The visual brain: Computing through multiscale complexity

Abstract

Information coding in sensory neurons is both digital, in terms of neuronal output spike timing and rate, and analog, produced by the irregular subthreshold changes in somatic and dendritic membrane potential resulting from synchronized volleys of synaptic inputs. Intracellular recordings give a unique access to a composite multiscale signal where the local microscopic integration process realized by a single neuron can be studied in the global mesoscopic context of the “unseen” units afferent to the recorded cell. This chapter shows how reverse engineering approaches can be used in the primary visual cortex of higher mammals to reveal the hidden complexity of visual processing and establish causal links between the functional dynamics of synaptic echoes in primary visual cortex and perceptual biases in low-level, non-attentive perception.