Technical Report #196

On the Dangers of Ignoring Noise in High-Level Perception Experiments

Leola Alfonso-Reese

Abstract

Ignoring perceptual noise can lead to misinterpretations of data. The author illustrates this by investigating effects of noise on a previous category learning experiment. Results indicate that two main effects in the previous study could be entirely explained by perceptual noise. To investigate effects of noise, the author describes a procedure for estimating the distribution of percepts. This technique can be used to investigate several characteristics of perceptual noise. One application tests the assumption that perceptual noise is normally distributed. The second tests for perceptual independence of stimulus components.

Download this paper in PDF format:

Download this paper in PostScript format:


To view PDF files, download Adobe Acrobat Reader (it's free) from http://www.adobe.com/prodindex/acrobat/readstep.html.