Cognitive project on human visual perception

Photo created by Psychopy toolbox

One of the best useful tools for psychophysicists to investigate visual motion detection is Random Dot Kinematograms. In this project, we study the effect of RDKs and the bistable moving paradigm one among the mechanisms controlling visual perceptual bistability. We wanted to show that by putting RDKs on the plaids of the moving paradigm and by changing two factors on the dots (noise and direction) we may obtain to change a human percept to perceive the direction of the ambiguous plaids according to the direction of the dots. we tried to implement and improve some methods on creating the plaids and designing the aperture as the dots only appear on the plaids and also improved the velocity of the dots and the plaids to be correlated to each other and it changed by time and some factor of the whole aperture.

After some objective tests and some try-and-error sessions instead of doing a psychophysical staircase procedure, we switched to the statistical analysis. Then the obtained data analyzed with ANOVA methods, however, no significant effect was obtained from noise and direction of the dots on human visual percept of the direction of the plaids. Also, there was no significant effect on the noise and the direction, there was still an effect based on the result of this experiment. If instead of defining the direction of the plaids in three different directions, defining them on coherence and transparent direction, based on the previous study that most of the time the subject perceives the direction as coherence, this study proves otherwise. Most of the subjects in this experiment expressed that they saw transparent motion and sometimes it was hard for them to choose between the direction but it was on the transparent motion on the period of the trial. This matter and some other factors may have a better effect such as the number of the dots on the plaids or putting the dots in the periphery instead of the center of the aperture.

Sayeh Gholipour Picha
Sayeh Gholipour Picha
PhD candidate in Explainable AI

My research interests include Visual understanding using Vision and Language Models.