Gate-Programmable Spectral Plasticity in Nanoporous Vertical Organic Synaptic Transistors for Adaptive Neuromorphic Vision

Artificial vision systems capable of dynamically adapting to changing spectral environments are essential for next-generation intelligent perception but remain fundamentally constrained by the fixed spectral responsivity of conventional photodetectors. While narrowband photodetectors effectively suppress optical background interference, their spectral selectivity is generally predetermined by material composition or device geometry, preventing adaptive spectral perception within a single hardware platform. Here, we report a nanoporous vertical organic synaptic transistor based on a highly diluted P3HT:PC61BM (100:1) blend that enables gate-programmable spectral plasticity, allowing reversible electrical switching between wavelength-selective narrowband detection and broadband panchromatic perception. Unlike conventional planar phototransistors, the vertically configured transport architecture spatially decouples optical absorption from electrostatic channel modulation, thereby overcoming the long-standing incompatibility between the thick photoactive layers (1.5 μm) required for charge-collection narrowing and the ultrathin semiconductor channels necessary for efficient gate control. By electrically reshaping the interfacial energy landscape and carrier collection length, gate modulation dynamically regulates the extraction of photogenerated carriers across different penetration depths, producing adaptive spectral responses without altering either the material composition or device structure. Quantitatively, the device achieves a competitive R of 41.7 A/W, a shot-noise-limited D* of 2.7 × 1013 Jones, and an EQE of 8.3 × 103%. Beyond spectral reconfigurability, the thick active layer naturally introduces persistent photoconductivity arising from trap-mediated carrier relaxation, enabling biologically analogous excitatory postsynaptic current generation together with paired-pulse facilitation (reaching an index of 290%), spike-number-, spike-duration-, and spike-intensity-dependent plasticity, as well as programmable transitions from short-term to long-term plasticity. These multidimensional synaptic dynamics provide intrinsic temporal information processing directly at the sensor level, eliminating the need for separate sensing and memory units. Furthermore, experimentally acquired synaptic conductance characteristics are integrated into convolutional neural networks through a hardware-algorithm codesign strategy to emulate adaptive retinal signal preprocessing. System-level simulations demonstrate that electrically programmable spectral weighting significantly improves both broadband image recognition and monochromatic color discrimination by dynamically matching spectral perception to different visual tasks. This work establishes a general strategy for integrating adaptive spectral sensing, electrically programmable phototransistors, and neuromorphic information processing within a unified vertical organic transistor architecture, providing a versatile materials platform for intelligent optoelectronic systems.