Flexible cortical gamma-band correlations suggest neural principles of visual processing

Eckhorn et al. review in their study two experiments in which they perform measurements in humans and monkeys while presenting visual stimuli. They apply a new measure of and thus view on synchrony to yield larger correlation. In addition the results are compared to similar measurements in neural network models confronted with comparable input. I want to consider, in which way the results of these experiments allow argumentation in favour of binding by neural synchrony. There are two major task in plausible binding-like information processing. For useful assembly coding, groups must not only be dynamically selected but also labelled in a way that subsequent processing stages can identify them as such. Neural synchrony does both in an intuitive way: The construction of groups of synchronously firing neurons can be imagined to be the consequence of similar behaviour following similar input (together with the right interaction). After that it is plausible that higher processing stages identify neurons belonging to the same assembly by simply comparing their behaviour over time. In the first experiment, Eckhorn et al. assess synchrony during a figureground segmentation task, asking: Do neurons in V1 representing together either figure or ground fire in a correlated way? As shown in preceding studies, measuring synchrony (by means of γ -coherence) would restrict binding to an area of a few millimetres on V1. To overcome this perceptoempirically implausible restriction Eckhorn et al. suggest γ-phase conti-