Cortical layer-specific imaging of context-dependent cognitive processing
Lead Research Organisation:
University of Glasgow
Abstract
A significant neuroscientific challenge lies in uncovering the neurophysiological processes and circuitry of cognitive processing. This objective becomes increasingly pressing as we gain a deeper understanding of mental health symptoms and witness the advancement of artificial intelligence seeking to emulate brain functions. Recent breakthroughs are converging towards a framework that elucidates and reframes the neural substrates of cognition and cognitive dysfunctions, based on context-specific cellular mechanisms. Context sensitivity is vital for coherent mental experiences, from accurately perceiving the sensory environment to adapting behaviour in social interactions. This framework describes how pyramidal neurons, the primary computational units of the cortex, have distinct basal and apical regions that process bottom-up (feedforward) and top-down (feedback) signals (different to the dominant integrate-and-fire concept). The integration of top-down inputs in the apical dendrites is pivotal for bestowing pyramidal neurons with context sensitivity, which forms the basis for fundamental cognitive processes and conscious awareness.
We recently led a consortium across Europe consisting of neuroscientists, computational modellers, and theorists in the Human Brain Project to study context-sensitive feedback in the mammalian brain. We deepened our understanding of contextual mechanisms for perceptual inference across species, in human, monkey and mouse visual cortex. Furthermore, our brain imaging, electrophysiological and calcium imaging results recorded across different labs in Europe contributed to refining theories of predictive processing. Predictive processing is the most influential theory of brain function today, providing a framework for how the brain interprets and acts in the world by making predictions and updating its models when errors occur. The exchange of predictions and prediction errors is central to healthy brain function, and it is often assumed these processes occur in separate neuron types. However, theories of context-sensitive cellular mechanisms compel us to advance to a sub-neuronal level, suggesting that errors and predictions are exchanged within a single neuron. Such mechanisms are challenging to study with non-invasive brain imaging, but understanding these mechanisms is essential for establishing baselines of healthy cognition.
Utilising our cutting-edge 7T fMRI brain imaging platform at Glasgow, and paradigms disentangling sensory from top-down processing, we capture sub-neuronal mechanisms of prediction error detection at different layers of cortical microcircuits. Our approach is characterised by two key empirical strategies:
(i) we design paradigms isolating internally generated contextual signals, by recording from retinotopic non-stimulated regions of primary visual cortex
(ii) we measure laminar fMRI in these non-stimulated regions to localise functional signals to cortical layers housing the apical dendritic compartment of pyramidal neurons receiving top-down inputs.
Specifically our objectives here are to reveal spatiotemporal interactions of predictive and sensory signals in visual microcircuits using state-of-the-art line scanning fMRI. We will also show how activity in laminar microcircuits relates to conscious experience. Lastly we will demonstrate how laminar microcircuit architectures are functionally modified by memory and learning.
Our integrated approach pushes boundaries at intersections between psychological processes and animal neuroscience, converging towards a new field of cellular cognitive neuroscience. With deeper interdisciplinary exchange, laminar fMRI is an ideal mesoscale bridge towards closing explanatory gaps between neuronal mechanisms and system-level modulations. Our findings are relevant for biological and artificial neurons, with context-dependent motifs from biological vision being incorporated into formal specifications of artificial computation. Investigating mesoscale circuit function will also be informative for computational psychiatry in cases where top-down contextual processing is dysfunctional, creating visual disturbances such as hallucinations.
We recently led a consortium across Europe consisting of neuroscientists, computational modellers, and theorists in the Human Brain Project to study context-sensitive feedback in the mammalian brain. We deepened our understanding of contextual mechanisms for perceptual inference across species, in human, monkey and mouse visual cortex. Furthermore, our brain imaging, electrophysiological and calcium imaging results recorded across different labs in Europe contributed to refining theories of predictive processing. Predictive processing is the most influential theory of brain function today, providing a framework for how the brain interprets and acts in the world by making predictions and updating its models when errors occur. The exchange of predictions and prediction errors is central to healthy brain function, and it is often assumed these processes occur in separate neuron types. However, theories of context-sensitive cellular mechanisms compel us to advance to a sub-neuronal level, suggesting that errors and predictions are exchanged within a single neuron. Such mechanisms are challenging to study with non-invasive brain imaging, but understanding these mechanisms is essential for establishing baselines of healthy cognition.
Utilising our cutting-edge 7T fMRI brain imaging platform at Glasgow, and paradigms disentangling sensory from top-down processing, we capture sub-neuronal mechanisms of prediction error detection at different layers of cortical microcircuits. Our approach is characterised by two key empirical strategies:
(i) we design paradigms isolating internally generated contextual signals, by recording from retinotopic non-stimulated regions of primary visual cortex
(ii) we measure laminar fMRI in these non-stimulated regions to localise functional signals to cortical layers housing the apical dendritic compartment of pyramidal neurons receiving top-down inputs.
Specifically our objectives here are to reveal spatiotemporal interactions of predictive and sensory signals in visual microcircuits using state-of-the-art line scanning fMRI. We will also show how activity in laminar microcircuits relates to conscious experience. Lastly we will demonstrate how laminar microcircuit architectures are functionally modified by memory and learning.
Our integrated approach pushes boundaries at intersections between psychological processes and animal neuroscience, converging towards a new field of cellular cognitive neuroscience. With deeper interdisciplinary exchange, laminar fMRI is an ideal mesoscale bridge towards closing explanatory gaps between neuronal mechanisms and system-level modulations. Our findings are relevant for biological and artificial neurons, with context-dependent motifs from biological vision being incorporated into formal specifications of artificial computation. Investigating mesoscale circuit function will also be informative for computational psychiatry in cases where top-down contextual processing is dysfunctional, creating visual disturbances such as hallucinations.