scholarly journals Deep learning models of cognitive processes constrained by human brain connectomes

2021 ◽  
Author(s):  
Yu Zhang ◽  
Nicolas et Farrugia ◽  
Pierre Bellec
Keyword(s):  
Deep Learning ◽  
Cognitive Processes ◽  
Brain Activity ◽  
Neural Substrates ◽  
Brain Regions ◽  
Human Cognition ◽  
Experimental Conditions ◽  
Cognitive States ◽  
Learning Techniques ◽  
Human Connectome Project

Brain decoding aims to infer cognitive states from recordings of brain activity. Current literature has mainly focused on isolated brain regions engaged in specific experimental conditions, but ignored the integrative nature of cognitive processes recruiting distributed brain networks. To tackle this issue, we propose a connectome-based graph neural network to integrate distributed patterns of brain activity in a multiscale manner, ranging from localized brain regions, to a specific brain circuit/network and towards the full brain. We evaluate the decoding model using a large task-fMRI database from the human connectome project. By implementing connectomic constraints and multiscale interactions in deep graph convolutions, the model achieves high accuracy of decoding 21 cognitive states (93%, chancel level: 4.8%) and shows high robustness against adversarial attacks on the graph architecture. Our study bridges human connectomes with deep learning techniques and provides new avenues to study the underlying neural substrates of human cognition at scale.

scholarly journals Functional Annotation of Human Cognitive States using Deep Graph Convolution

2020 ◽  
Author(s):  
Yu Zhang ◽  
Loïc Tetrel ◽  
Bertrand Thirion ◽  
Pierre Bellec
Keyword(s):  
Human Brain ◽  
Brain Activity ◽  
Brain Regions ◽  
Test Accuracy ◽  
Experimental Conditions ◽  
Cognitive Domains ◽  
Brain Decoding ◽  
Cognitive States ◽  
Human Connectome Project ◽  
Fmri Response

AbstractA key goal in neuroscience is to understand brain mechanisms of cognitive functions. An emerging approach is “brain decoding”, which consists of inferring a set of experimental conditions performed by a participant, using pattern classification of brain activity. Few works so far have attempted to train a brain decoding model that would generalize across many different cognitive tasks drawn from multiple cognitive domains. To tackle this problem, we proposed a domain-general brain decoder that automatically learns the spatiotemporal dynamics of brain response within a short time window using a deep learning approach. By leveraging our prior knowledge on network organization of human brain cognition, we constructed deep graph convolutional neural networks to annotate cognitive states by first mapping the task-evoked fMRI response onto a brain graph, propagating brain dynamics among interconnected brain regions and functional networks, and generating state-specific representations of recorded brain activity. We evaluated the decoding model on a large population of 1200 participants, under 21 different experimental conditions spanning 6 different cognitive domains, acquired from the Human Connectome Project task-fMRI database. Using a 10s window of fMRI response, the 21 cognitive states were identified with a test accuracy of 89% (chance level 4.8%). Performance remained good when using a 6s window (82%). It was even feasible to decode cognitive states from a single fMRI volume (720ms), with the performance following the shape of the hemodynamic response. Moreover, a saliency map analysis demonstrated that the high decoding performance was driven by the response of biologically meaningful brain regions. Together, we provide an automated tool to annotate human brain activity with fine temporal resolution and fine cognitive granularity. Our model shows potential applications as a reference model for domain adaptation, possibly making contributions in a variety of domains, including neurological and psychiatric disorders.

scholarly journals Spatial Patterns for Discriminative Estimation

2019 ◽  
Author(s):  
Alberto Llera ◽  
Roselyne Chauvin ◽  
Peter Mulders ◽  
Jilly Naaijen ◽  
Maarten Mennes ◽  
...  
Keyword(s):  
Spatial Patterns ◽  
Brain Connectivity ◽  
Covariance Structure ◽  
Brain Regions ◽  
Linear Filters ◽  
Cognitive Changes ◽  
Experimental Conditions ◽  
Cognitive States ◽  
Human Connectome Project ◽  
Perfect Discrimination

