Unidirectional brain-computer interface: Artificial neural network encoding natural images to fMRI response in the visual cortex

TL;DR

This paper introduces VISION, an artificial neural network that predicts fMRI responses to natural images with high accuracy, aiding neuroscientific understanding of visual perception and reducing analysis costs.

Contribution

The study presents VISION, a multimodal neural network that outperforms existing models in predicting fMRI responses and offers tools for interpretability and hypothesis generation.

Findings

Achieved 45% higher accuracy than previous models in predicting fMRI responses.

Revealed representational biases across different visual cortical areas.

Provided an interpretable metric linking neural representations to cortical functions.

Abstract

While significant advancements in artificial intelligence (AI) have catalyzed progress across various domains, its full potential in understanding visual perception remains underexplored. We propose an artificial neural network dubbed VISION, an acronym for "Visual Interface System for Imaging Output of Neural activity," to mimic the human brain and show how it can foster neuroscientific inquiries. Using visual and contextual inputs, this multimodal model predicts the brain's functional magnetic resonance imaging (fMRI) scan response to natural images. VISION successfully predicts human hemodynamic responses as fMRI voxel values to visual inputs with an accuracy exceeding state-of-the-art performance by 45%. We further probe the trained networks to reveal representational biases in different visual areas, generate experimentally testable hypotheses, and formulate an interpretable metric to associate these hypotheses with cortical functions. With both a model and evaluation metric, the cost and time burdens associated with designing and implementing functional analysis on the visual cortex could be reduced. Our work suggests that the evolution of computational models may shed light on our fundamental understanding of the visual cortex and provide a viable approach toward reliable brain-machine interfaces.