Over-Reliance on Prior Expectations in Relapsing-Remitting Multiple Sclerosis

Cognitive impairment is a common and disabling feature of multiple sclerosis (MS). Bayesian models of perception and action provide a powerful framework to better understand how cognitive processes are altered in MS. In this case-control study, we employed a time reproduction paradigm within a Bayesian framework to investigate the underlying mechanisms of cognitive dysfunction in patients with relapsing-remitting MS. We applied a modified Bayesian observer model, which partitions time reproduction into three stages: sensory measurement, time estimation, and motor response. We found that both MS and control groups showed a systematic bias in reproducing time, overestimating short intervals and underestimating long intervals. This bias was significantly larger in patients with MS compared with controls, reflecting an over-reliance on prior expectations relative to sensory-motor information. Computational modeling indicated that this increased bias in the MS group was driven by greater measurement noise during the sensory stage. Moreover, central tendency bias increases with age in healthy participants as reliance on prior expectations becomes stronger than sensory-motor evidence. Interestingly, we found that this age-related effect on bias was absent in patients with MS. Further analysis showed both younger and older patients performed equally biased, and their performance was similar to older healthy participants.