Autonomous Decision-Making Machine Vision System Enabled by a Low-Voltage, Photoadaptive Organic Synaptic Transistor.

Abstract

Artificial visual systems often require external circuits to detect intensity changes and switch biases, which limits integration and efficiency. Here, we report an organic synaptic transistor that executes autonomous decision-making solely on the basis of input light intensity, eliminating the need for external closed-loop feedback. Operating across the visible and near-infrared regions, the device exhibits bidirectional plasticity governed by light intensity, where weak light enhances channel conductance, whereas strong light suppresses it. This behavior originates from the competitive dynamics between photocarrier accumulation and trap-assisted recombination. The device therefore forms a closed loop of self-perception, self-decision, and self-modulation that emulates human visual adaptation. Crucially, the decision threshold is tunable via the PVA concentration, gate voltage, and excitation wavelength, enabling versatile in-sensor calibration. By emulating human visual adaptation through a closed loop of self-perception and self-modulation, this work paves the way for compact, energy-efficient, and autonomous neuromorphic vision systems.