Neural Heterogeneity Enhances Reliable Neural information Processing: Local Sensitivity and Globally Input-slaved Transient Dynamics

Abstract

Cortical neuronal activity varies over time and across repeated stimulation trials, yet consistently represents stimulus features. The dynamical mechanism underlying this reliable representation and computation remains elusive. This study uncovers a mechanism that achieves reliable neural information processing, leveraging a biologically plausible network model with neural heterogeneity. We first investigate neuronal timescale diversity in reliable computation, revealing it disrupts intrinsic coherent spatiotemporal patterns, enhances local sensitivity, and aligns neural network activity closely with inputs. This leads to local sensitivity and globally input-slaved transient dynamics, essential for reliable neural processing. Other neural heterogeneities, such as non-uniform input connections and spike threshold heterogeneity, plays similar roles, highlighting neural heterogeneitys role in shaping consistent stimulus representation. This mechanism offers a potentially general framework for understanding neural heterogeneity in reliable computation and informs the design of new reservoir computing models endowed with liquid wave reservoirs for neuromorphic computing. TeaserNeural diversity disrupts spatiotemporal patterns, aligning network activity with inputs for reliable information processing.