Autobiographical memory in Alzheimer's disease: a systematic review

The study investigates how Alzheimer's disease (AD) affects autobiographical memory (AM), focusing on memory specificity, temporal gradients (remote vs recent memories), and emotional processing. AD is a progressive neurodegenerative condition with early involvement of memory systems and widespread cognitive, social, and self-related impairments. AM comprises episodic (event-specific, vivid recollection) and personal semantic knowledge; episodic AM is considered the hallmark of AM and appears disproportionately impaired in AD. Key phenomenological features of AM—vividness, belief in accuracy, sensory details, emotional valence and intensity, accessibility, frequency of sharing, and narrative coherence—are relevant to understanding AM decline in AD. Temporal distribution phenomena (childhood amnesia, reminiscence bump, recency effect) and neuroanatomical changes (DMN disruption, hippocampal/medial temporal atrophy, prefrontal dysfunction) provide context. Despite the significance of AM for identity and quality of life, recent comprehensive synthesis was lacking; this systematic review aims to integrate current evidence on AM impairment in AD, its cognitive/neural correlates, and implications for self-identity and intervention.

Prior research consistently shows that AD patients have impaired autobiographical memory relative to healthy controls, with episodic components more affected than personal semantic memory. Anosognosia studies reveal discrepancies between subjective memory awareness and objective performance. The typical temporal distribution of AM includes childhood amnesia, the reminiscence bump (especially self-defining, emotionally salient events from ages ~10–30), and recency effects; in AD, the reminiscence bump often persists but recent memories are disproportionately impaired, aligning with Ribot’s law. Neuroanatomically, AM decline in AD corresponds to default mode network disruption, hippocampal/medial temporal atrophy, and prefrontal dysfunction, with compensatory shifts from episodic to semantic content possibly involving left prefrontal overactivation. Comparative work with other dementias (semantic dementia, bvFTD, LPA) highlights syndrome-specific AM profiles and temporal gradients. Despite episodic deficits, aspects of emotional intensity and self-referential processing may be relatively preserved in early/mild AD.

Design: Systematic review adhering to PRISMA guidelines (pre-registered in PROSPERO: CRD42024596837). Data sources: PubMed, Scopus, ScienceDirect, and Web of Science; search conducted May 24 (no predefined start year). Search strategy: Combinations of terms related to Alzheimer’s and autobiographical memory (e.g., "Autobiographical memory," "self-defining memory," "memory phenomenology"), applied to titles, abstracts, and keywords. Screening and data management: Citations imported into Rayyan to deduplicate and support screening. Two independent reviewers screened titles/abstracts/full texts against predefined criteria; disagreements resolved by consensus, with a third reviewer as arbiter. Inclusion criteria: (a) direct comparison of AD subsample to healthy controls; (b) English-language; (c) quantitative/qualitative cognitive data (no case studies); (d) peer-reviewed empirical studies; (e) diagnosis of AD or probable AD. Quality assessment: QUADAS-C (Yang et al., extension of QUADAS-2) assessing risk of bias across Patient Selection, Index Test, Reference Standard, Flow and Timing, including blinding, consistency of diagnostic criteria, and potential flow/timing biases. Classification: Studies narratively synthesized into eight categories—general AM deficits; specificity; temporal gradient; emotional components; stimulus effects on retrieval; AM–self relations; comparisons with other neurological disorders; neural correlates. Search results: Initial records n=1,987; duplicates removed n=1,049; records screened n=938; records excluded n=827 (foreign language n=28; wrong publication type n=468; wrong population n=243; wrong outcome n=88); full-text assessed n=111; full-text excluded n=28 (reasons included wrong population n=6; foreign language n=5; wrong publication type n=3; duplicate n=2; no control group n=5; wrong topic n=3; no assessment of AD n=1; full text not available n=3); studies included n=83. Disease severity coverage: MCI (15%), mild AD (54%; MMSE ~21–26), moderate AD (13%; MMSE 16–20), advanced AD/severe (5%; MMSE <16), multiple levels (13%). AM assessment tools across studies included AMI, TEMPau, AI, AMT, SDM, EAMI, Remember/Know paradigm, sensory-evoked (music, odors), and life narrative methods.

