Controversies in dealing with visual noise S Klein, J Ding, D Levi |
|---|
School of Optometry, UC Berkeley, CA, United States
|
| We will consider three outstanding problems related to visual noise: 1) The great difficulty of distinguishing multiplicative noise from contrast gain control (Klein vs Katkov-Sagi ‘singularity’). 2) Assessing the errors that are made when replacing stochastic noise with an analytic model when nonlinearities are present (Klein vs Lu-Dosher). 3) The role of uncertainty in accounting for the differences between experimental results across labs (Klein-Levi vs Dosher-Lu). This presentation will summarize our previous work on dealing with these controversies. In addition we will discuss three recommendations for resolving these issues: a) Rating responses should always be used so that the ROC slope can be used to assess multiplicative noise. b) Different stimulus conditions should be intermixed and blocked to determine the uncertainty effects. c) Monocular vs binocular experiments can reveal the role of noise. Finally, we will offer new data on three binocular summation tasks that are useful for dealing with these questions: contrast matching, contrast discrimination, location matching. By dealing with all three tasks together most simple models can be eliminated. The discrimination task is the one task that specifically measures noise, the other tasks are needed to constrain the models. |
| Up Home |
|---|