Temporal binding at gamma frequencies in the brain: paving the way to epilepsy?

Fast (beta-gamma band 20-100 Hz) rhythms of electrical activity of the brain have been suggested to play an important role in perception, cognition and consciousness providing temporal binding of neural activities and allowing the formation of mental representations. The recent advances in the concept of temporal binding and their relation to the theory of neural networks (connectionism) are reviewed here as well as some experimental results concerning the intensified gamma rhythms and epilepsy. The hippocampal-neocortical gamma rhythms are extremely intense and hypersynchronous at onset of the epileptiform discharges induced by systemic kainic acid in the rat. Those gamma rhythms are followed by a slow rhythm of epileptiform spikes/sharp waves or spike-wave complexes ('spike-wave' activity). During spike-wave activity, gamma synchronisation is significantly decreased. A novel unifying concept is proposed which relates the associative principle of neural networks to the mechanism of temporal binding at high frequencies. It suggests that for each memory stored in an associative network there is a corresponding quasi-stable state of synchronous oscillation at some frequency within the gamma band. It also suggests that excessive temporal binding ("over-binding") occurs at seizure onset when abnormally intensified and globally synchronous fast activity is often observed. "Over-binding" may cause the undesirable formation of false associations due to inadequate synaptic modifications. To prevent this process, spike-wave discharge develops as an extreme activation of the mechanism capable to desynchronise and eventually suppress fast activity and erase the spurious modes of activity associated with hypersynchronous gamma rhythms. Thus, spike-wave activity is suggested to be the "anti-binding" mechanism. This mechanism is also related to the spikes/sharp waves normally occurring in the brain mostly in sleep. It is qualitatively similar to the "unlearning" mechanism of Crick and Mitchison presumably associated with the PGO spikes of the REM sleep.