Abstract:Mechanisms of top-down modulation in sensory perception and their relation to underlying connectivity are not completely understood. We present here a biophysically-detailed computational model of two interconnected cortical areas, representing the first steps in a cortical processing hierarchy, as a tool for potential discovery. The model integrates a large body of data from rodent primary somatosensory cortex and reproduces biological features across multiple scales: from a handful of ion channels defining a diversity of electrical types in hundreds of thousands of morphologically detailed neurons, to local and long-range networks mediated by hundreds of millions of synapses. Notably, long-range connectivity in the model incorporates target lamination patterns associated with feed-forward and feedback pathways. We use the model to study the impact of inter-areal interactions on sensory processing. First, we exhibit a cortico-cortical loop between the two model areas (X and Y), wherein sensory input to area X produces a response with two components in time, the first driven by the stimulus and the second by feedback from area Y. We perform a structural and functional characterization of this loop, finding a differential impact of layer-specific pathways in the feed-forward and feedback directions. Second, we explore stimulus discrimination by presenting four different spatially-segregate stimulus patterns. We observe well-defined temporal sequences of functional cell assembly activation, with stimulus specificity in early but not late assemblies in area X, i.e., in the stimulus-driven component of the response but not in the feedback-driven component. We also find the earliest assembly in area Y to be specific to pairs of patterns, consistent with the topography of connections. Finally, we examine the integration of bottom-up and top-down signals. When presenting a second stimulus coincident with the feedback-driven component, we observe an approximate linear superposition of responses. We find the implied lack of interaction consistent with the naive connectivity in the model and the absence of plasticity mechanisms that would underlie the learning of top-down influences. This work represents a first step in the study of inter-areal interactions with biophysically-detailed simulations.

Predicting proprioceptive cortical anatomy and neural coding with topographic autoencoders

Sequential sparse autoencoder for dynamic heading representation in ventral intraparietal area

Variational Autoencoder: An Unsupervised Model for Modeling and Decoding fMRI Activity in Visual Cortex

Modeling Category-Selective Cortical Regions with Topographic Variational Autoencoders

Body Topography Parcellates Human Sensory and Motor Cortex

Task-driven neural network models predict neural dynamics of proprioception

Multi-Modal Latent Variables for Cross-Individual Primary Visual Cortex Modeling and Analysis

A biologically inspired computational model of human ventral temporal cortex

Posteromedial cortical networks encode visuomotor prediction errors.

A GPU-accelerated cortical neural network model for visually guided robot navigation

The homunculus for proprioception: Toward learning the representation of a humanoid robot's joint space using self-organizing maps

Neural Geometry from Mixed Sensorimotor Selectivity for Predictive Sensorimotor Control

A biophysically-detailed model of inter-areal interactions in cortical sensory processing

Unraveling somatotopic organization in the human brain using machine learning and adaptive supervoxel-based parcellations

Vicarious Somatotopic Maps Tile Visual Cortex

A sensory-motor theory of the neocortex

Area 2 of primary somatosensory cortex encodes kinematics of the whole arm

Spatial Evolution of Information Dynamics in the Primary Motor Cortex During Reach

Automated construction of cognitive maps with visual predictive coding

pAE: An Efficient Autoencoder Architecture for Modeling the Lateral Geniculate Nucleus by Integrating Feedforward and Feedback Streams in Human Visual System

Automated mapping of virtual environments with visual predictive coding