AbstractFunctional connectivity between brain regions is modulated by cognitive states or experimental conditions. A multivariate methodology that can capture fMRI connectivity maps in light of different experimental conditions would be of primary importance to learn about the specific roles of the different brain areas involved in the observed connectivity variations. Here we detail, adapt, optimize and evaluate a supervised dimensionality reduction model to fMRI timeseries. We demonstrate the strength of such an approach for fMRI data using data from the Human Connectome Project to show that the model provides close to perfect discrimination between different fMRI tasks at low dimensionality. The straightforward interpretability and relevance of the model results is demonstrated by the obtained linear filters relating to anatomical areas well known to be involved on each considered task, and its robustness by testing discriminatory generalization and spatial reproducibility with respect to the number of subjects and fMRI time-points acquired. We additionally suggest how such approach can provide a complementary view to traditional task fMRI analyses by looking at changes in the covariance structure as a substitute to changes in the mean signal. We conclude that the presented methodology provides a robust tool to investigate brain connectivity alterations across induced cognitive changes and has the potential to be used in pathological or pharmacological cohort studies. A publicly available toolbox is provided to facilitate the end use and further development of this methodology to extract Spatial Patterns for Discriminative Estimation (SP♠DE).

scholarly journals Inter-subject phase synchronization and the dynamics of human cognition

2017 ◽  
Author(s):  
Taylor Bolt ◽  
Jason S. Nomi ◽  
Shruti G. Vij ◽  
Catie Chang ◽  
Lucina Q. Uddin
Keyword(s):  
Cognitive Processes ◽  
Phase Synchronization ◽  
Brain Activity ◽  
Methodological Approach ◽  
Brain Networks ◽  
Human Cognition ◽  
Bold Signal ◽  
Top Down ◽  
Whole Brain ◽  
Human Connectome Project

AbstractMassive whole-brain blood-oxygen-level dependent (BOLD) signal modulation (up to 95% of brain voxels) in response to task stimuli has recently been reported in functional MRI investigations. These findings have two implications. First, they highlight inability of a conventional ‘top-down’ general linear model approach to capture all forms of task-driven brain activity. Second, as opposed to a static ‘active’ or ‘non-active’ localization theory of the neural implementation of cognitive processes, functional neuroimaging should develop and pursue dynamical theories of cognition involving the dynamic interactions of all brain networks, in line with psychological constructionist theories of cognition. In this study, we describe a novel exploratory, bottom-up approach that directly estimates task-driven brain activity regardless of whether it follows an a priori reference function. Leveraging the property that task-driven brain activity is associated with reductions in BOLD signal variability, we combine the tools of instantaneous phase synchronization and independent component analysis to characterize whole-brain task-driven activity in terms of group-wise similarity in temporal signal dynamics of brain networks. We applied this novel framework to task fMRI data from a motor, theory of mind and working memory task provided through the Human Connectome Project. We discovered a large number of brain networks that dynamically synchronized to various features of the task scan, some overlapping with areas identified as ‘active’ in the top-down GLM approach. Using the results provided through this novel approach, we provide a more comprehensive description of cognitive processes whereby task-related brain activity is not restricted to dichotomous ‘active’ or ‘non-active’ inferences, but is characterized by the temporal dynamics of brain networks across time.Significance StatementThis study describes the results of a novel exploratory methodological approach that allows for direct estimation of task-driven brain activity in terms of group-wise similarity in temporal signal dynamics, as opposed to the conventional approach of identifying task-driven brain activity with a hypothesized temporal pattern. This approach applied to three different task paradigms yielded novel insights into the brain activity associated with these tasks in terms of time-varying, low-frequency dynamics of replicable synchronization networks. We suggest that this exploratory methodological approach provides a framework in which the complexity and dynamics of the neural mechanisms underlying cognitive processes can be captured more comprehensively.

