MinedDeep

Patients with Hippocampal Amnesia Cannot Imagine New Experiences

3 min read

Patients with Hippocampal Amnesia Cannot Imagine New Experiences

Type: paper Slug: patients-with-hippocampal-amnesia-cannot-imagine-new-experiences—hassabis Sources: patients-with-hippocampal-amnesia-cannot-imagine-new-experiences—hassabis Last updated: 2026-05-13

Summary

Hassabis, Kumaran, Vann, and Maguire (2007) demonstrated that patients with bilateral hippocampal damage are markedly impaired at constructing richly imagined new experiences, despite intact semantic retrieval and normal subjective sense of presence. The critical deficit was spatial coherence — patients produced fragmented image collections rather than integrated scenes. This paper, published in PNAS and communicated by Endel Tulving, established the imagination-memory link that became the foundation for Hassabis’s construction system hypothesis.

Core content

Research question: Does the hippocampus play a necessary role in imagining new experiences, paralleling its established role in episodic memory recollection?

Method: 5 patients with bilateral hippocampal damage (mostly from limbic encephalitis) and 10 matched controls completed a construction task: 10 short verbal cues (beach, museum, pub, port, market, forest, castle, plus 3 self-relevant future scenarios) elicited free-form imagined experiences, scored on an experiential index (0–60) combining content, participant ratings, spatial coherence, and quality judgment (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Methods).

Key findings:

  • Patients scored significantly lower on the overall experiential index (mean 27.54 vs 45.06, P = 0.002) (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Results).
  • All four content subcategories were impaired: spatial references (P = 0.002), entities present (P = 0.003), sensory descriptions (P = 0.005), thought/emotion/actions (P = 0.001) (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Results).
  • Sense of presence and perceived salience were not impaired — patients felt they were “there” but produced sparse descriptions (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Results).
  • Spatial coherence index was the most dramatic deficit: patients scored 0.10 vs controls 3.68 (P = 0.007), with patients explicitly describing their experiences as “a collection of separate images” (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Results).
  • An “assisted” condition providing pictures, sounds, and smells did not improve performance (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Results).
  • One patient (P01) was unimpaired, potentially due to residual hippocampal tissue confirmed by fMRI (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Discussion).

Theoretical significance: Challenges the standard consolidation model (hippocampus time-limited) by showing that hippocampal damage impairs constructing novel scenes, not just retrieving old memories. Proposes that the hippocampus provides spatial context that binds disparate elements into coherent experiences — a mechanism that applies equally to remembering and imagining (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences §Discussion).

Connections- Theme: theme—memory-imagination, theme—hippocampal-construction — this is the foundational empirical paper for both themes

  • Theme: episodic-memory — challenges time-limited consolidation theory
  • Collaborators: Eleanor A. Maguire (senior author), Dharshan Kumaran, Seralynne D. Vann
  • Era: phd-period — conducted during Hassabis’s PhD at UCL
  • Precedes: paper—deconstructing-episodic-memory-with-construction (2007, BBS target article extending these ideas)
  • Precedes: paper—the-construction-system-of-the-brain (2008, full review articulating the construction system hypothesis)

Honest Gaps

  • The metadata in _extraction_raw.json lists venue as “Current-Biology” but the paper was published in PNAS (vol. 104, no. 5, pp. 1726–1731). The co-author list in metadata also omits Dharshan Kumaran and Seralynne D. Vann who appear on the paper.
  • The PDF extraction contains scanning artefacts (reversed header/footer text on each page) but the body text is intact.
  • The scoring was not fully blind — the primary scorer knew subject status, though a secondary blind scorer showed high inter-rater reliability.

Graph View

Backlinks