The Future of Memory: Remembering, Imagining, and the Brain

Type: paper Slug: the-future-of-memory-remembering-imagining-and-the-brain—hassabis Sources: the-future-of-memory-remembering-imagining-and-the-brain—hassabis Last updated: 2026-05-13

---

Summary

Schacter, Addis, Hassabis, Martin, Spreng, and Szpunar (2012) published a major review in Neuron synthesizing five years of research since the 2007 “mental time travel” breakthrough. The review makes four key arguments: (1) temporal and nontemporal factors must be distinguished in memory-imagination studies, (2) important differences between remembering and imagining have emerged despite their similarities, (3) the default network’s component processes are being identified (scene construction vs. self-reflection subsystems), and (4) the default network flexibly couples with the frontoparietal control network for goal-directed simulation.

Core content

Structure: Organized around four key points, with concluding sections on emerging directions (paper—the-future-of-memory-remembering-imagining-and-the-brain §Introduction).

Point 1 — Temporal vs. nontemporal factors:

• Early studies conflated “remembering the past” with “imagining the future,” but imagined future events could reflect imagination independent of temporal projection (paper—the-future-of-memory-remembering-imagining-and-the-brain §Understanding the Relation).
• Evidence for nontemporal view: Addis et al. (2009a) found nearly identical default network activity for past and future imagined events; de Vito et al. (2012a) found no subjective/objective differences between future and atemporal imagined episodes (paper—the-future-of-memory-remembering-imagining-and-the-brain §Evidence for a Nontemporal Perspective).
• Evidence for temporal view: Parkinson’s patients impaired for future but not atemporal imagining (de Vito et al., 2012b); older adults show greater deficits for future than atemporal scenarios (Rendell et al., 2012); Nyberg et al. (2010) isolated chronesthesia to left lateral parietal and frontal regions (paper—the-future-of-memory-remembering-imagining-and-the-brain §Evidence for a Temporal Perspective).

Point 2 — Differences between remembering and imagining:

• Greater hippocampal and frontopolar activity for future > past during construction phase (Addis et al., 2007); left anterior hippocampus selectively responsive to future event detail (Addis & Schacter, 2008) (paper—the-future-of-memory-remembering-imagining-and-the-brain §Evidence for Differences).
• Remembered events contain more sensory-perceptual detail; imagined events require more constructive operations (paper—the-future-of-memory-remembering-imagining-and-the-brain §Evidence for Differences).
• Neuropsychological dissociations: semantic dementia patients impaired at future simulation but not past recall (Irish et al., 2012); prefrontal lesion patients impaired at constructed experiences but not autobiographical memory (Berryhill et al., 2010) (paper—the-future-of-memory-remembering-imagining-and-the-brain §Evidence for Differences).
• Debate over whether hippocampal damage alone causes imagination deficits: Squire et al. (2010) vs. Hassabis et al. (2007b) and Maguire et al. (paper—the-future-of-memory-remembering-imagining-and-the-brain §Evidence for Differences).

Point 3 — Component structure of the default network:

• Hassabis et al. (2007a) decomposed the network into scene construction (hippocampus, PHG, retrosplenial, posterior parietal, vmPFC) and self-reflection (amPFC, PCC, precuneus) subsystems (paper—the-future-of-memory-remembering-imagining-and-the-brain §Component Structures).
• Andrews-Hanna et al. (2010b) confirmed an MTL subsystem (scene construction) and dMPFC subsystem (affective self-referential processing) with resting-state and task-based fMRI (paper—the-future-of-memory-remembering-imagining-and-the-brain §Component Structures).
• Repetition suppression studies linked MPFC to social scenarios, inferior frontal/premotor to objects, PCC/retrosplenial/parahippocampal to locations (paper—the-future-of-memory-remembering-imagining-and-the-brain §Component Structures).

Point 4 — Flexible network coupling:

• Spreng et al. (2010): autobiographical planning engaged default + frontoparietal control networks; visuospatial planning (Tower of London) engaged dorsal attention + frontoparietal control (paper—the-future-of-memory-remembering-imagining-and-the-brain §Default Network Can Couple Flexibly).
• The frontoparietal control network acts as a flexible hub, coupling with default or dorsal attention networks depending on task demands (paper—the-future-of-memory-remembering-imagining-and-the-brain §Default Network Can Couple Flexibly).
• Older adults failed to suppress default network during visuospatial planning and failed to decouple it from the frontoparietal control network (Spreng & Schacter, 2012) (paper—the-future-of-memory-remembering-imagining-and-the-brain §Default Network Can Couple Flexibly).

Emerging directions: Episodic simulation reduces temporal discounting (Benoit et al., 2011; Peters & Büchel, 2010); semantic memory contributes scaffolding for future thinking; counterfactual thinking shares neural substrates with future simulation; Monte Carlo tree search in AI as a parallel to mental simulation (paper—the-future-of-memory-remembering-imagining-and-the-brain §Concluding Comments).

Connections- Theme: theme—hippocampal-construction, episodic-memory, theme—memory-imagination, scene-construction

• Collaborators: Daniel L. Schacter (first/senior), Donna Rose Addis, Victoria C. Martin, R. Nathan Spreng, Karl K. Szpunar
• Era: deepmind-era — Hassabis’s only neuroscience review as senior author at DeepMind
• Venue: venue—Neuron (NOT Science as metadata states)
• Synthesizes: paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory, paper—deconstructing-episodic-memory-with-construction, paper—the-construction-system-of-the-brain, paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences
• Precedes: paper—imagine-all-the-people-how-the-brain-creates-and-uses-personality-models — unpublished data cited in this review
• Open questions: Why do some hippocampal patients show dissociated memory/imagination deficits while others show parallel deficits? What is the precise hippocampal subfield contribution to imagination?

Honest Gaps

• Metadata incorrectly lists venue as Science; published in Neuron 76(4):677–694.
• Metadata incorrectly lists year as 2007; published November 2012.
• As a review, contains no original data — all findings are second-order interpretations of others’ work.
• The Monte Carlo tree search parallel to mental simulation (§Concluding Comments) is a brief analogy rather than a substantive computational claim.
• Review focuses exclusively on human studies; non-human animal work on future planning (e.g., scrub jays) is acknowledged but not reviewed.
• Several findings cited as “unpublished data” (including the Imagine All the People study) may have been subsequently modified in publication.