Summary
A bioRxiv preprint reports that six days of audiovisual timing training reduced susceptibility to the sound-induced flash illusion in both children with ADHD and typically developing children. A Bayesian model indicated that the groups reached this similar behavioral outcome through different changes in perceptual precision and causal expectations.
A bioRxiv preprint reports that children with attention-deficit/hyperactivity disorder (ADHD) and typically developing children became less susceptible to an audiovisual illusion after six days of perceptual training. Although the behavioral improvement was similar, a computational analysis indicated that the two groups changed different parts of the process used to combine sensory information.
The study examined audiovisual binding: how the brain decides whether signals arriving through different senses belong together. Its results suggest that a common improvement on a perceptual task can arise from changes at different levels of this process.
What the experiments tested
The researchers conducted three experiments involving children with ADHD and typically developing children. They used the sound-induced flash illusion, in which the sounds accompanying a visual flash can influence how many flashes a person reports seeing. The illusion is a way to study how the timing and reliability of auditory and visual information affect perception.
In the first experiment, children with ADHD were more susceptible to the illusion than typically developing children. In the second, the ADHD group showed markedly flatter temporal tuning of the illusion. Temporal tuning describes how strongly the perceptual effect changes as the timing between the sound and the flash changes.
The researchers also calculated conventional temporal binding window estimates. These estimates describe the range of timing differences over which auditory and visual signals are likely to be perceptually combined. Those estimates did not distinguish the two groups, indicating that the behavioral measures did not by themselves identify which part of multisensory processing differed.
The children then completed six days of simultaneity-judgment training. After training, both groups became comparably less susceptible to the sound-induced flash illusion.
Similar behavior, different computational changes
To examine the learning process more closely, the researchers applied a temporally explicit two-dimensional Bayesian causal-inference model. In this framework, perception depends partly on the precision of sensory timing information and partly on a prior expectation about whether signals share a common cause.
The model assigned the typically developing group a reduction in its prior expectation that auditory and visual signals came from the same cause. In practical terms, the model indicated that training changed a higher-level expectation about how the two sensory signals were related.
For children with ADHD, the model instead indicated a marked reduction in temporal uncertainty. Their sensory evidence became more temporally precise, while the model did not show the corresponding recalibration of the prior expectation about a shared cause.
These are model-derived changes rather than measurements of separate brain regions or directly observed internal beliefs. Their significance is that the same reduction in illusion susceptibility was associated with different parameter changes in the computational account.
Why the result matters
The findings challenge the idea that multisensory perceptual learning has one uniform mechanism. A person may perform better on the same task because sensory evidence has become more precise, because expectations about the causes of sensory signals have changed, or because both processes have changed to different degrees.
For research on ADHD, this distinction may help explain why conventional behavioral summaries can miss differences in how children process and learn from multisensory information. It also provides a framework for separating changes in immediate perceptual precision from slower, more abstract expectations about the environment.
The work is a bioRxiv preprint and therefore an early research report. Its evidence comes from audiovisual laboratory tasks in children, so implications for diagnosis, treatment or everyday functioning require separate study. The abstract supplied with the preprint does not report sample sizes or numerical effect estimates, leaving the size and precision of the reported group differences unavailable here.