Multimodal Optoelectronic Synapses for Neuromorphic Intelligence: Coupling Light, Electrical, Thermal, and Beyond

ABSTRACT Intelligent systems operate in environments where multiple physical stimuli coexist and interact, yet most optoelectronic synaptic devices have primarily been limited to single‐modal operation. This review expands optoelectronic synapses beyond conventional optical‐electrical modulation toward a multisensory neuromorphic framework that incorporates thermal, mechanical, chemical, acoustic, and other physical stimuli. In multimodal optoelectronic synaptic systems, sensing, memory, and computation are coupled through conductance‐state evolution. External stimuli perturb internal variables such as photocarrier density, trap occupancy, ion distribution, interfacial polarization, adsorption‐induced charge, or phase state. The resulting changes in these internal states modulate conductance, producing volatile responses for transient sensory encoding or nonvolatile states for memory storage and weighted readout. We systematically review multimodal plasticity in optoelectronic synapses from both mechanism‐oriented and architecture‐based perspectives. Fundamental mechanisms are classified into five types: molecular structural transformation, filamentary switching, band structure engineering, field‐induced charge modulation, and carrier generation pathways. In this review, the reported multimodal optoelectronic neuromorphic devices are classified into five architectural types according to where transduction, memory, and computation occur, including single‐device integration, parallel multimodal processing, sensor‐coupled preprocessing, cross‐modal conversion, and stimulus‐dependent bidirectional plasticity. Representative applications in pattern recognition, adaptive learning, and in‐sensor/in‐memory computing are discussed as fundamental demonstrations toward realizing hardware‐level multisensory intelligence.