"The holistic nature of perceptual judgments"
Alan Stocker
University of Pennsylvania
Abstract: “What’s the orientation of this line?” — Perceptual judgments of human observers are commonly viewed as reflecting the outcome of an inference process of the stimulus feature (e.g. orientation). In this talk I show recent results from my laboratory that challenge this widely established reductionist view. I propose that perceptual judgments are rather the result of a holistic inference process that operates not only at the feature but across all levels of a representational hierarchy. We tested this hypothesis in the context of a commonly used psychophysical matching task in which subjects are asked to report their perceived visual orientation of a test stimulus by adjusting a probe stimulus (method-of-adjustment). We introduce a holistic matching model that assumes that subjects' reports reflect an optimal match between the test and probe stimulus, both in terms of their inferred feature (orientation) but also their higher-level representation (orientation category). Validation of our model against five existing datasets demonstrates that the model accurately and comprehensively predicts subjects' response behavior, and outperforms previous models both quantitatively and qualitatively. Our results suggest that categorical effects in perceptual judgments are ubiquitous and can be parsimoniously explained as optimal behavior based on holistic sensory representations. These findings have substantial implications for our understanding of the neural information pathways and mechanisms underlying perceptual judgments.
Bio: Alan Stocker earned a PhD in Physics in 2002 at the Swiss Federal Institute of Technology (ETH). After postdoctoral training at NYU he took on a faculty job at the University of Pennsylvania in 2009, where he is currently an Associate Professor in Psychology. His research interests are in understanding the transformations that match sensory information onto hierarchical representations in the brain, and the computations by which these representations determine perceptually guided behavior. A particular focus of his research is on how contextual and statistical regularities in the environment shape these sensory representations and computations, as well as on the constraints the naturally limited metabolic and structural resources of the brain impose on them.