Why do some people see the illusion more than others?
The sound-induced flash illusion (SIFI) is a simple test: you see one flash but hear two beeps, and you may report seeing two flashes (fission), or you see two flashes with one beep and report one (fusion). How strongly you fall for it depends partly on how your brain binds sights and sounds in time. People with a wider temporal binding window—the window of time within which a sound and a flash are treated as one event—tend to be more susceptible to audiovisual illusions [2].
Pilots illustrate this. In a study of 66 participants, pilots were less susceptible to both fission and fusion versions of the SIFI than controls, and they showed a narrower temporal binding window in the fusion condition, meaning their susceptibility changed more gradually as the timing between flash and beep increased [1]. The authors suggest that in the fission condition the group difference was mainly driven by visual sensitivity, while in the fusion condition it likely also reflected different multisensory integration mechanisms [1]. This is a correlation between a trained group and illusion strength, not proof that training caused the difference.
Does age or experience shift how strongly the illusion works?
Susceptibility also changes with age. In 118 children aged 4 to 17, older children were less susceptible to the SIFI at longer delays between flash and beep, which the authors interpret as developmental refinement in audiovisual integration [5]. Exploratory analyses found that children with dance training and those who found the task easier were slightly less susceptible [5]. These are associations, not demonstrations that practice or training reduces illusion strength.
The pilot findings point in a similar direction: a group with extensive visual experience shows weaker illusions, at least under some conditions [1]. But neither study manipulated visual experience or visual statistics, so they cannot tell us whether adapting to altered visual input would change illusion susceptibility.
Is there any evidence that adapting to altered visual input changes illusion strength?
One study comes closest. In a virtual-environment experiment, visual cues influenced haptic perception, and the user's visual-motor system adapted and adjusted processing of visual inputs and haptic outputs, leaving the user temporarily adapted [4]. That shows cross-modal adaptation can occur, but it does not measure susceptibility to a multisensory illusion before and after adaptation, so it cannot say whether adaptation changes illusion strength.
Adding sound can also change how strongly people feel a body-resizing illusion: in 44 participants, non-naturalistic auditory input increased subjective illusion strength when there was no tactile feedback, but not when tactile feedback was present, and it had no effect on performance tasks [3]. This is about adding a sound, not about adapting vision, so it does not answer the adaptation question either.
What can we reasonably conclude?
The honest answer is that the direct experiment has not been reported in these studies. What exists are clues: illusion susceptibility tracks individual differences in temporal binding [2], differs between groups with different visual experience such as pilots [1], and changes across childhood [5]. Cross-modal adaptation to altered visual input has been observed in a virtual environment [4], and auditory input can modulate a bodily illusion [3].
Putting these together, it is plausible—but not demonstrated—that adapting to altered visual statistics could shift multisensory illusion susceptibility by changing visual sensitivity or the temporal binding window. Testing that would require measuring illusion strength before and after a controlled period of visual adaptation, which none of these studies did.
Sources used in this answer
Altered multisensory integration in pilots: Examining susceptibility to fission and fusion sound-induced flash illusions
Pilots were less susceptible to both fission and fusion sound-induced flash illusions than controls, with a narrower temporal binding window only in the fusion condition; in fission the difference was mainly driven by visual sensitivity, while fusion likely also reflected different multisensory integration mechanisms.
Individual differences in the multisensory temporal binding window predict susceptibility to audiovisual illusions.
Individual differences in the multisensory temporal binding window predict susceptibility to audiovisual illusions including the McGurk effect and the sound-induced flash illusion.
Multisensory processing and proprioceptive plasticity during resizing illusions
Adding non-naturalistic auditory input increased subjective strength of a hand-resizing illusion when tactile feedback was absent but not when it was present, and had no facilitatory effect on performance tasks.
Visual touch in virtual environments: An exploratory study of presence, multimodal interfaces, and cross-modal sensory illusions
In a virtual environment, visual cues influenced haptic perception, and the user's visual-motor system adapted and adjusted processing of visual inputs and haptic outputs, leaving the user temporarily adapted.
Investigating developmental changes in susceptibility to temporal audiovisual illusions on balance and sensorimotor function in children.
Older children (aged 4–17) were less susceptible to the sound-induced flash illusion at longer stimulus onset asynchronies, and exploratory analyses found children with dance training and those who found the task easier were slightly less susceptible; no strong evidence linked SIFI performance to balance or motor performance.