About

Marvin Chun is the Richard M. Colgate Professor of Psychology and Professor of Neuroscience at Yale University. He earned his Ph.D. from MIT in 1994. His laboratory employs functional magnetic resonance imaging (fMRI) to study various aspects of cognition, including visual attention, memory, decision-making, perception, and performance. A primary focus of his research is to use fMRI to decode brain activity in order to understand how people perceive, remember, and make decisions. For example, his work involves guessing which faces people are viewing or determining whether individuals are attentive or distracted based on brain activity patterns. Additionally, he aims to use fMRI to predict individual differences in behavior, such as whether it is possible to forecast how well someone will perform a task even when they are not actively engaged in it while being scanned.

Research topics

• Psychology
• Cognitive psychology
• Computer science
• Artificial intelligence
• Neuroscience

Selected publications

• Feature consistency in transdiagnostic connectome-based models of sustained attention and autism symptoms

  medRxiv · 2026-04-03

  articleOpen access

  Abstract Sustained attention is an important neurobiological process. Difficulties with attention play a key role in neurodevelopmental disorders, such as attention-deficit/hyperactivity disorder (ADHD) and autism. Here, we identified functional connections consistently associated with sustained attention across datasets, participant populations, and fMRI scan types. We interrogated five transdiagnostic, previously published connectome-based models predicting attention and autistic phenotypes. All models were related to sustained attention, including in samples comprising participants with autism. We found that model similarity was associated with participant characteristics—including age and clinical diagnosis—and predicted behavioral measure. As expected, models predicting attention phenotypes shared more similar features with each other than models predicting autism symptoms. Furthermore, predictive features overlapped more between datasets that included participants of similar ages (i.e., youth vs. adult) and diagnostic status (autism diagnosis vs. no diagnosis). This suggests that functional connectivity patterns predicting individual differences in behavior are phenotype-specific and may vary as a function of age and clinical diagnosis. Highlights We interrogated five previously published, transdiagnostic models of sustained attention and autistic phenotypes Functional connectome-based models shared edges and networks Phenotype, age, and diagnosis were associated with model similarity

  PublisherOA PDFDOI

• Biological validity, test–retest reliability, and behavioral relevance of the single-subject brain volumetric similarity network

  NeuroImage · 2026-03-13

  articleOpen access

  The T1-weighted brain magnetic resonance imaging (MRI)-based volumetric similarity network (VSN) offers an advantage in clinical settings due to its ease of acquisition and widespread availability. However, its validity, reliability, and behavioral relevance remain unclear. The present study aimed to assess the reproducibility and utility of the VSN as a foundation for future research and clinical applications. Here, we analyzed three datasets (total N = 354), with two datasets having repeated MR runs (Dataset 1: n = 86; Dataset 2: n = 49) and two having an attention measure (Datasets 1 and 3: n = 219). For each run and participant, the VSN was generated using interregional morphological similarity metrics. We examined whether the VSN reflects the brain's cytoarchitecture and assessed its test-retest reliability by using connectome fingerprints in Datasets 1 and 2. We also examined the VSN's behavioral relevance and further tested its predictive utility using connectome-based predictive modeling in Datasets 1 and 3. The VSN defined using the z-transformed interregional correlation showed significant spatial similarity with the cytoarchitectonic covariance network (rhos = 0.23 and 0.22 in Datasets 1 and 2, respectively; p < 0.01). The VSN also yielded high test-retest reliability, demonstrated by high identification accuracy (91% and 100% in Datasets 1 and 2, respectively). However, unlike the functional connectome (r > 0.31, p < 0.01), VSNs did not reliably predict individual differences in attention (r < 0.1, p > 0.3). This study demonstrates the biological validity and high reliability of the VSN to support brain fingerprinting of individual subjects, but not individual differences in attention.

  PublisherDOI

• Optimizing functional connectivity scanning conditions for predicting autistic traits

  medRxiv · 2025-01-17 · 1 citations

  preprintOpen access

  Autism is a heterogeneous condition, and functional magnetic resonance imaging-based studies have advanced understanding of neurobiological correlates of autistic features. Little work has focused on the optimal brain states to reveal brain-phenotype relationships. Using connectome-based predictive modelling, we interrogated four datasets to determine scanning conditions that boost prediction of clinically relevant phenotypes and assess generalizability. In dataset one, a sample of youth with autism and neurotypical participants (n = 63), we found that a sustained attention task resulted in high prediction performance of autistic traits compared to a free-viewing social attention task and a resting-state condition. In dataset two (n = 25), we observed the predictive network model of autistic traits generated from the sustained attention task generalized to predict measures of attention in neurotypical adults. In datasets three and four, we determined the same predictive network model further generalized to predict measures of social responsiveness in the Autism Brain Imaging Data Exchange (n = 229) and the Healthy Brain Network (n = 643). Our data suggest an in-scanner sustained attention challenge can help delineate robust markers of autistic traits.

