R. Benyishai et al., TRAVELING WAVES AND THE PROCESSING OF WEAKLY TUNED INPUTS IN A CORTICAL NETWORK MODULE, Journal of computational neuroscience, 4(1), 1997, pp. 57-77
Recent studies have shown that local cortical feedback can have an imp
ortant effect on the response of neurons in primary visual cortex to t
he orientation of visual stimuli. In this work, we study the role of t
he cortical feedback in shaping the spatiotemporal patterns of activit
y in cortex. Two questions are addressed: one, what are the limitation
s on the ability of cortical neurons to lock their activity to rotatin
g oriented stimuli within a single receptive field? Two, can the local
architecture of visual cortex lead to the generation of spontaneous t
raveling pulses of activity? We study these issues analytically by a p
opulation-dynamic model of a hypercolumn in visual cortex. The order p
arameter that describes the macroscopic behavior of the network is the
time-dependent population vector of the network. We first study the n
etwork dynamics under the influence of a weakly tuned input that slowl
y rotates within the receptive field. We show that if the cortical int
eractions have strong spatial modulation, the network generates a shar
ply tuned activity profile that propagates across the hypercolumn in a
path that is completely locked to the stimulus rotation. The resultan
t rotating population vector maintains a constant angular lag relative
to the stimulus, the magnitude of which grows with the stimulus rotat
ion frequency. Beyond a critical frequency the population Vector does
not lock to the stimulus but executes a quasi-periodic motion with an
average frequency that is smaller than that of the stimulus. In the se
cond part we consider the stable intrinsic state of the cortex under t
he influence of isotropic stimulation. We show that if the local inhib
itory feedback is sufficiently strong, the network does not settle int
o a stationary state but develops spontaneous traveling pulses of acti
vity. Unlike recent models of wave propagation in cortical networks, t
he connectivity pattern in our model is spatially symmetric, hence the
direction of propagation of these waves is arbitrary. The interaction
of these waves with an external-oriented stimulus is studied. It is s
hown that the system can lock to a weakly tuned rotating stimulus if t
he stimulus frequency is close to the frequency of the intrinsic wave.