Mean-field models of propagating waves in visual cortex

Alain Destexhe, Matteo di Volo, Sandrine Chemla, Laurent Perrinet,
Frederic Chavane

UNIC, CNRS, Gif sur Yvette, France;
INT, CNRS, Marseille, France.

Propagating waves are widely seen in the nervous system, and their
presence was disputed in primary visual cortex (V1) of the monkey.
Using a combination of voltage-sensitive dye (VSD) imaging in awake
monkey V1 and model-based analysis, we showed previously that
virtually every visual input is followed by a propagating wave
(Muller et al., Nat Comm 2014). The wave was confined within V1,
and was consistent and repeatable for a given input. More recently,
we showed that two propagating waves interact in a suppressive
fashion, and are always sublinear. Here, we show that a mean-field
model can reproduce such features based on two mechanisms. First
mean-field models must include the duality of cell properties,
regular-spiking and fast-spiking for excitatory and inhibitory
cells, respectively. It is necessary that inhibitory cells have a
higher gain than excitatory cells, and networks that do not display
this duality show negligible suppression. Second, the
conductance-based shunting effect of the two waves onto one another
is important, as the suppression is almost abolished in
current-based models. This suggests that the suppressive effect is
a general feature of propagating wave activity, in agreement with
the general suppressive effect seen in other circumstances (Bair et
al., J Neurosci 2003; Reynaud et al., J Neurosci 2012).