Longitudinal Trajectories of Cognitive Control and Theory of Mind in Amnestic and Non-Amnestic Mild Cognitive Impairment: Prospective Associations of Objective Sleep Duration

Background/Objectives: Sleep disturbances are increasingly implicated in age-related cognitive decline and may contribute to heterogeneity in cognitive and social–cognitive functioning in Mild Cognitive Impairment (MCI). Nevertheless, it remains unclear whether longitudinal actigraphy-derived sleep parameters predict social cognition, including theory of mind (ToM), across amnestic (aMCI) and non-amnestic (naMCI) MCI subtypes. This study examined longitudinal trajectories of cognitive control and ToM in individuals with aMCI and naMCI and healthy older adults, and investigated whether actigraphy-derived sleep indices prospectively predict subsequent cognitive control and higher-order social cognition across these groups. Methods: A total of 179 participants (46 healthy controls, 75 aMCI, 58 naMCI) were assessed across three waves over approximately 16–20 months. Actigraphy-derived sleep indices were obtained using wrist actigraphy. Cognitive control was assessed with Delis–Kaplan Executive Function System subtests, whereas ToM was evaluated through tasks involving non-literal language comprehension, metaphor and proverb interpretation, higher-order mentalizing, and emotion recognition. Longitudinal trajectories were examined using mixed-design ANOVAs, and prospective sleep effects were tested using longitudinal path models. Results: Both cognitive control and ToM declined over time in the MCI groups, whereas healthy controls showed relative stability. Subtype differences in cognitive control emerged primarily in higher-order planning and rule-monitoring processes, with naMCI showing greater impairment. ToM revealed a clearer and more consistent gradient (healthy controls > aMCI > naMCI), particularly in higher-order inferential tasks. Among the actigraphy-derived sleep indices, only TST prospectively predicted later ToM performance, whereas prospective associations with cognitive control were limited and inconsistent. Conclusions: Reduced sleep duration was prospectively associated with poorer subsequent performance in higher-order social cognition, independently of the executive-control measures examined. These findings highlight phenotypic heterogeneity across MCI subtypes and suggest that objectively measured sleep duration may serve as a clinically relevant longitudinal marker of social–cognitive vulnerability. Future studies are needed to determine whether interventions targeting sleep duration are associated with improved social–cognitive outcomes.