Bistable, Irregular Firing and Population Oscillations in a Modular Attractor Memory Network

Top Cited Papers
Open Access
Publisher Website
Google Scholar
Abstract
Attractor neural networks are thought to underlie working memory functions in the cerebral cortex. Several such models have been proposed that successfully reproduce firing properties of neurons recorded from monkeys performing working memory tasks. However, the regular temporal structure of spike trains in these models is often incompatible with experimental data. Here, we show that the in vivo observations of bistable activity with irregular firing at the single cell level can be achieved in a large-scale network model with a modular structure in terms of several connected hypercolumns. Despite high irregularity of individual spike trains, the model shows population oscillations in the beta and gamma band in ground and active states, respectively. Irregular firing typically emerges in a high-conductance regime of balanced excitation and inhibition. Population oscillations can produce such a regime, but in previous models only a non-coding ground state was oscillatory. Due to the modular structure of our network, the oscillatory and irregular firing was maintained also in the active state without fine-tuning. Our model provides a novel mechanistic view of how irregular firing emerges in cortical populations as they go from beta to gamma oscillations during memory retrieval. The basic computational principles of the brain are still unknown, and one major reason for this is related to the difficulties in simultaneously measuring detailed data from a sufficiently large number of cells. In techniques where populations of cells are monitored, resolution is low. Computational models have no such measurement limitations and can be constrained by several experiments at different levels of granularity, enabling testing of the biological plausibility of different computational theories. One such theory, the attractor network paradigm, has gained increasing support over the past twenty years by, for instance, comparing the output of attractor memory models to population data and spike frequency modulations of neocortical neurons. We take this comparison further by also looking at the fine-structure of activity in a network model with a novel modular structure also seen in vivo. This allows the network to operate in a new dynamic regime. In particular, we reproduce the irregular low-rate spiking of single cells in vivo, which has previously been a challenge for attractor network models. Oscillations in field potentials at gamma and beta frequencies, again believed to be connected to, or even essential for, attention and consciousness, emerge as a feature of the underlying dynamics of the model.