2.1.3 - Episodic and semantic memory

Remembering events and knowing facts

Tulving’s 1972 distinction concerns the kind of information remembered, rather than simply how long a memory lasts.

An episode is a particular occurrence: the time you missed a train, where you were standing and what happened next. Its content connects the event with a personal time and place. In Tulving’s later development of the theory, recollecting an episode includes the subjective experience of mentally returning to one’s past. This is different from merely knowing that an event happened.

Semantic memory lets you understand that trains run on tracks, what “platform” means or which city is the capital of France. You can use that knowledge without recalling the specific lesson, conversation or encounter in which you learned it. A fact acquired during a personal experience does not remain episodic merely because it came from experience.

Both kinds are long-term memories and can usually be expressed consciously. The difference is not “episodic equals short-term” or “semantic equals schoolwork”. Knowing a neighbour’s name may be semantic; recollecting the classroom moment in which you first understood a concept may be episodic. Nor is emotion a defining requirement: an episode can be ordinary, while factual knowledge can be emotionally important.

How events and knowledge work together

Distinguishing two kinds of memory does not mean that they operate in isolation.

Imagine your first visit to a planetarium. An episodic memory may include sitting beside a friend, seeing the ceiling display and hearing the presenter’s explanation. Semantic memory supplies the meanings of “planet” and “orbit”, enabling you to understand the experience. You may later know that a planet orbits a star while no longer remembering when you learned it. Repeated encounters can establish knowledge that is available independently of one particular learning episode.

The same event can therefore contribute both an episode and factual knowledge. Classification depends on what the person is retrieving. “I know what the presenter explained” does not establish that the person re-experiences the visit; conversely, recollecting the visit does not guarantee that every scientific fact from it was learned correctly.

This distinction has practical consequences for interpreting a memory difficulty. Someone who struggles to recollect a recent family visit may still know relatives’ names and relationships. A conversation can draw on preserved knowledge rather than assuming that every kind of memory has failed. That is an illustrative implication of the distinction, not a diagnosis or a promise that any particular form of support will work. Individual assessment is necessary because impairments vary.

It also refines the multi-store model. Describing all these representations as “long-term memory” identifies their enduring availability but says little about the differences within that store. Tulving’s distinction gives a more detailed question: can an event-recollection function be affected while factual knowledge remains available? It supplements an explanation of short-term versus long-term memory rather than replacing the need for temporary processing.

Evidence and the limits of the distinction

Uneven memory impairment supports functional differences, but an observed difference does not automatically establish completely independent brain systems.

Tulving’s later primary account describes K.C., a man with extensive brain damage following a motorcycle accident. Researchers compared recollection of personal experiences with factual knowledge. K.C. could report many facts about his earlier life and the world but could not recollect past personal events. His everyday new learning was also severely impaired: it would be wrong to say all semantic learning was normal. In structured experiments, repeated teaching nevertheless allowed some new factual knowledge to be acquired without recollection of the learning occasions. This shows a difference between knowing information and re-experiencing its acquisition, within the same person.

This is a dissociation: one function is much more impaired than another. It challenges the view that all long-term remembering is one uniform ability. Retaining facts alongside severe event-recollection difficulty also makes it less plausible that the person merely lacks effort or all understanding of the questions. The value lies in the particular pattern, not in the claim that one patient proves the entire theory.

There are important limits. K.C.’s injury affected multiple brain regions. The case cannot identify one isolated site for episodic memory, and one unusual patient does not establish the pattern in everyone. Earlier factual knowledge may also have been repeated and practised more often than particular episodes. Unequal learning histories could partly explain unequal survival, so controlled comparisons of learning and retrieval are valuable. The structured new-learning evidence helps address this issue, without establishing that every influence was removed.

A further challenge is measurement. Researchers can observe a correct answer, but whether the person consciously recollects an event partly depends on a report of subjective experience. A correct factual response alone does not reveal whether it came from recollection, general knowledge or a strategy. Different task instructions, cues and difficulty can create performance differences as well. Strong evidence combines carefully designed tasks with clinical and brain-activity findings, rather than equating a task label with a pure memory system.

The distinction is consequently valuable at two levels: it describes a meaningful difference in what remembering feels like and concerns, and it generates tests of whether the underlying functions are separable. Evidence supports distinguishable but interacting functions. It does not require us to imagine two sealed containers, two single brain locations, or an event memory free of knowledge about the world.