Entropic Theory Explains Insistence on Sameness in Autism

The phenomenon of "insistence on sameness" observed in autism is a complex behavioral pattern. This research introduces a novel framework, rooted in information theory, that attempts to explain this behavior as a fundamental strategy to minimize surprise and uncertainty. The paper redefines autism as a condition where cognitive functions are primarily limited to the discrimination, memorization, and prediction of tangible environmental properties. The framework draws an analogy between insistence on sameness and the constrained minimization of an entropy metric, $D_H(R, M) = H(R|M) + H(M|R)$, where R represents stimuli sequences and M is memory. This metric interprets conditional entropies as surprise and uncertainty. The theory infers that to minimize this metric, an individual can either learn about new stimuli or restrict their environment to what is already known, with insistence on sameness being a manifestation of the latter. Beyond explanation, the proposed framework offers several practical implications. It helps quantify concepts like surprise, uncertainty, sensory overload, and comfort zones. Crucially, it leads to a list of guidelines for learning therapies and daily care routines, suggesting they can be defined as optimization algorithms and implemented for robotic caregivers. The framework also proposes a Turing test-like validation method that avoids direct experiments with individuals with autism.