• Included 83 studies; robust evidence of AM deficits in AD across episodic and semantic components, with disproportionate impairment in episodic details and a tendency toward overgeneralized narratives (reduced internal details, increased external/semantic details). - Temporal gradient: Remote memories better preserved than recent ones, supporting Ribot’s law; reminiscence bump generally retained but with reduced specificity; some exceptions (e.g., relatively better recall for very recent events in specific contexts). - Emotional processing: Mixed valence patterns; reports of positivity bias; some studies show emotional neutrality with fewer positive/negative memories; emotional intensity can be preserved despite reduced specificity. - Stimulus effectiveness: Sensory cues (self-chosen music, odors, concrete objects, photographs, films) often enhance AM retrieval and narrative quality; odors and music particularly effective; self-selected cues outperform experimenter-selected ones. - Neural correlates: AM deficits associated with hippocampal atrophy, prefrontal dysfunction, and posterior cortical involvement (precuneus, retrosplenial cortex); disrupted connectivity between hippocampal and prefrontal/temporal-frontal networks; compensatory activation observed in left inferior frontal gyrus, ventromedial prefrontal cortex, left lingual gyrus. - Cognitive relationships: Significant correlations between AM performance and executive functions (organization, planning, strategic search), attention, and processing speed; semantic fluency positively correlates with episodic AM; working memory and phonemic fluency less predictive. - Sense of self: Despite AM impairment, mild-to-moderate AD patients can retrieve self-defining memories and maintain aspects of self-continuity; AM-rated centrality to identity comparable to healthy controls in some studies. - Methodology distribution and severity: Predominantly cross-sectional case-control designs; studies span MCI to severe AD; severity distribution—MCI 15%, mild AD 54%, moderate AD 13%, advanced AD 5%, multiple levels 13. - Screening statistics: Initial records 1,987; duplicates removed 1,049; screened 938; excluded 827; full-text assessed 111; included 83.

AM impairment in AD spans multiple cognitive domains and impacts identity, social functioning, and quality of life. The consistent reduction in episodic specificity, altered temporal gradients (remote > recent), and modified emotional processing provide converging evidence for characteristic AM pathology in AD, aligning with hippocampal–prefrontal–posterior cortical network dysfunction. These findings address the review’s research questions by clarifying which components of AM are most vulnerable, how temporal distribution is altered, and which stimulus modalities can leverage preserved systems to enhance retrieval. Clinical implications include the potential of AM features (e.g., specificity, temporal gradients) as early diagnostic markers, the value of multimodal, patient-tailored cues (music, odors, concrete objects, photos) in rehabilitation, and the importance of targeting executive processes to support AM reconstruction. Standardization of AM assessment protocols is needed to improve comparability and facilitate clinical translation. The relationship between AM and self-identity suggests interventions aimed at self-defining memories may bolster self-continuity. Early detection of AM changes, particularly in MCI, may inform prognostic monitoring and timely intervention.

This systematic review synthesizes contemporary evidence showing pervasive autobiographical memory deficits in AD, marked by reduced episodic specificity, altered temporal gradients with preservation of remote and reminiscence bump memories, and distinctive emotional processing patterns. It identifies effective retrieval-enhancing strategies using sensory and personally meaningful cues, and delineates neural substrates implicating hippocampal, prefrontal, and posterior cortical regions. Contributions include: consolidating heterogeneous findings into thematic domains; highlighting AM’s links to executive function and self-identity; and proposing standardized assessment needs. Future research should employ larger, diverse cohorts; harmonized AM protocols; longitudinal designs tracking progression from MCI to AD; and intervention trials leveraging multimodal cues (music, odors, images) and executive function training. Integrating construction vs elaboration and direct vs generative retrieval frameworks may guide phase-specific interventions and biomarker development, ultimately improving diagnostic precision and quality of life in AD.

• Considerable methodological heterogeneity across studies (diverse AM instruments and protocols) limiting direct comparability and precluding meta-analysis. - Many studies with relatively small sample sizes, reducing generalizability. - Variability in disease severity operationalization and assessment tools across studies. - Potential biases addressed via QUADAS-C, but differences in blinding, reference standards, and flow/timing may remain. Future work should standardize AM assessment, use larger and more diverse samples, and adopt longitudinal designs to strengthen inference and comparability.