Does color affect grouping in Gabor snakes?

Katarina Bulovic

Gestalt theory proposes a list of principles that guide the ways in which people interpret ambiguous visual stimuli. One of these Gestalt principles is continuation, meaning we generally group together stimuli which could be interpreted as a continuous line or object. One visual stimulus which demonstrates this phenomenon is Gabor snakes. Gabor filters are randomly positioned and oriented, except for one line (or "snake) of Gabor filters which are positioned and oriented in a way so that they can be interpreted as being continuous. As a result, our visual system groups them together into a "snake".

Previous research has explored different changes to the Gabor snakes which do or do not weaken the grouping effect. For example, varying the angle of the Gabors in the snake can weaken the effect, but varying the phase does not. In this experiment, I vary the color of the Gabor filters. The first two images shown are the control conditions, where every Gabor filter in the image is colored in the same way (red/green in (a) and blue/yellow in (b)). Image (c) is the experimental condition, where the color of each Gabor filter is randomly chosen to be either red/green or blue/yellow. I predict that it will be slightly more difficult to group the Gabors together into a continuous snake in the randomized color condition.

(a)


(b)


(c)


Field, David J., Hayes, Anthony, Hess, Robert F. (1992). Contour Integration by the Human Visual System: Evidence for a Local “Association Field”.

Comments


Katarina Bulovic

c) The reports don't completely answer my question because different observers saw different results.
d) To answer the question more completely, it might help to have more observers, and to have a more controlled way of measuring exactly how long it takes each observer to find the snake in each condition.

Griffin Leonard

a) In the control conditions I could easily spot the snakes. In the third picture, a snake popped out for me in a backward C shape on the left side of the screen, just as quickly as in the control conditions. However, after looking at each image a second time I noticed the entire snake for the second control condition and the third picture (which looks like a S shape), which I didn't immediately notice before.
b) This seems to show that changing the colors doesn't really effect the the grouping of the gabor filters. However, it is super hard to judge how quickly I notice the snakes. After I know the snakes are there, they pop out just as easily in the experimental condition as in the controls. Yes, this experiment answers the question.

Maddie C

I saw all of them right away (maybe I've looked at too many of these in my life?) But I may have been lucky... I think my gaze just happened to land on where there were a couple blue/yellow ones in a row and then I saw the whole thing.

Richard McWalter

This was a very nice illusion. I found the randomized color has a little change on the "strength" of the Gabor snake. But it still pops out quite well and forms a continuous line.

Nicholas Guiliano

a) For both control conditions, my ability to group the gabors was fine and I could distinguish patterns. The randomized color gabors were very chaotic and nearly completely without patten, with only extremely mimimal grouping occuring at local phenomena where the randomization put the same color together a few times.

b) This validates the experiement's hypothesis, as varying the colors of the gabor filters prevented their grouping. This also indicates that unrelated colors override relatable contours, as one cannot see a multicolored snake.