Árni Kristjánsson, University of Iceland
Rapid, detailed learning of feature distributions
Features such as color or orientation are rarely uniform in the envrionment, yet, little is known about how their distibutions are represented by the visual system. I will introduce a new method for studying feature ensembles based on intertrial learning in visual search. Observers looked for the oddly colored, or oddly oriented item among diamonds taken from either uniform or Gaussian distributions. On test trials the targets had various distances in feature space from the mean of the preceding distractor distribution. Targets on test trials therefore served as probes into probabilistic representations of distractor features. Test-trial response times revealed a striking similarity between the physical distribution of colors and orientations and their internal representations. The results demonstrate that the visual system represents color and orientation ensembles in a far more detailed way than previously thought, coding not only mean and variance but most surprisingly, the actual shape (uniform, Gaussian, bimodal or skewed) of feature distributions in the environment.
Overall, the results show that representations of distributions can be remarkably detailed.
Additional Information
Colloquium Talk of the GRK "Sensory Computation in Neural Systems"
Organized by
Zampeta Kalogeropoulou / Martin Rolfs