Bio-Inspired Wide-Field Visual Neuron Implemented with Ultra-Low Information Loss Population Coding

Area of Science:

• Neuromorphic Engineering
• Materials Science
• Computational Neuroscience

Background:

• Modern neuromorphic systems struggle with emulating biological vision, facing challenges in wide-spectrum perception, distortion-free encoding, and population-level signal processing.
• Existing systems often lack the efficiency and adaptability of biological visual processing.
• Macaque visual neurons exhibit stochastic-resilient population coding, offering a model for advanced neuromorphic designs.

Purpose of the Study:

• To develop an advanced visual neuron prototype that overcomes current limitations in neuromorphic vision systems.
• To integrate broadband photodetection with biomimetic spike population encoding in a monolithic architecture.
• To improve visual perception range, signal fidelity, and processing efficiency in artificial vision hardware.

Main Methods:

• Development of a visual neuron prototype utilizing a photoelectric multi-stimulation field-effect transistor and a parallel threshold-switch architecture.
• Integration of a photosensitive MoSe2/MoS2 heterojunction within the field-effect transistor to extend spectral sensitivity.
• Implementation of a parallel threshold-switching design to enable cooperative population coding from single-unit encoding.

Main Results:

• The visual neuron prototype achieved broadband photodetection from 350-1000 nm, doubling the perception field.
• Photocurrent response increased by 1.36-fold under identical conditions.
• The parallel threshold-switching design reduced signal distortion by 82.1% and improved pattern recognition accuracy by 12.1% in a spiking neural network.
• Information processing time was maintained at the biological scale (<200 ms).

Conclusions:

• The developed visual neuron prototype successfully integrates van der Waals heterostructure photonics with macaque-derived neural population coding principles.
• This work establishes a transformative framework for bioinspired vision hardware, significantly enhancing neuromorphic system capabilities.
• The findings bridge the gap between advanced neuromorphic materials and cortical processing efficiency, paving the way for next-generation artificial vision.