Inefficient Encoding as an Explanation for Age-Related Deficits in Recollection-Based Processing

2014 ◽  
Vol 28 (3) ◽  
pp. 148-161 ◽  
Author(s):  
David Friedman ◽  
Ray Johnson
Keyword(s):  
Older Adults ◽  
Young Adults ◽  
Cognitive Processes ◽  
Brain Activity ◽  
Memory Performance ◽  
Brain Regions ◽  
Semantic Retrieval ◽  
Age Related ◽  
The Face ◽  
Episodic Memories

A cardinal feature of aging is a decline in episodic memory (EM). Nevertheless, there is evidence that some older adults may be able to “compensate” for failures in recollection-based processing by recruiting brain regions and cognitive processes not normally recruited by the young. We review the evidence suggesting that age-related declines in EM performance and recollection-related brain activity (left-parietal EM effect; LPEM) are due to altered processing at encoding. We describe results from our laboratory on differences in encoding- and retrieval-related activity between young and older adults. We then show that, relative to the young, in older adults brain activity at encoding is reduced over a brain region believed to be crucial for successful semantic elaboration in a 400–1,400-ms interval (left inferior prefrontal cortex, LIPFC; Johnson, Nessler, & Friedman, 2013 ; Nessler, Friedman, Johnson, & Bersick, 2007 ; Nessler, Johnson, Bersick, & Friedman, 2006 ). This reduced brain activity is associated with diminished subsequent recognition-memory performance and the LPEM at retrieval. We provide evidence for this premise by demonstrating that disrupting encoding-related processes during this 400–1,400-ms interval in young adults affords causal support for the hypothesis that the reduction over LIPFC during encoding produces the hallmarks of an age-related EM deficit: normal semantic retrieval at encoding, reduced subsequent episodic recognition accuracy, free recall, and the LPEM. Finally, we show that the reduced LPEM in young adults is associated with “additional” brain activity over similar brain areas as those activated when older adults show deficient retrieval. Hence, rather than supporting the compensation hypothesis, these data are more consistent with the scaffolding hypothesis, in which the recruitment of additional cognitive processes is an adaptive response across the life span in the face of momentary increases in task demand due to poorly-encoded episodic memories.

scholarly journals Decoding brain activity using a large-scale probabilistic functional-anatomical atlas of human cognition

2016 ◽  
Author(s):  
Timothy N. Rubin ◽  
Oluwasanmi Koyejo ◽  
Krzysztof J. Gorgolewski ◽  
Michael N. Jones ◽  
Russell A. Poldrack ◽  
...  
Keyword(s):  
Large Scale ◽  
Latent Dirichlet Allocation ◽  
Topic Model ◽  
Brain Activity ◽  
Human Cognition ◽  
Brain Images ◽  
Whole Brain ◽  
Context Sensitive ◽  
Cognitive States ◽  
Small Set

AbstractA central goal of cognitive neuroscience is to decode human brain activity--i.e., to infer mental processes from observed patterns of whole-brain activation. Previous decoding efforts have focused on classifying brain activity into a small set of discrete cognitive states. To attain maximal utility, a decoding framework must be open-ended, systematic, and context-sensitive--i.e., capable of interpreting numerous brain states, presented in arbitrary combinations, in light of prior information. Here we take steps towards this objective by introducing a Bayesian decoding framework based on a novel topic model---Generalized Correspondence Latent Dirichlet Allocation---that learns latent topics from a database of over 11,000 published fMRI studies. The model produces highly interpretable, spatially-circumscribed topics that enable flexible decoding of whole-brain images. Importantly, the Bayesian nature of the model allows one to “seed” decoder priors with arbitrary images and text--enabling researchers, for the first time, to generative quantitative, context-sensitive interpretations of whole-brain patterns of brain activity.

scholarly journals Mind wandering and attention during focused meditation: A fine-grained temporal analysis of fluctuating cognitive states

