Exploring the content and quality of episodic future simulations in semantic dementia

Abstract

Semantic dementia (SD) is a progressive neurodegenerative disorder characterised by the amodal loss of semantic knowledge in the context of relatively preserved recent episodic memory. Recent studies have demonstrated that despite relatively intact episodic memory the capacity for future simulation in SD is profoundly impaired, resulting in an asymmetric profile where past retrieval is significantly better than future simulation (referred to as a past>future effect). Here, we sought to identify the origins of this asymmetric profile by conducting a fine-grained analysis of the contextual details provided during past retrieval and future simulation in SD. Participants with SD (n=14), Alzheimer's disease (n=11), and healthy controls (n=14) had previously completed an experimental past-future interview in which they generated three past events from the previous year, and three future events in the next year, and provided subjective qualitative ratings of vividness, emotional valence, emotional intensity, task difficulty, and personal significance for each event described. Our results confirmed the striking impairment for future simulation in SD, despite a relative preservation of past episodic retrieval. Examination of the contextual details provided for past memories and future simulations revealed significant impairments irrespective of contextual detail type for future simulations in SD, and demonstrated that the future thinking deficit in this cohort was driven by a marked decline in the provision of internal (episodic) event details. In contrast with this past>future effect for internal event details, SD patients displayed a future>past effect for external (non-episodic) event details. Analyses of the qualitative ratings provided for past and future events indicated that SD patients' phenomenological experience did not differ between temporal conditions. Our findings underscore the fact that successful extraction of episodic elements from the past is not sufficient for the generation of novel future simulations in SD. The notable disconnect between objective task performance and patients' subjective experience during future simulation likely reflects the tendency of SD patients to recast entire past events into the future condition. Accordingly, the familiarity of the recapitulated details results in similar ratings of vividness and emotionality across temporal conditions, despite marked differences in the richness of contextual details as the patient moves from the past to the future.