Solid-State Synapses Modulated by Wavelength-Sensitive Temporal Correlations in Optic Sensory Inputs

Recently, inspired by neurobiological information processing, correlation-based learning has been expressed physically in nonbiological systems by exploiting the time causality of electric signals. Yet, the capability to learn from visual events requires extending these concepts to optical stimuli. Here we show a solid-state system that is sensitive to 100 ms-scale timing of pairs of light stimuli with complementary short/long visible wavelengths, causing asymmetric changes of photoconductance. This property endows optical signals with time causality, leading to wavelength-sensitive time correlations with time scales comparable with those of perceptual recognition. On the basis of these observations, we propose that complex information can be extracted from visual patterns imprinted as spatiotemporal modulations of persistent photoconductance. We suggest that this capability may stimulate neuromorphic hybrid electronic/photonic systems to construct biomimetic spatial memory and navigation maps inspired from neurobiology.