Does the Ebbinghaus Illusion Affect Optimal Pointing?

We investigated speeded pointing movements (Trommershauser, Maloney, and Landy, 2003) in the presence of the Ebbinghaus illusion. Franz et al (2000) reported systematic effects of the Ebbinghaus illusion on grasping movements, contradicting earlier results (Aglioti, et al, 1995) and casting doubt on a strict dissociation between action and perception (Goodale and Milner, 1992). Here, we try to extend this finding by explicitly manipulating the consequence of motor actions in a speeded pointing task. We hypothesised that large, medium, and small circular surrounds induce the perception of smaller, unchanged, and larger target and penalty areas at the centre. If subjects take the visual illusion into account, then on average they should overshoot, hit, and undershoot the optimal point in the three illusion conditions, respectively. Subjects were asked to hit a circular target area on a touch screen with their right index finger. They were awarded +100 points for hitting the target, and in separate conditions either lost 0 or 500 points when touching an overlapping red penalty area. Feedback was provided at the end of each trial so the subjects could monitor their total score and gauge their performance. Participants were trained in 240 trials before performing the experiment in two blocks of 120 trials for each penalty condition with large, medium, and small surround stimuli randomly intermixed. Preliminary results from 12 observers indicate a significant effect of penalty on pointing position (F1,11=21.3, p=0.001) but no statistically significant effect for the Ebbinghaus illusion (F2,22=1.14, p=0.34). Results from a size adjustment task at the end of the experiment suggest that the perceived size of target and penalty area changed as predicted (large: −1.5, medium: 0.7, and small: 1.5 pixels) but that the effect was too weak to systematically influence pointing. In conclusion, we found no clear evidence that the Ebbinghaus illusion can bias optimal motor decisions, but our stimuli and design need to be improved before we can confirm these findings