Background gamma rhythmicity and attention in cortical local circuits: a computational study

Abstract

We describe a simple computational model, based on generic features of cortical local circuits, that links cholinergic neuromodulation, gamma rhythmicity, and attentional selection. We propose that cholinergic modulation, by reducing adaptation currents in principal cells, induces a transition from asynchronous spontaneous activity to a "background" gamma rhythm (resembling the persistent gamma rhythms evoked in vitro by cholinergic agonists) in which individual principal cells participate infrequently and irregularly. We suggest that such rhythms accompany states of preparatory attention or vigilance and report simulations demonstrating that their presence can amplify stimulus-specific responses and enhance stimulus competition within a local circuit.

Fig. 1.

Adaptation currents disrupt weak PING. (a) Weak PING. (b) Adaptation currents suppress E-cells and reduce coherence of I-cells. (c) Loss of rhythmicity alone leads to suppression of the E-cells, even when there are no adaptation currents.

Fig. 2.

Weak PING facilitates response to specific input. (a) Specific excitation leads to specific strong PING. (b) Adaptation currents weaken rhythm and weaken response to specific excitation. (c) Loss of coherence, even in the absence of any adaptation currents, greatly weakens response to specific excitation. (d) In a single simulation, intervals of rhythmicity (arhythmicity) are intervals of strong (weak) response to specific excitation.

Fig. 3.

Synchrony facilitates stimulus competition. (a) When L (E-cells 121–140) is excited more strongly than D (E-cells 11–30), D is suppressed. (b) E → E synapses within assemblies improve suppression. (c) Overlapping ensembles: L (E-cells 121–140) indicated by solid lines, D (E-cells 107–126) indicated by dash–dots. The set of cells D \ L (cells belonging to D but not to L) is suppressed, but L \ D is partially suppressed as well. (d) Recurrent excitation within assemblies counteracts suppression of L \ D. (e) When asynchronous, nonrhythmic inhibition is used to suppress D, contrast is reduced.

Fig. 4.

Results for networks with only two E-cells, L and D. The intrinsic frequency of L is fixed at 75 Hz. The intrinsic frequency of D is varied (horizontal axis). In each case, inhibition is strong enough for D to be suppressed but not larger. The graphs show the resulting network frequencies of L in the cases of synchronous, rhythmic inhibition (solid) and asynchronous inhibition (dashes).