When:
2009-12-03 at 10:00
Where:
University Pierre&Marie Curie Paris VI, Building B, 1st floor, Room 120 (How to come)
Details:
TITLE
Expectations, Sensory Adaptation and the Origin of Perceptual Biases.
ABSTRACT
I will talk about 2 recent projects.
The first project (Series, Stocker and Simoncelli, Neural Computation, 2009) is a theoretical work investigating the relationship between the physiological effects of adaptation and their perceptual consequences. At the physiological level, adaptation is known to correspond to changes in neurons' properties -- it is usually found that the responsivity of neurons that are sensitive to the adaptor stimulus decrease. We are interested in understanding how the rest of the brain interpret these changes. Does the read-out adapt at the same time as the sensory neurons ? Or does it continue to use a fixed strategy, `unaware' of the adaptation changes? We modelled adaptation in a population of neurons, compared a variety of read-out schemes that are either fixed or adaptive, and studied their ability to explain estimation biases (e.g. after-effets) and discrimination performance in motion and contrast adaptation. We show that the psychophysical data can be easily explained in terms of a mismatch between the changing response properties and a fixed read-out scheme.
The second project (Chalk, Seitz and Series, submitted) is an
experimental investigation of the influence of expectations on the
perception of simple stimuli.
Expectations broadly influence our experience of the world. However,
the process by which expectations are acquired and then shape our
sensory experiences is not well understood.
Using a simple task involving estimation and detection of motion
random dots displays, we examined whether expectations can be
developed implicitly through in a fast statistical learning procedure.
We found that participants quickly and automatically developed
expectations for the most frequently presented directions of motion,
and that this strongly altered their perception of new motion
directions, inducing attractive biases in perceived direction as well
as visual hallucinations in the absence of a stimulus.
We show that in this case, the perceptual biases are consistent with
optimal Bayesian Inference and discuss their possible biological
substrate.