Brain complexity represents uncertainty in sequence learning and corroborates habituation deficit in Parkinson disease patients.

Information processing is a core element of cognitive processes in the brain. Rooted in the idea postulated by John von Neumann that the stochasticity of neural operations might essentially be a reflection of the probabilistic nature of information, it can be argued that fluctuations in the complexity of the brain operations should reflect aspects of information processing activities undertaken to infer nuances embedded in the sensory data. The current study elucidates these aspects by examining the intricate relationship between information processing in the brain, reflected in the complexity of brain activities, and information-theoretical measures of uncertainty in the Bayesian models of sequence learning. A key finding of our study is that the brain complexity shows a significantly superior correlation with the uncertainty about the input, captured by the entropy of the posterior distribution over all input values, compared to its correlation with the novel information stored in the current input as measured by the Shannon or Bayesian models of surprise. Additionally, we found that the brain complexity follows the habituation process known to occur during learning by demonstrating a decreasing trend and also corroborates the process of spontaneous recovery at the onset of a repeated learning task. As further evidence for this relationship, distinguished complexity trends are observed in patients of Parkinson disease, reflecting cognitive disorders in habituation and habit formation caused by the disease.