Font Size: a A A

Dynamics and flexibility of population coding in the middle temporal area

Posted on:2008-07-17Degree:Ph.DType:Dissertation
University:Stanford UniversityCandidate:Cohen, Marlene RochelleFull Text:PDF
GTID:1440390005450474Subject:Biology
Abstract/Summary:
To be useful for guiding behavior, populations of neurons must encode sensory information quickly and accurately. Furthermore, sensory information must be read out by neurons involved in planning the behavioral response, and the mapping between sensory representations and behavioral responses must be flexible to allow for multiple responses to the same sensory stimulus. We studied the ability of populations of neurons in the middle temporal area (MT) to encode information about motion direction on the timescale of behavioral decisions, and how the sensory representation changes with the demands of the perceptual task at hand. We first studied the sensitivity of single MT neurons, and found that monkeys are more sensitive to motion direction than even optimally tuned single neurons, suggesting that monkeys must pool information from multiple MT neurons in order to achieve the observed psychophysical performance. Second, we studied dynamic changes in the functional decision-making circuitry by analyzing the noise correlations of pairs of simultaneously recorded MT neurons in two behavioral contexts. We found that identical visual stimuli give rise to different noise correlations depending on whether the spatial structure of the task promotes cooperate or competitive interactions between the two neurons. This result suggests that MT neurons receive inputs of central origin whose strength changes with the task structure. The changes in noise correlation appear to reflect differences in how MT neurons are pooled for the purpose of discrimination in the two task structures. Our data are consistent with a model in which the pre-motor neurons that read out MT responses for the purpose of decision-making feed back onto MT neurons, and the strength of feedback is dependent on the spatial structure of the task. This feedback serves to separate MT responses into pools that are relevant for the task parameters on a particular trial, and may be a general mechanism by which the responses of sensory neurons are flexibly separated into evidence in favor of particular perceptual decisions.
Keywords/Search Tags:Neurons, Sensory, Responses, Information
Related items