability. We describe a two-part algorithm that draws on specific features of neural computation to achieve OOD generalization, and provide a proof of concept by evaluating performance on two challenging cognitive tasks. First we draw on the fact that the mammalian brain represents metric spaces using grid cell code (e.g., in the entorhinal cortex): abstract representations of relational structure, organized in recurring motifs that cover the representational space. Second, we propose an attentional mechanism that operates over the grid cell code using determinantal point process (DPP), that we call DPP attention (DPP-A) – a transformation that ensures maximum sparseness in the coverage of that space. We show that a loss function that combines standard task-optimized error with DPP-A can exploit the recurring motifs in the grid cell code, and can be integrated with common architectures to achieve strong OOD generalization performance on analogy and arithmetic tasks. This provides both an interpretation of how the grid cell code in the mammalian brain may contribute to generalization performance, and at the same time a potential means for improving such capabilities in artificial neural networks.

Understanding the mechanisms enabling the learning and flexible use of knowledge in context-appropriate ways has been a major focus of research in the study of both semantic cognition and cognitive control. We present a unified model of semantics and control that addresses these questions from both perspectives. The model provides a coherent view of how semantic knowledge, and the ability to flexibly access and deploy that knowledge to meet current task demands, arises from end-to-end learning of the statistics of the environment. We show that the model addresses unresolved issues from both literatures, including how control operates over features that covary with one another and how control representations themselves are structured and emerge through learning, through a series of behavioral experiments and simulations. We conclude by discussing the implications of our approach to other fundamental questions in cognitive science, machine learning, and artificial intelligence.