Justin Junge

Tufts University

In principle, information theory is capable of describing all signal processing, but we are
still uncovering the specific algorithms that do the work in human vision.  In the current
studies, we randomly generated a large sample of stimuli, constrained to exhibit a substantial
range of perceived complexity, and used these stimuli in several tasks.  After collecting a first
round of data, an extensive exploratory analysis yielded a new metric that precisely predicts
response time in a same/different task (e.g. r = 0.99), and also predicts subjective reports of
stimulus simplicity and complexity (e.g. r = .95).  This metric appears to map onto one or more
central capacity limits in human vision, and the stimulus set may prove useful for controlled
manipulations of visual complexity.