Musicians are less prone to the ventriloquism illusion, study finds

An experiment comparing musicians and nonmusicians found that musicians tend to be less susceptible to the ventriloquism effect. When a sound came from one location and a visual flash from another, musicians’ judgments of where the sound came from were pulled less toward the flash than nonmusicians’ judgments were. The paper was published in the Journal of Experimental Psychology: Human Perception and Performance.

The ventriloquism effect is a multisensory illusion in which a sound is perceived as coming from the location of a nearby visual stimulus rather than from its true source. It occurs because the brain normally combines information from different senses and gives greater weight to vision when judging where something is located.

This ability to integrate sights and sounds is essential in everyday life, but the brain must also recognize when sensory signals come from different sources and should remain separate. Musical training may be especially relevant to this process because musicians spend years learning to coordinate highly precise auditory, visual, and motor information. Previous studies have found that musicians are better than nonmusicians at detecting mismatches between simple sounds and visual events, and two earlier experiments by the same team found that musicians showed less ventriloquism when asked to locate only the sound.

Study author Matthew Paul O’Donohue of Queensland University of Technology in Australia and his colleagues wanted to understand why. Musicians might be less susceptible to the illusion because they perceive sounds or sights more precisely, or because they have a weaker built-in expectation that a sound and a flash come from the same source. Based on their earlier work, the researchers predicted, in a plan registered before data collection, that the second explanation would hold.

Study participants were 35 musicians and 35 nonmusicians, most of them first-year psychology students. The musicians averaged 22 years of age and the nonmusicians 20, and each group included 26 women. Musicians were defined as people with at least 7 years of experience playing an instrument who had played for an average of 1 hour a week over the past 12 months; on average, they had played their main instrument for about 12 years. Nonmusicians had no more than 12 months of experience playing an instrument outside school and no more than 2 years of compulsory music lessons in school.

Participants sat in a dark room facing a screen that hid five speakers. On each trial, a brief flash, a burst of noise, or both appeared from one of five positions along a horizontal line, and participants used a mouse to mark where each stimulus had come from. When both appeared, participants located each one separately, which required them to divide their attention between hearing and vision. On these combined trials, the flash and the sound came from the same position only 20% of the time.

Participants completed the full set of 420 trials twice. Before the second run, they were told that the flash and the sound came from the same location only 20% of the time, and they were reminded of this at the start of each block. This allowed the researchers to test whether a simple instruction could change people’s expectations, and whether musicians’ advantage stemmed from assumptions formed during the experiment itself, in which case the instruction should have erased the difference between the groups.

Musicians and nonmusicians were equally precise at locating the flash or the sound when only one was presented. For both groups, flashes were located less consistently the farther they appeared from straight ahead.

When both stimuli were presented, musicians’ sound judgments were pulled about 1.6 degrees less toward the flash than nonmusicians’ were, on average. This difference was modest; it held up in one of the two statistical approaches the researchers used but fell just short in the other. After participants learned the 20% figure, the ventriloquism effect shrank overall, but musicians still showed less of it than nonmusicians.

To explain these results, the researchers used a mathematical model of how the brain decides whether to combine information from two senses. The model attributed the group difference to musicians’ lower prior expectation that the sound and the flash came from the same place, rather than to differences in how precisely they could hear or see. The instruction lowered this expectation overall, but nearly all participants who had started out with an expectation below 20% raised it, suggesting that people adjusted their expectations to match the information they were given.

“Our results suggest that musicians tend to exhibit less spatial ventriloquism than nonmusicians even when distributing their attention across audition and vision. … Finally, we have shown that this prior expectation changes when participants are informed of the true probability of a common location for the stimuli, demonstrating that top–down factors can exert a strong influence on this Bayesian prior,” the study authors concluded.

The study contributes to the scientific understanding of the ventriloquism effect. However, because the researchers compared existing musicians with nonmusicians rather than assigning people to musical training, the study cannot show that musical training itself reduces the illusion, as opposed to other traits that differ between the groups. The difference between musicians and nonmusicians was also small, and numerically much smaller than in the team’s earlier experiments, in which participants only had to locate the sound.

The authors also note that the musicians played a wide range of instruments, and that some subgroups, such as orchestral musicians, might show stronger auditory spatial skills or combine visual cues differently. In addition, it is not yet known whether the instruction changed what participants perceived or simply changed how they chose to respond.

The paper, “A Multisensory Causal Inference Prior is Attenuated in Musicians and is Further Attenuated Following Instruction,” was authored by Matthew Paul O’Donohue, Naohide Yamamoto, Brian Odegaard, and Philippe Lacherez.