hide
Free keywords:
-
Abstract:
When a subject is dichoptically presented with two conflicting images, only one image is perceived at a time while the other is suppressed from awareness; this paradigm of multistable perception, is known as Binocular Rivalry (BR). Perception, therefore, alternates between the two visual patterns allowing a dissociation of sensory stimulation from conscious visual perception.
From a theoretical point of view, most of the computational models proposed to account for BR are rate-like models. The need is nevertheless apparent to employ biophysically
plausible neuronal network models in order to connect psychophysics experiments with neurophysiological data. Here we adopted a spiking network with biophysically realistic AMPA, NMDA, and GABA receptor-mediated synaptic dynamics, as well as spike-frequency adaptation mechanisms based on Ca++-activated K+ after-hyperpolarization currents.
Noise due to the probabilistic spike times of neurons is crucial for rivalry. It has been shown that competition models based on cross-inhibition and adaptation explain the observed alternations in perception when noise operates in balance with adaptation. In order to gain insights into
the cortical microcircuit dynamics mediating spontaneous perceptual alternations in BR, we derived a consistently reduced four-variable population rate model from a recurrent attractorbased biologically realistic spiking network used to model working memory, attention, and
decision-making, where neuronal adaptation is implemented, using mean-field techniques.
The model accounts for experimental data, collected from human subjects during BR, such as mean dominance duration, coefficient of variation, shape parameter of the gamma distribution of dominance durations and agrees with Levelt’s second and fourth proposition. The model replicates the observed data when it operates near the bifurcation that separates the noise-driven-transitions from the adaptation-driven-oscillations dynamical regime. Moreover, we show that spike-frequency adaptation of interneurons is not crucial for the spontaneous perceptual alternations, but affects the optimal parametric space of the system by decreasing the overall level of neuronal adaptation necessary for the bifurcation to occur and generates oscillations in resting state, such as in the absence of external stimuli.
Furthermore, we consider recent experimental data from the macaque lateral prefrontal cortex collected during Binocular Flash Suppression a paradigm of externally induced perceptual alternation. They show a decrease in correlated variability across pairs of neurons sharing similar stimulus preferences when their preferred stimulus is perceived during rivalrous visual stimulation compared to the magnitude of correlation when the same stimulus is perceived without competition. Employing the biophysically plausible spiking network with spike-frequency adaptation, we explore distinct possible computational strategies responsible for the noise correlation decrease under visual competition.