Expectation and Attention in Visual Cognition

Duke University, Durham NC

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): An accurate interpretation of the visual environment is crucial to our survival. However, visual perception is faced with two major constraints. First, computational limitations of the visual system render it impossible to simultaneously analyze all aspects of our surroundings with high fidelity. To meet this challenge, vision relies on mechanisms of attention, prioritizing the processing of those aspects of the environment deemed most relevant to our wellbeing. Second, visual information is inherently ambiguous, as many distinct objects can cast an identical retinal image, while a single object can cast many different images. The brain meets this challenge by exploiting statistical regularities in the environment to form perceptual expectations, providing context- sensitive guidance regarding the most probable visual data. Though expectation and attention are closely interwoven in everyday life, they have typically either been investigated in isolation or confounded with each other, such that their relation remains poorly understood. We here argue that the disentangling of the twin influences of attention (stimulus relevance) and expectation (stimulus probability) on perception is a key to major advances in our understanding of visual cognition, including the resolution of longstanding debates in the attention literature (does attention act in an additive o a multiplicative fashion? Does attention act early or late?). Moreover, this question is deeply relevant to clinical conditions in which attentive and predictive processes appear to be deficient, such as attention deficit hyperactivity disorder and schizophrenia. The overall goal of this projec is to determine the computational and neural mechanisms of expectation and attention in visual cognition. We break down this goal into three specific aims. Firstly, we will dissect the computational mechanisms by which expectation and attention modulate visual perception, by examining the timing (early vs. late) and nature (additive vs. multiplicative) of their respective influences on signal detection and visual neural responses. Secondly, we will determine whether and how attention and expectation interact in their modulation of visual processing. Finally, we will exploit the computational metrics developed in this work to lay the foundations for translational computational neuropsychiatry applications, by linking individual differences in attention and expectation model parameters in healthy subjects to variance in personality traits that constitute known risk-factors for clinical diseases whose etiology involves deficits in attentive and predictive processing. These aims will be addressed with a combination of computational simulations, psychophysical testing, self- report, functional magnetic resonance imaging (fMRI), electro-, and magneto-encephalography (EEG, MEG). The proposed work bridges traditionally segregated research on attention and expectation, and advances our knowledge of how humans make sense of, and prioritize, their visual environment. It is furthermore directly relevant to improving our understanding of potential 'failure modes' of visual cognition in patient populations who have difficulty with controlling attention or with accurately predicting and interpreting sensory information.