The role of luminance transients in the generation of the sound-induced flash illusion

Abstract

Listen

Translate article

The illusion of a single pulse of light seen as two flashes under certain viewing conditions has been known for nearly a century. The sound-induced flash illusion (SIFI, Shams 2000) is another example of extra flashes being reported beyond what is presented (‘fission’), but here in the context of co-occuring beeps. There have also been reports of ‘fusion’ in the SIFI: fewer flashes reported than presented, though Shams and colleagues did not find this. In the initial series of five experiments, the frequency of fusion vs fission was measured as a function of relative timings between the beeps and the flashes (SOAs). Particular care was taken to eliminate artefactual influences on the results: potential M pathway deficits in participants, afterimages, temporal resolution of flashes and beeps, expectancy effects and response biases. Without any cross-modal stimulation, there were small fission effects in reporting one flash and larger fusion effects in reporting three flashes. In comparison to this control condition, significant fission was found for SOAs within 100 msec, and fusion for SOAs beyond this value. Bowen (1989)’s explanation of unimodal multiple flash illusion was then applied to explaining SIFI fission effects. To do so, we modelled the biphasic temporal responses of actual fast ON- and OFF- midget parasol M cells to visual stimuli used in SIFI research. Whether auditory stimuli – beeps – direct attention to short duration peaks in neural responding, thus biasing the response of the observer in reporting the number of ‘transient’ events was considered. It is shown that for a square wave temporal profile, such response bias could easily be the case, especially given that three or four beeps rarely yield an illusion of more than two flashes. To eliminate these potential artefacts, a novel approach was developed: the use of a sawtooth temporal profile (either on or off) to create one physical transient in luminance for the sound to effect.