Purchase this article with an account.
Miconi Thomas, VanRullen Rufin; Gamma oscillations decompose the visual scene into object-based perceptual cycles: a computational model. Journal of Vision 2010;10(7):1270. doi: https://doi.org/10.1167/10.7.1270.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Several hypotheses exist regarding the functional role of oscillations in the brain. Among these is the well-known binding-by-synchrony hypothesis, suggesting that neurons coding for a specific object or concept tend to fire in synchrony (i.e. within the same cycle), creating a neural code for binding together features of a same object (Gray et al., 1989). Recently gamma oscillations have also been proposed as a competitive “gating” process, implementing a winners-take-all mechanism, allowing only the most excited neurons to fire at each cycle (Almeida, Idiart & Lisman 2009). Here we suggest that both mechanisms can be seen as two aspects of a single process, namely the decomposition of a visual scene into “perceptual cycles”, with neurons for each objects firing at successive cycles of the oscillation. We describe a simple model of V1 in which oscillations, binding by synchrony and scene decomposition all emerge automatically and spontaneously from the interactions between three simple, well-known mechanisms, namely feedback inhibition, refractory periods, and lateral (or feedback) connections implementing gestalt principles. Despite its extreme simplicity, the system gives rise to spontaneous decomposition of the scene into perceptual cycles, such that neurons encoding different objects tend to fire on different cycles. The system is applied both to artificial and natural images. We then show that these results persist when oscillations are exogenously imposed from a separate source (as opposed to endogenously generated, “gamma-like” oscillations). This suggests that the basic principle described here may extend to other oscillatory regimes besides gamma.
This PDF is available to Subscribers Only