Individual differences in internal noise are consistent across two measurement techniques

Greta Vilidaite, Daniel Hart Baker

Research output: Contribution to journalArticlepeer-review

Abstract

Internal noise is a fundamental limiting property on visual processing. Internal noise has previously been estimated with the equivalent noise paradigm using broadband white noise masks and assuming a linear model. However, in addition to introducing noise into the detecting channel, white noise masks can suppress neural signals, and the linear model does not satisfactorily explain data from other paradigms. Here we propose estimating internal noise from a nonlinear gain control model fitted to contrast discrimination data. This method, and noise estimates from the equivalent noise paradigm, are compared to a direct psychophysical measure of noise (double-pass consistency) using a detailed dataset with seven observers. Additionally, contrast discrimination and double-pass paradigms were further examined with a refined set of conditions in 40 observers. We demonstrate that the gain control model produces more accurate double-pass consistency predictions than a linear model. We also show that the noise parameter is strongly related to consistency scores whereas the gain control parameter is not; a differentiation of which the equivalent noise paradigm is not capable. Lastly, we argue that both the contrast discrimination and the double-pass paradigms are sensitive measures of internal noise that can be used in the study of individual differences.
Original languageEnglish
Pages (from-to)30 - 39
Number of pages10
JournalVision Research
Volume141
Early online date22 Dec 2016
DOIs
Publication statusPublished - 10 Dec 2017

Bibliographical note

© 2016 Published by Elsevier Ltd. All rights reserved. This is an author-produced version of the published paper. Uploaded in accordance with the publisher’s self-archiving policy.

Keywords

  • contrast perception
  • internal noise
  • double-pass
  • equivalent noise
  • gain control model

Cite this