Abstract
The concept of attractor neural networks (ANN) has led to biologically plausible models for neocortical working-associative memory, depending on excitatory feedback within modules of 103-105 neurons, asynchronous updating by units within discrete intervals approximating the spike refractory period and the stabilisation of attractors as different modular patterns of correlated burst firing which would persist for many seconds (of Hebbian reverberatory circuits). Recent studies of up to 11 units simultaneously, in nonspecific cortex of behaving monkeys, confirm traditional impressions of relatively low firing rates and apparently little bursting correlation in these conventionally ‘silent areas’ (in contrast to sensorimotor-specific neocortex) during a bimodal-sensory, complex localisation task. Considering the supposed cognitive functions of nonspecific cortex, the intracortical pathways involved, and not least the data from population-vector and oscillatory slow-wave studies in sensory-specific areas of the same and other species performing comparable tasks, this relative action-potential silence of associative areas is not surprising. Possibly some cognitive functions of neocortex, especially those partitioned for ‘entity and event binding’ rather than feature detection and fast locomotor response, may be mediated by quasiperiodic oscillations of membrane potential and their associated extracellular currents, which are below spike-bursting thresholds.