  PublisherOA PDFDOI

• Representation of objects, attention, and load in human prefrontal cortex

  Journal of Vision · 2025-07-15

  articleOpen access

  Visual processing often involves attending to different object features under various task loads. Past studies reported that some object and task features are encoded independently, while others are represented interactively in the human occipitotemporal cortex (OTC) and posterior parietal cortex (PPC). However, the mechanisms underlying their coding in the prefrontal cortex (PFC) are not fully understood. As a hub for cognitive control, it is possible that PFC primarily encodes task-related features, such as attention and load, but not object identity. Alternatively, it may simultaneously encode object identity alongside attention and load. If the latter holds, delineating how these distinct features are coded together in PFC can provide a mechanistic understanding of how it processes diverse information during visual tasks. In this fMRI study, 12 human participants performed an n-back task on colored visual stimuli, with task load (1-back and 2-back) and attention content (color and shape) varied orthogonally. Using multi-voxel pattern analysis, we found that PFC can decode not only task features such as attention and load, but also object identity. Comparison across brain regions showed that load representation was the strongest in PFC while object representation was the strongest in OTC. Attention representation, however, appears to be weaker in PFC than in OTC. Additional analyses revealed that while object representation only interacts with attention in OTC and PPC, it is further modulated by task loads in PFC. Apart from these differences, the three regions converge in that attention representation is modulated by load and object, while load representation remains independent of attention and object. Overall, our results extend prior findings and show a representational gradient across the human brain where object identity information decreases, and load information increases from posterior to anterior regions. We further show that PFC employs distinctive mechanisms to encode objects interactively along with task demands.

  PublisherDOI

• Modularity Measures of Functional Brain Networks Predict Individual Differences in Long‐Term Memory

  European Journal of Neuroscience · 2025-03-01 · 2 citations

  article

  Long-term memory (LTM) is crucial to daily functioning, and individuals show a wide range in LTM capacity. In this study, we ask: How does the brain's functional organization explain individual differences in LTM? We focused on two important, widely studied forms of LTM, general recognition and recollection memory. Inspired by recent work on graph theory and modularity of the brain, we explored how modularity measures of brain activity during encoding could predict individual differences in later LTM performance. Specifically, we examined two modularity measures that describe distinct aspects of network functioning: diversity-the extent a node connects with different modules-and locality-the extent a node has more connections within its own modules. Combining modularity measures and connectome-predictive modeling (CPM), a powerful framework for predicting individual differences in behavior from brain functional connectivity, we found that diversity and locality measures together significantly predicted individual differences in both general recognition and recollection memory. Modularity-based predictions were less strong than CPM models using only connectivity features. With regard to predictive neuroanatomy, we found that the default mode network was the most consistently selected brain network across our models. Our findings extend previous work on how the modularity of the brain is related to cognition and demonstrate that successful LTM is supported by critical connector hubs coordinating between and within networks during encoding. More broadly, they demonstrate the utility of a graph-based approach to reveal how modularity of brain networks relates to individual differences in LTM.

  PublisherDOI

• The coding of spiky objects in human occipitotemporal and posterior parietal cortices

  Journal of Vision · 2025-07-15

  articleOpen accessSenior author

  An important aspect of understanding primate vision is to determine the key features in visual object coding. In the macaque inferotemporal (IT) cortex, spikiness and animacy are two principal features mediating the coding of a diverse array of objects, with distinctive IT subregions preferring unique combinations of these features (Bao et al., 2020). Similar results, however, were not observed in the human brain. Here, we constructed well-matched spiky and stubby stimuli and reexamined their coding in the human brain with fMRI. Using the output of a convolutional neural network shown to mirror the macaque IT cortex in spiky and stubby object representations, we selected pairs of inanimate spiky and stubby objects matched for semantic category (e.g., tripod vs. camera). The object images were further equated in low-level visual features, including luminance, contrast, and spatial frequency. Across 12 participants, contrasting spiky with stubby objects consistently revealed three areas preferring spiky objects: a small bilateral ventral activation between face- and scene-selective areas (proximal to the location of the macaque’s inanimate-spiky area), a large bilateral lateral activation within a separately localized body-selective area, and a bilateral dorsal activation along superior/anterior intra-parietal sulcus. Meanwhile, no areas were found to prefer stubby objects. The same three spiky areas were also activated when participants viewed matching animate spiky-stubby stimulus pairs (e.g., crane vs. penguin), indicating their spiky-object preference is animacy-independent. Furthermore, within the lateral body-selective area, only the posterior part truly preferred animacy/bodies, as it was activated by both the animate spiky and stubby objects compared to the inanimate ones; the anterior part of the lateral body-selective area prefers spikiness but not animacy/bodies. Together, these results document for the first time a network of human brain areas preferring spiky over stubby objects and highlight the potential importance of spikiness in human visual object perception.

