KurisutinaSarah Steinvorth, Brian Levine, and Suzanne Corkin. Medial temporal lobe structures are needed to re-experience remote autobiographical memories: evidence from H.M. and W.R. Neuropsychologia 43(4), 479–496. Peer-reviewed primary human lesion study. DOI: 10.1016/j.neuropsychologia.2005.01.001. PubMed record; public author-institution PDF.
Evidence from this source
Two amnesic patients, H.M. and W.R., were compared with eight older controls matched on age, education, and IQ. Interviews requested one event from each of five life periods, with free recall, general prompts, then structured specific probes. Scoring distinguished event-specific internal details from external, largely semantic material. The patients showed remote autobiographical impairment without a clear temporal gradient, alongside substantially better semantic and public-event knowledge. Crucially, autobiographical group differences emerged after combining all three retrieval stages; limited prompting could obscure them. H.M. nevertheless produced one richly detailed, dated teenage episode. Authors favor multiple-trace theory over standard consolidation. Lesions extended beyond the hippocampus, making hippocampus-specific necessity uncertain. Only one event per period and two heterogeneous patients limit inference. Public-event narratives controlled some demands of storytelling; they were not pure episodic-memory measures. Blinding is reported for a second scorer checking 24% of memories, with internal/external reliability coefficients .86/.90. Self-rated vividness did not consistently track detail impairment. Full paper.
Original interpretation and competing explanations
The practical inference is that an extraction procedure partly determines what appears to remain in memory. A sparse answer can reflect a difficult retrieval route, an unsuitable cue, or an impoverished representation. A database containing everything said in a first interview therefore cannot be assumed to contain everything the person could remember under other conditions.
These observations favor separating episodic specificity from autobiographical gist. They do not uniquely identify whether damaged tissue stored an episode, indexed distributed components, or enabled construction during retrieval. Nor do they establish that every remote episode requires the same tissue. Residual tissue and variability across memories make an isolated preserved recollection compatible with several accounts. Conversely, fewer reported details need not mean every underlying detail has been erased.
The controls weaken a general inability-to-tell-stories account. They cannot guarantee equivalent search difficulty, imagery demands, or original learning across autobiographical and public-event tasks. Normal-range performance against a small, variable control sample also does not establish equivalence. Selection of a memorable event is different from sampling all experienced events. These are reasons to narrow the inference, not to disregard the dissociation.
Proposed experiment for functional transfer
Evaluate acquisition under several standardized cue families: open invitation, temporal/place cues, perceptual cues, and emotion/thought cues. Separate spontaneously accessible material from material requiring support, and retain the cue-to-response relation. On held-out events and new cues, test both the content recovered and the support needed to recover it. Compare a semantic life summary, an event record, and a record with observed retrieval relations, with matched information budgets.
Score externally corroborated details, event/source binding, unsupported additions, and uncertainty separately from narrative richness. Rich prose alone is easy to manufacture. Counterbalance probe order and measure changes caused by earlier probing, since an interview can construct or modify the material later used as a target. A successful system would reproduce assessed retrieval behavior; it would not thereby reproduce subjective re-experiencing or prove preservation of identity.
Reading and acquisition record
Read the complete introduction, methods, results, discussion, conclusions, tables, and figure captions from the published 18-page PDF's extracted text (main paper pp. 479–493). Appendix A's illustrative interview was only inspected at its opening, not fully read. Bibliography entries were not followed individually. No separate supplement was identified. Brain images and plotted values were not independently remeasured; no raw data or reanalysis was attempted.
The publisher-formatted PDF is publicly hosted by an author's institution and states Elsevier copyright, all rights reserved; no open reuse license was identified. Local research retention does not imply redistribution permission. Accessed 22 September 2026.
Local sources: PDF, readable text, provenance and SHA-256 hashes. The text uses two-column PDF layout; consult the PDF when column order is ambiguous.