2017 ◽  
Author(s):  
Wendy Hasenkamp ◽  
Christine Wilson-Mendenhall ◽  
Erica Duncan ◽  
Lawrence Barsalou
Keyword(s):  
Cognitive Processes ◽  
Sustained Attention ◽  
Temporal Analysis ◽  
Mind Wandering ◽  
Brain Regions ◽  
Attentional Processing ◽  
Default Mode ◽  
Cognitive Correlates ◽  
Cognitive States ◽  
Task Positive Network

Studies have suggested that the default mode network is active during mind wandering, which is often experienced intermittently during sustained attention tasks. Conversely, an anticorrelated task-positive network is thought to subserve various forms of attentional processing. Understanding how these two systems work together is central for understanding many forms of optimal and sub-optimal task performance. Here we present a basic model of naturalistic cognitive fluctuations between mind wandering and attentional states derived from the practice of focused attention meditation. This model proposes four intervals in a cognitive cycle: mind wandering, awareness of mind wandering, shifting of attention, and sustained attention. People who train in this style of meditation cultivate their abilities to monitor cognitive processes related to attention and distraction, making them well suited to report on these mental events. Fourteen meditation practitioners performed breath-focused meditation while undergoing fMRI scanning. When participants realized their mind had wandered, they pressed a button and returned their focus to the breath. The four intervals above were then constructed around these button presses. We hypothesized that periods of mind wandering would be associated with default mode activity, whereas cognitive processes engaged during awareness of mind wandering, shifting of attention and sustained attention would engage attentional subnetworks. Analyses revealed activity in brain regions associated with the default mode during mind wandering, and in salience network regions during awareness of mind wandering. Elements of the executive network were active during shifting and sustained attention. Furthermore, activations during these cognitive phases were modulated by lifetime meditation experience. These findings support and extend theories about cognitive correlates of distributed brain networks.

Neurocognitive Aging and Functional Connectivity Using Functional Magnetic Resonance Imaging

2018 ◽  
Author(s):  
Hana Burianová
Keyword(s):  
Functional Connectivity ◽  
Cognitive Processes ◽  
Large Scale ◽  
Healthy Aging ◽  
Brain Activity ◽  
Brain Regions ◽  
Effective Strategies ◽  
Cognitive Improvement ◽  
Age Related ◽  
Neurocognitive Aging

Determining the mechanisms that underlie neurocognitive aging, such as compensation or dedifferentiation, and facilitating the development of effective strategies for cognitive improvement is essential due to the steadily rising aging population. One approach to study the characteristics of healthy aging comprises the assessment of functional connectivity, delineating markers of age-related neurocognitive plasticity. Functional connectivity paradigms characterize complex one-to-many (or many-to-many) structure–function relations, as higher-level cognitive processes are mediated by the interaction among a number of functionally related neural areas rather than localized to discrete brain regions. Task-related or resting-state interregional correlations of brain activity have been used as reliable indices of functional connectivity, delineating age-related alterations in a number of large-scale brain networks, which subserve attention, working memory, episodic retrieval, and task-switching. Together with behavioral and regional activation studies, connectivity studies and modeling approaches have contributed to our understanding of the mechanisms of age-related reorganization of distributed functional networks; specifically, reduced neural specificity (dedifferentiation) and associated impairment in inhibitory control and compensatory neural recruitment.

scholarly journals Data-driven models reveal the organization of diverse cognitive functions in the brain

2019 ◽  
Author(s):  
Tomoya Nakai ◽  
Shinji Nishimoto
Keyword(s):  
Cognitive Processes ◽  
Brain Activity ◽  
Quantitative Model ◽  
Task Type ◽  
Hierarchical Organization ◽  
Human Cognition ◽  
Cognitive Tasks ◽  
Cortical Organization ◽  
Brain Functions ◽  
Complete Representations