  PublisherDOI

• Author response for "Modularity Measures of Functional Brain Networks Predict Individual Differences in Long-Term Memory"

  2025-02-08

  peer-review
  PublisherDOI

• Ongoing thoughts at rest reflect functional brain organization and behavior

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-08-19 · 4 citations

  preprintOpen access

  Abstract Resting-state functional connectivity (rsFC)—brain connectivity observed when people rest with no external tasks—predicts individual differences in behavior. Yet, rest is not idle; it involves streams of thoughts. Are these ongoing thoughts reflected in FC and do they contribute to the relationship between rsFC and behavior? To test this question, we developed an annotated rest paradigm where participants rated and verbally described their thoughts after each rest period during functional MRI ( N = 60). Our findings revealed rich and idiosyncratic thoughts across individuals. Similarity in thoughts was associated with more similar FC patterns within and across individuals. In addition, both thought ratings and topics could be decoded from FC. Furthermore, neuromarkers of these thoughts generalized to unseen individuals in the Human Connectome Project dataset ( N = 908), where decoded thought patterns during rest predicted positive vs. negative trait-level individual differences. Together, our findings reveal that ongoing thoughts at rest are reflected in brain dynamics and these network patterns predict everyday cognition and experiences. Understanding subjective in-scanner experiences is thus crucial in characterizing the relationship between individual differences in functional brain organization and behavior.

  PublisherOA PDFDOI

• Using fMRI Representations of Single Objects to Predict Multiple Objects in Working Memory in Human Occipitotemporal and Posterior Parietal Cortices

  Journal of Neuroscience · 2025-11-14

  articleOpen accessSenior author

  Research in visual perception has shown that in sensory areas, neural responses to a pair of objects presented together can be approximated by the linear average of the responses of each object shown alone. In this study, we ask if such an averaging relationship is unique to perceptual representations or if it also applies to representations maintained in visual working memory (VWM). By examining fMRI response pattern averaging across two experiments in both male and female human participants, we found that after properly accounting for task factors such as load, an averaging relationship also applies to representations formed in VWM. Specifically, VWM representations for two items can be approximated by the linear average of the VWM representations of each component item in both human occipitotemporal cortex (including early visual areas) and posterior parietal cortex. Although response averaging was originally proposed as a mechanism to combat distortion in representation due to neuronal response saturation in perception, the present study shows that even when response amplitudes were much lower in VWM compared with those in visual perception, an averaging relationship is still present for neural representations formed in VWM. This likely stems from the need to reduce interference among the concurrently stored items in VWM to maintain their representational independence. As an experimental method, response averaging may constitute an efficient yet simple tool to probe response independence in the human brain beyond perception and VWM.

  PublisherOA PDFDOI

• Representing Visual Objects, Attention, and Load in Human Occipito-temporal and Posterior Parietal Cortices

  Journal of Cognitive Neuroscience · 2025-01-01 · 5 citations

  articleSenior author

  Stability and adaptability are two essential components of everyday vision, enabling us to maintain an object's identity as we attend to different features under varying task loads. We hypothesize that these two components of vision are supported by the interactions among object, attention, and load representations, and the interplay between the human occipito-temporal cortex (OTC), given its visual representation invariance, and the posterior parietal cortex (PPC), an adaptive visual processing center. To test this, human participants performed four tasks on the same stream of colored objects with varying attention (attending to color or shape) and load (1-back or 2-back repetition detection). Although the exact neural mechanisms differ in how object, attention, and load modulate neural responses, by placing them as different factors in the same visual representational space using fMRI pattern decoding, we directly compared their effects on visual responses and interactions. We found significant object, attention, and load representations across OTC and PPC, with a gradual transition from more object-sensitive representations in OTC to more attention- and load-sensitive representations in PPC. Notably, object, attention, and load representations showed significant interactions and generalizations across changes with each other in both OTC and PPC. When objects were held constant, attention and load were represented independently of each other, showing their neural separability. Together, the invariant and adaptive nature of object, attention, and load representations in OTC and PPC provides both stability in visual processing and adaptation to the ever-changing visual input and task demands.

  PublisherDOI

Recent grants

• Whole Brain Functional Connectivity Measures of Attention

  NSF · $492k · 2016–2019

• NIH Grant R01EY014193

  NIH · $3.2M · 2012

• Whole Brain Functional Connectivity Measures of Attention

  NIH · $1.7M · 2016–2021

• NIH Grant F32EY006592

  NIH · $41k

Frequent coauthors

• Monica D. Rosenberg

  48 shared

• Kwangsun Yoo

  Sungkyunkwan University

  28 shared

• R. Todd Constable

  Yale University

  26 shared

• Dustin Scheinost

  Yale University

  23 shared

• Yuhong Jiang

  China University of Geosciences

  23 shared

• Nicholas B. Turk‐Browne

  Yale University

  20 shared

• Do-Joon Yi

  19 shared

• Yaoda Xu

  Yale University

  18 shared

Education

• Ph.D., Brain and Cognitive Science

  Massachusetts Institute of Technology