Abstract
What is the time course of brightness induction? Rossi & Paradiso (1996) measured the strength of this illusion when the inducing region's luminance was sinusoidally modulated over time. The illusion decreased with increasing temporal frequency, eventually disappearing at between 2–4Hz. The higher the spatial frequency of the illusion, the higher the temporal frequency at which it could still be seen. From this, they theorized that brightness induction is due to a neural filling-in signal that propagates slowly. Here we introduce a new paradigm for investigating the timecourse of this illusion and find results inconsistent with slow filling-in.
METHOD: In our paradigm the stimulus was shown briefly (OnTime=58, 82, 117, or 1120ms, depending on condition), and then covered by a noise mask for 900ms. The display alternated between showing the noise mask and the stimulus while subjects adjusted the brightness of a constantly visible patch to match the perceived strength of brightness induction. Subjects typically took 30 to 60 seconds to find a satisfactory match.
RESULTS: We found the brightness induction illusion was visible when OnTime was just 58 ms. In addition, the illusion was actually stronger at short OnTimes, not weaker. Below 58ms subjects could not reliably see the whole stimuli, and thus could not make brightness matches at all. The same patterns of results were seen for high and low spatial frequency stimuli (0.5 cpd and 0.05 cpd, respectively), suggesting little difference in the speed of induction as a function of spatial frequency.
CONCLUSIONS: These results suggest that the brightness induction illusion develops much faster than previously thought. This suggests that filling-in may not be involved, or that if it is, filling-in is much faster than previously thought.
Funding: NSF Grant DGE-0333451 to GW Cottrell and NSF CAREER Award 0133996 to VR de Sa.