Using Imagination to Understand the Neural Basis of Episodic Memory

Type: paper Slug: using-imagination-to-understand-the-neural-basis-of-episodic-memory—hassabis Sources: using-imagination-to-understand-the-neural-basis-of-episodic-memory—hassabis Last updated: 2026-05-13

---

Summary

Hassabis, Kumaran, and Maguire (2007) used a novel fMRI paradigm in which subjects recalled real episodic memories, retrieved previously imagined fictitious experiences, and constructed new fictitious scenes. A conjunction analysis revealed a core “scene construction” network (bilateral hippocampi, parahippocampal gyrus, retrosplenial cortex, posterior parietal cortex, vmPFC) common to all three conditions, while anterior medial PFC, posterior cingulate, and precuneus were selectively engaged for real episodic memories, supporting self-schema and familiarity processes.

Core content

Research question: Can imagination of fictitious experiences be used as a comparison task to deconstruct the episodic memory network into its constituent processes?

Design: 21 participants completed 7 conditions (3 scene: real memory recall, imagined scene recall, new scene construction; 3 object controls; baseline). Scenes were contrasted against context-free object visualization to isolate spatial scene construction from simple imagery (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory §Materials and Methods).

Key findings:

• Conjunction of RM>RO, IS>IO, and NS>NO revealed the core scene construction network: bilateral hippocampi, PHG, retrosplenial cortex, posterior parietal cortex, vmPFC (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory §Commonalities).
• (RM−RO)>(IS−IO) isolated episodic-memory-specific regions: anterior medial PFC (BA10), posterior cingulate cortex, and precuneus — linked to self-schema, familiarity, and mental time travel (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory §Differences).
• Precuneus showed a graded response: highest for real memories, intermediate for recalled imagined scenes, lowest for new constructions — consistent with a familiarity signal (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory §Differences).
• Objects>Scenes activated lateral occipital cortex and intraparietal sulcus, dissociating simple object imagery from complex scene construction (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory §Results).
• The findings decomposed the classic episodic memory network into two components: a hippocampus-centered scene construction system and an amPFC/PCC/precuneus self-projection system (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory §Discussion).

Theoretical contribution: Proposed that the brain network commonly activated across navigation, future thinking, theory of mind, and the default mode may in fact reflect a general role for scene construction, with additional regions co-opted in a task-specific manner.

Connections- Precedes: paper—deconstructing-episodic-memory-with-construction (2007) — the theoretical twin paper

• Precedes: paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences (2007) — provides the neuropsychological complement
• Theme: theme—hippocampal-construction, episodic-memory, theme—memory-imagination, scene-construction
• Collaborators: Dharshan Kumaran, Eleanor A. Maguire
• Era: phd-period
• Venue: venue—Journal-of-Neuroscience (NOT Current Biology as metadata states)
• Influences: Buckner & Carroll (2007) self-projection theory

Honest Gaps

• Metadata incorrectly lists venue as Current Biology; the paper was published in The Journal of Neuroscience, 27(52):14365–14374.
• Metadata year is correct (2007).
• 1.5T scanner limits spatial resolution compared to the 3T and 7T used in later studies by the same group.
• The fictitious scene condition was explicitly designed to lack temporal and self-referential features, which is a strength for dissociation but limits ecological validity of the comparison.