Objectives/Aims Dissociative amnesia refers to loss of autobiographical memory with a presumed psychological, rather than neurological cause. The neurocognitive mechanisms underlying the memory loss are not fully understood. One theory suggests that prefrontal control mechanisms ‘block’ access to autobiographical memories. A parallel body of experimental work has defined a prefrontally-mediated memory control network which can be engaged voluntarily to inhibit memory retrieval. We aimed to examine the possible role of this retrieval control system in dissociative amnesia.

Methods We re-analysed fMRI data from two previously reported cases of dissociative amnesia (Kikuchi et al. 2009). Both patients had extensive focal retrograde amnesia covering several years, with preserved anterograde memory. In the scanner, they were presented with reminders of people from the forgotten period (current colleagues), as well as a period they could still remember (high school friends). Contrasting the forgotten with the remembered condition, we examined the extent to which the activated regions overlapped with regions involved in retrieval control, derived from a meta-analysis of retrieval inhibition tasks (the ‘think/no-think task’). We used dynamic causal modelling to examine the effective connectivity between the right dorsolateral prefrontal cortex (DLPFC) and hippocampus when patients were reminded of people they could not recognise. One patient later recovered their memories, whilst the other remained amnesic. Both were scanned a second time, and we repeated the same analyses on these data to examine changes associated with recovery.

Results We observed increased DLPFC activation and decreased hippocampal activation in response to reminders of forgotten colleagues, relative to remembered high school friends. There was a high degree of overlap with the regions engaged during typical laboratory-based tasks of memory inhibition. Using dynamic causal modelling, we found clear evidence that the right DLPFC downregulated the right hippocampus when the patients were reminded of forgotten colleagues. This pattern persisted in the patient who remained amnesic at follow-up, but was no longer evident in the patient who recovered their memories.

Conclusions These findings provide preliminary evidence that retrieval control systems may be maladaptively recruited to inhibit memory retrieval in dissociative amnesia. This is often temporary, and may occur on a semi- or sub-conscious level. These findings may provide a mechanistic target for current and future treatment approaches.