September 2011
Volume 11, Issue 11
Free
Vision Sciences Society Annual Meeting Abstract  |   September 2011
A cascade-correlation model of bistable perception
Author Affiliations
  • Caitlin Mouri
    McGill University, Canada
  • Avi Chaudhuri
    McGill University, Canada
Journal of Vision September 2011, Vol.11, 982. doi:10.1167/11.11.982
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Caitlin Mouri, Avi Chaudhuri; A cascade-correlation model of bistable perception. Journal of Vision 2011;11(11):982. doi: 10.1167/11.11.982.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

The phenomenon of bistability arises from physical ambiguities in the stimulus that lend themselves to two mutually exclusive interpretations. Many properties have been shown to affect switching rate, including stimulus interruptions (Kornmeier et al., 2007), attention (Meng & Tong, 2004), and eye movements (Ellis & Stark, 1978). However, relatively little research has addressed the role of individual differences. Previous studies have found that subjects fall into two groups: fast switchers and slow switchers (Borsellino et al., 1982). It has been suggested that these differences arise from variations in individual experience with the stimulus (Sakai et al., 1995). In this study, we use a sibling-descendant cascade-correlation neural network (Baluja & Fahlman, 1994; Shultz, 2004) to examine this hypothesis. We trained the network on a set of unambiguous stimuli, then tested it on an ambiguous stimulus, modeled after the Necker cube. Networks with extensive training showed high switching rates, while networks with shorter training regimes showed significantly lower switching rates. In addition, we found that strong positive feedback yielded lower switching rates, while weak positive feedback resulted in higher switching rates. Dynamical models support the latter result, where rivalry depends on a balance between positive self-feedback and mutually inhibitory connections between neural populations (Wilson, 1999). Our model suggests that switching rates may also depend on the underlying neural architecture, which in turn depends on early network training and experience.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×