AbstractOur daily life is realized by the complex orchestrations of diverse brain functions including perception, decision, and action. One of the central issues in cognitive neuroscience is to reveal the complete representations underlying such diverse functions. Recent studies have revealed representations of natural perceptual experiences using encoding models1–5. However, there has been little attempt to build a quantitative model describing the cortical organization of multiple active, cognitive processes. Here, we measured brain activity using functional MRI while subjects performed over 100 cognitive tasks, and examined cortical representations with two voxel-wise encoding models6. A sparse task-type encoding model revealed a hierarchical organization of cognitive tasks, their representation in cognitive space, and their mapping onto the cortex. A cognitive factor encoding model utilizing continuous intermediate features by using metadata-based inferences7 predicted brain activation patterns for more than 80 % of the cerebral cortex and decoded more than 95 % of tasks, even under novel task conditions. This study demonstrates the usability of quantitative models of natural cognitive processes and provides a framework for the comprehensive cortical organization of human cognition.

scholarly journals A-GHN: Attention-based Fusion of Multiple GraphHeat Networks for Structural to Functional Brain Mapping

2021 ◽  
Author(s):  
SUBBA REDDY OOTA ◽  
Archi Yadav ◽  
Arpita Dash ◽  
Surampudi Bapi Raju ◽  
Avinash Sharma
Keyword(s):  
Deep Learning ◽  
Functional Connectivity ◽  
Complex Dynamics ◽  
Brain Activity ◽  
Structural Connectivity ◽  
Reaction Diffusion ◽  
Complex Nature ◽  
Relational Structures ◽  
Human Connectome Project ◽  
The Brain

Over the last decade, there has been growing interest in learning the mapping from structural connectivity (SC) to functional connectivity (FC) of the brain. The spontaneous fluctuations of the brain activity during the resting-state as captured by functional MRI (rsfMRI) contain rich non-stationary dynamics over a relatively fixed structural connectome. Among the modeling approaches, graph diffusion-based methods with single and multiple diffusion kernels approximating static or dynamic functional connectivity have shown promise in predicting the FC given the SC. However, these methods are computationally expensive, not scalable, and fail to capture the complex dynamics underlying the whole process. Recently, deep learning methods such as GraphHeat networks along with graph diffusion have been shown to handle complex relational structures while preserving global information. In this paper, we propose a novel attention-based fusion of multiple GraphHeat networks (A-GHN) for mapping SC-FC. A-GHN enables us to model multiple heat kernel diffusion over the brain graph for approximating the complex Reaction Diffusion phenomenon. We argue that the proposed deep learning method overcomes the scalability and computational inefficiency issues but can still learn the SC-FC mapping successfully. Training and testing were done using the rsfMRI data of 100 participants from the human connectome project (HCP), and the results establish the viability of the proposed model. Furthermore, experiments demonstrate that A-GHN outperforms the existing methods in learning the complex nature of human brain function.

scholarly journals Linking Models with Brain Measures

2020 ◽  
Author(s):  
Bradley C. Love
Keyword(s):  
Information Flow ◽  
Cognitive Processes ◽  
Brain Activity ◽  
Cognitive Model ◽  
Brain Regions ◽  
Learning Models ◽  
Model Based ◽  
Theoretical Significance ◽  
Brain Data ◽  
The Brain

Linking models and brain measures offers a number of advantages over standard analyses. Models that have been evaluated on previous datasets can provide theoretical constraints and assist in integrating findings across studies. Model-based analyses can be more sensitive and allow for evaluation of hypotheses that would not otherwise be addressable. For example, a cognitive model that is informed from several behavioural studies could be used to examine how multiple cognitive processes unfold across time in the brain. Models can be linked to brain measures in a number of ways. The information flow and constraints can be from model to brain, brain to model, or reciprocal. Likewise, the linkage from model and brain can be univariate or multivariate, as in studies that relate patterns of brain activity with model states. Models have multiple aspects that can be related to different facets of brain activity. This is well illustrated by deep learning models that have multiple layers or representations that can be aligned with different brain regions. Model-based approaches offer a lens on brain data that is complementary to popular multivariate decoding and representational similarity analysis approaches. Indeed, these approaches can realise greater theoretical significance when situated within a model-based approach.

Sign in / Sign up

Export Citation Format

Share Document