Lenfesty, B. | Search bioRxiv

bioRxiv2026

Decision Confidence Neuron in Echo State Network for Continual Evaluation of EEG Motor Imagery Classification Quality

Echo state networks (ESNs) are efficient, neuro-inspired computational frameworks well suited to time-series data. However, ESN decision confidence is typically quantified in limited ways. We propose an explicit decision-confidence readout neuron, trained from decision readout outputs, to continuously monitor confidence as decisions form. In a simulated decision task, confidence activity increased with stimulus strength, linking greater discriminability to higher confidence. We then evaluated the model on EEG-based motor imagery classification, showing that confidence activity increased with decision accuracy and discriminated correct from error decisions, particularly in higher-performing participants, reflecting human-like metacognition. Overall, this approach enables continual monitoring of decision confidence, supporting more trustworthy ESN decisions, particularly in biomedical applications.

neuroscience↗

bioRxiv2025

A canonical gated neural circuit model for flexible perceptual decisions

Flexible perceptual decision-making requires rapid, context-dependent adjustments, yet the neural circuit mechanisms underlying its parsimonious representations remain unclear. Here, we propose a minimal mean-field neural circuit model that integrates sensory evidence and selects actions via distributed neuronal encoding, guided by data from a task that dissociates perceptual choice from motor response - abstract perceptual decision-making. The models nonlinear gating of action selective (AS) neurons replicates parietal cortical activity observed during task performance. Critically, recurrent excitation within the evidence integration (EI) population supports sensory evidence accumulation, working memory for sequential sampling, and reward rate optimisation. Moreover, the dynamics of EI and AS neuronal activities in the same model respectively mirror parietal neuronal activities related to sensory evidence encoding and ramping-to-threshold firing in a separate reaction-time task, while suggesting that decision readout engages both neuronal populations. The model also predicts decision interference in a novel two-stage decision version of the task, accounting for choice accuracy decrements observed in other experiments while predicting slower decisions. Together, these findings propose a minimal mean-field circuit-level mechanism unifying perceptual, memory-based, and abstract decision-making.

neuroscience↗