Modeling Neurocognitive Mechanisms in Multistable Perception: Insights Into Variability in Autism

Multistable perception, characterized by spontaneous alternations between perceptual states, has garnered increasing attention in autism research, because atypical performance in multistable perception tasks has been suggested as a behavioral marker of this population. However, previous studies in this field have yielded variable and conflicting results leading to inconsistent conclusions. Here we present a computational model integrating key neurocognitive mechanisms, all of which have been considered critical in accounting for atypical behavior in autism, such as bottom-up/top-down processes, the excitation-inhibition (E/I) balance, and neural noise. By systematically altering these mechanisms within an artificial neural network, we explored their functional roles on perceptual fluctuations. Our approach provides a clearer understanding of how specific behaviors may correlate with varied neural and cognitive alterations in autism. This model offers a perspective on key debates in autism research, such as the direction for the excitation/inhibition imbalance, and the possible dominance of bottom-up processing. Moreover, this model challenges heuristic single-mechanism theories of autism, favoring neurodiversity approaches by illustrating the spectrum’s complexity, where identical behaviors might result from diverse neurocognitive mechanisms. We highlight the need for future research to employ multiple behavioral tasks to map neurocognitive alterations in autism, aiming for more precise identification and management in therapeutic scenarios.