Eye-Tracking Study Using Cellular Automata Patterns as Visual Stimuli

Anirudh Tiwathia & Cristián Opazo

Vassar College

The visual system, like other mechanisms of human perception, limits and selects the information to be subsequently analyzed by higher cognitive processes. Sophisticated mechanisms at the initial level select only a small amount of the available visual information to be deeply processed. In this selection process, two additional processes have been identified: bottom-up (stimulus driven) and top-down (observer driven). Their relationship and influence are still subjects of debate. This study examined goal-free viewing of cellular automata (CA) images to address the nature of the bottom-up process, the robustness of salience as a framework for explaining fixation points, and the particular features that can characterize salience. The influence of familiarity on oculomotor strategy was also addressed.

A qualitative study of the results showed promising trends. Prior experience with CA images affected the eye-movement patterns of observers, suggesting the presence of some top-down influences. Higher-level structural features such as pockets of regularity within randomness or localized structures within regularity were salient for most participants. These results raise questions about the kinds of visual features that can be used to characterize salience. In another unexpected result, many fixations occurred in blank regions within images featuring nested (fractal) structures. Many of these findings escape current psychophysical models of oculomotor strategy, and pose interesting avenues for further research.

[presentation materials]

Created by Mathematica  (May 11, 2006)