Pattern separation and pattern completion for human face identity recognition in prodromal Alzheimer’s disease

Abstract

AbstractBackgroundThe integrity of face identity recognition (FIR) in prodromal Alzheimer’s disease (prAD) is inconsistent in the literature, and the cognitive processes underpinning a potential impairment remain elusive. The concepts of pattern separation (PS) and pattern completion (PC) offer an interesting insight in this respect, since efficient FIR implies that facial identities are stored in at least partly distinct representations (corresponding to a PS process) and to fill‐in the currently perceived facial pattern based on these mnemonic representations (corresponding to the PC process). The goal of our study was therefore to investigate FIR in prAD patients with a PC and PS paradigm. In addition, we aimed to verify that prAD patients were able to individuate faces in a task that does not involve any long‐term memory component.MethodThirty‐two healthy older participants and 17 prAD patients were familiarized with 40 facial identities, through a picture‐matching task (Fig. 1.A). Participants then performed a forced‐choice recognition task aiming at assessing and strengthening the memory encoding of these 40 faces (Fig. 1.B). They then underwent separate yes/no recognition tasks involving parametrically controlled blurred or morphed faces (Fig. 2), designed to solicit PC and PS processes respectively (Fig. 1.C‐D). Finally, in two separate discrimination tasks (Fig. 1.E‐F), participants were instructed to determine whether two simultaneously displayed faces were identical or not. These discrimination tasks included the blurred and morphed faces displayed in the PC and PS recognition tasks respectively.ResultsprAD patients showed significantly lower performance than controls in both the PC and PS recognition tasks (Fig. 3). However, this group difference was not significant when the recognition accuracy index for the non‐manipulated faces (not blurred nor morphed) displayed in these tasks was included as a covariate. In addition, group differences were unexpectedly found for the familiarization and discrimination tasks.ConclusionOur study failed to highlight specific PC or PS deficits in prAD patients. However, it showed that prAD patients performed poorly on tasks that required comparing and matching simultaneously presented facial patterns, in the absence of any long‐term memory component. Future research should investigate whether this impairment could represent an early cognitive marker of AD that could be easily assessed in clinical practice.