Cheng, Y.-A. | Search bioRxiv

bioRxiv2025

Attention Improves Population Codes by Warping Neural Manifolds in Human Visual Cortex

Decades of research on visual attention have revealed its numerous effects on neural responses. Two competing models have been proposed for how these effects lead to improved population representations: one highlights changes in neural tuning, while the other points towards changes in trial-by-trial noise correlations. Here, we develop a neural population manifold framework that interprets changes in neural responses as geometric transformations in high-dimensional neural space, allowing us to disentangle and quantify the effects of tuning changes and correlation changes induced by attention. Applying this framework to extensive measurements of cortical responses during different attentional tasks, we find that tuning changes are the primary driver of improved population representations. In contrast, correlation changes, though present, have minimal--or even detrimental--effects to information content due to its strong interactions with other changes (e.g., tuning, variability). Our results support the "tuning change" model of visual attention and demonstrate a general framework for adjudicating how different aspects of neural coding affect information processing.

neuroscience↗

bioRxiv2023

Noise reduction as a unified mechanism of perceptual learning in humans, macaques, and convolutional neural networks

Visual perceptual learning (VPL), defined as long-term improvement in a visual task, is considered a crucial tool for elucidating underlying visual and brain plasticity. However, the identification of a unified theory of VPL has long been controversial. Multiple existing models have proposed diverse mechanisms, including improved signal-to-noise ratio, changes in tuning curves, and reduction of noise correlations, as major contributors to improved neural representations associated with VPL. However, each model only accounts for specific aspects of the empirical findings, and there exists no theory that can comprehensively explain all empirical results. Here, we argue that all neural changes at single units can be conceptualized as geometric transformations of population response manifolds in a high-dimensional neural space. This approach enables conflicting major models of VPL to be quantitatively tested and compared within a unified computational theory. Following this approach, we found that changes in tuning curves and noise correlations, as emphasized by previous models, make no significant contributions to improved population representations by visual training. Instead, we identified neural manifold shrinkage due to reduced trial-by-trial neural response variability, a previously unexplored factor, as the primary mechanism underlying improved population representations. Furthermore, we showed that manifold shrinkage successfully accounts for learning effects across various domains, including artificial neural responses in deep neural networks trained on typical VPL tasks, multivariate BOLD signals in humans, and multi-unit activities in monkeys. These converging results suggest that our neural geometry theory offers a quantitative and comprehensive approach to explain a wide range of empirical results and to reconcile previously conflicting models of VPL.

neuroscience↗