Global resistance to local perceptual adaptation in texture discrimination

Abstract

Intensive training or testing reduces performance on perceptual tasks. These effects are specific to basic image features, implicating early stages of the visual stream rather than general fatigue. Recent results show that such adaptation-like performance decrements are practically eliminated following practice with a small number of trials and sleep. This long-term learning effect suggests a link between perceptual deterioration and learning at the neuronal connectivity level: training strengthens task related connections, with further training leading to saturation of these connections along with strengthening of less efficient connections corresponding to accumulated noise in the network. Such saturation in network connectivity and reduction of signal-to-noise ratio consequently affects the readout of the network, causing deterioration in discrimination performance. Resistance to such deterioration is achieved by sleep-dependent consolidation of unsaturated connectivity resulting from short training. Here we show that such training-induced resistance to perceptual decrements generalizes across retinal locations, while suppressive effects due to extensive training were shown to be local. Furthermore, we show that these local suppressive effects are long-term, implying consolidation of these effects into what we term as an “adaptational state” in local visual networks. These experiments, revealing the different transfer properties of performance decrements and increments, allow us to identify local and global components of perceptual learning and their interactions, suggesting mechanisms that induce modifications of higher brain areas which interact with local early visual networks and enable improvement of perceptual abilities.