PhD Period (2007–2009)
Type: period Slug: period—phd-period Sources: decoding-neuronal-ensembles-in-the-human-hippocampus—hassabis, deconstructing-episodic-memory-with-construction—hassabis, imagine-all-the-people-how-the-brain-creates-and-uses-personality-models—hassabis, patients-with-hippocampal-amnesia-cannot-imagine-new-experiences—hassabis, the-construction-system-of-the-brain—hassabis, the-future-of-memory-remembering-imagining-and-the-brain—hassabis, tracking-the-emergence-of-conceptual-knowledge-during-human-decision-making—hassabis, using-imagination-to-understand-the-neural-basis-of-episodic-memory—hassabis, when-fear-is-near-threat-imminence-elicits-prefrontal—hassabis Last updated: 2026-05-13
Summary
Hassabis’s PhD at UCL under Eleanor Maguire produced 9 publications (2007–2009) that established the construction system hypothesis — the claim that the hippocampus does not store memories but actively constructs episodic scenes from component elements, and that this same mechanism supports imagination, future thinking, and conceptual abstraction. This period accounts for 4 of the 5 most-cited neuroscience papers in the corpus and generated the intellectual framework that later informed DeepMind’s approach to imagination-based planning and world models.
Core content
The amnesia-imagination discovery (2007): The foundational finding that patients with hippocampal amnesia cannot imagine new experiences (paper—patients-with-hippocampal-amnesia-cannot-imagine-new-experiences). Four patients with bilateral hippocampal damage were asked to imagine new scenarios; their descriptions were strikingly deficient in spatial coherence, sensory detail, and narrative continuity. This single experiment linked memory and imagination at a mechanistic level.
The construction hypothesis (2007): Published as a BBS target article (paper—deconstructing-episodic-memory-with-construction), this proposed that episodic memory is not retrieval of a stored trace but an act of reconstruction — the hippocampus recombines stored elements (people, places, objects) into scene-like representations. This directly challenged the standard “store and retrieve” model of episodic memory.
The construction system of the brain (2008): A Phil Trans B review (paper—the-construction-system-of-the-brain) that elaborated the neural architecture — hippocampus as scene constructor, vmPFC as schema integrator, posterior parietal cortex as spatial framework, temporal pole as semantic scaffold. This became the canonical reference for the framework.
Imagination-based memory testing (2007–2009): A series of papers using imagination tasks to probe hippocampal function: personality model construction (paper—imagine-all-the-people), imagination-based memory paradigms (paper—using-imagination-to-understand-the-neural-basis-of-episodic-memory), and a review of the memory-imagination link (paper—the-future-of-memory-remembering-imagining-and-the-brain).
Single-neuron evidence (2009): Direct recordings from epilepsy patients (paper—decoding-neuronal-ensembles-in-the-human-hippocampus) showing that individual hippocampal neurons encode specific concepts (people, objects) that could be recombined — neural-level evidence for the construction mechanism.
Conceptual abstraction (2009): fMRI evidence that the hippocampus tracks the transition from associative to conceptual decision strategies (paper—tracking-the-emergence-of-conceptual-knowledge-during-human-decision-making), extending the construction account beyond episodic memory into abstract knowledge.
Fear and threat (2009): The only non-hippocampal paper from this period (paper—when-fear-is-near-threat-imminence-elicits-prefrontal) examined prefrontal-periaqueductal gray shifts during threat imminence — a departure that foreshadowed later interest in hierarchical control systems.
Connections
- Theme: theme—hippocampal-construction, theme—memory-imagination, theme—episodic-memory
- Project: project—hippocampus-research
- Collaborators: Eleanor A. Maguire (supervisor, co-author on 5/9), Catriona D. Bruce (2 papers), Raymond Dolan (1 paper), Dharshan Kumaran (1 paper), Jennifer J. Summerfield (1 paper)
- Venues: venue—Current-Biology (4), venue—Behavioral-and-Brain-Sciences (1), venue—Philosophical-Transactions-B (1), venue—Science (1), venue—Neuron (1), venue—Proceedings-of-the-National-Academy-of-Sciences (1)
- Precedes: period—postdoc-period (single-paper continuation), period—deepmind-ascent (neuroscience-AI bridge papers pick up these themes)
- Key claim: The hippocampus is a construction system, not a memory store (paper—deconstructing-episodic-memory-with-construction)
Honest Gaps
- The metadata year for tracking-the-emergence was wrong (2006→2009) — suggests other dates in this period may need verification.
- No thesis document is in the corpus — the actual UCL PhD thesis (2009) would contain more detailed argumentation than the published papers.
- The fear/threat paper (when-fear-is-near) is an outlier thematically — its connection to the construction system is unclear and may reflect a side project.
- All papers are single-lab work from the Maguire group — no multi-site or large-scale collaborations in this period.
- The BBS target article (deconstructing-episodic-memory) likely had extensive commentary from peers that is not in the corpus.