Home > Positioning > Persons > Tulving

Endel Tulving (1927–2023)

In Tulving’s hands memory stopped being one thing. His 1972 distinction cut what had been a single faculty in two: episodic memory — the remembering of particular experienced events, with their time, place, and the self who was there — and semantic memory — knowledge extracted from experience and freed of its occasion: that Paris is the capital of France, remembered by people who cannot recall any moment of learning it. The two came apart, on his argument, in what they store, how they fail, and what kind of awareness retrieval brings with it. He then kept cutting: the mature framework is a taxonomy of five systems, and it gave the psychology and later the neuroscience of memory its working vocabulary — contested at its joints from his day to this.

Life

Born 1927 in Petseri, Estonia. The war broke his family in two: sent west in 1944, he did not see his parents again for two decades, finishing his schooling in Germany and emigrating to Canada in 1949. PhD at Harvard (1957); from 1956 he was at the University of Toronto, where he remained for the whole career, latterly at the Rotman Research Institute. The 1972 chapter “Episodic and semantic memory” made the distinction; Elements of Episodic Memory (1983) is the systematic statement. Officer of the Order of Canada, Fellow of the Royal Society, and winner of the Gairdner International Award — by the end of his life the distinction he drew was standard vocabulary across psychology, neuropsychology, and brain imaging.

The architecture of memory

Episodic and semantic. The founding observation is that knowing and remembering are different achievements. Semantic memory serves up what one knows with no experience attached; episodic memory re-presents a particular past happening as one’s own past. Tulving later sharpened the difference into kinds of consciousness: retrieval from semantic memory is noetic — knowing — while episodic retrieval is autonoetic — a self-knowing awareness in which the rememberer mentally travels back to the event. The remember/know experimental paradigm operationalised the difference, and mental time travel became the formulation under which the distinction is now widely known: episodic memory as the backward half of a capacity whose forward half is imagining one’s future. Why one system should do both is the question the constructive turn answered — on Daniel Schacter and Donna Rose Addis’s constructive episodic simulation hypothesis, episodic memory is a recombination engine, storing the past in pieces so that the pieces can be flexibly reassembled, for remembering and for simulating futures alike — an account that also reads the system’s characteristic errors as the cost of its flexibility, and that carries Bartlett’s reconstruction into the very system Tulving named.

The full taxonomy. The two-way distinction is where the cutting began, not where it stopped. Tulving’s mature framework holds five systems — procedural, the perceptual representation system, semantic, primary (working) memory, and episodic — arranged in the SPI model: encoding is serial (information enters higher systems through lower ones), storage is parallel, retrieval is independent. The monotonic arrangement, with episodic memory at the top depending on all the rest, is a stronger and more contestable claim than the original binary, and he made it deliberately.

Encoding specificity. With Donald Thomson he established that remembering is a joint product of the trace and the cue: what is retrievable depends on the relation between how an event was encoded and what the retrieval situation provides — a principle under which forgetting can be cue failure rather than trace loss, and a “stored” memory is only half of what a remembering needs.

The evidence of dissociation. The case that episodic and semantic memory are different systems leaned on their coming apart. The frame was set by the amnesia literature Tulving inherited — above all H.M., whose surgery established the medial temporal lobes as the ground the disputes still run through. Tulving’s emblematic case was the patient K.C., who after a motorcycle accident retained his general knowledge — including knowledge about his own life — while losing entirely the ability to re-experience a single personal event, past or future. Faraneh Vargha-Khadem’s developmental amnesia cases cut from the other side: children with early hippocampal damage whose episodic memory never developed but who acquired language, schooling, and semantic knowledge in the normal range — acquisition running without the system it was supposed to depend on. Brain-imaging work followed, including Tulving’s own HERA proposal on the hemispheric asymmetry of encoding and retrieval.

The live disputes. The deepest dispute is whether systems are the right unit at all. Fergus Craik and Robert Lockhart’s levels-of-processing framework (1972), and Henry Roediger’s transfer-appropriate processing after it, hold that memory phenomena are explained by what the mind does at encoding and retrieval — depth of processing, match between study and test — not by which store an item sits in; the processing view ran alongside Tulving’s programme for decades as its standing alternative, and encoding specificity is claimed as evidence by both sides. Within the systems view, Larry Squire’s declarative-memory framework treats episodic and semantic memory as two expressions of one medial-temporal system, reading the dissociations as differences of degree; the multiple-systems and single-system readings of the same patient data remain in open argument. And Tulving’s claim that autonoetic episodic memory is uniquely human — evolutionarily recent, absent in other animals — met Nicola Clayton’s scrub-jay experiments, in which birds remember what they cached, where, and how long ago. Tulving’s reply was that “episodic-like” what-where-when memory does not establish autonoetic awareness; the constructive account sharpens the question from both sides — if episodic memory is a recombination engine, the test becomes whether an animal can flexibly recombine its past, not whether it can report an awareness — and that is where the argument now stands.

Where Tulving stops

The framework is a taxonomy with a phenomenology — kinds of memory, distinguished by what they store and the awareness their retrieval carries — and Tulving kept its claims at that level. How the systems are realised, and what mechanism turns an experienced event into semantic knowledge freed of its occasion, are questions the framework poses rather than answers: the interplay he named — episodic memory feeding semantic memory, semantic scaffolding episodic — is a standing research programme, not a settled account. His own criterion for the distinction’s core, autonoetic consciousness, is also the piece hardest to carry across the boundaries where the disputes run — into other species, into patients, into imaging — since it is identified in humans by report, and the framework does not say what stands in for report where report is missing.

Key works

See also: Bartlett · Baars · Edelman