Episodic vs semantic memory: knowing is not remembering
Ask someone what job they had three years ago and the answer comes in a second. Ask them to describe a specific Tuesday from that period and something shifts. They know the job. They know the city they lived in, the names of their colleagues, the rough arc of the year. What they cannot do, without extraordinary effort, is reconstruct the Tuesday.
The first question taps one kind of memory. The second taps another. The mind treats them as neighbors, and the language we use for recall does not separate them. Psychology has separated them for more than fifty years, and the separation is where most of this essay lives.
The 1972 distinction
In 1972, the Estonian-Canadian psychologist Endel Tulving wrote a chapter titled “Episodic and semantic memory” for a book he was editing with Wayne Donaldson. The chapter did something the field had not done cleanly before. It split long-term memory into two systems with different inputs, different outputs, and different failure modes.
Semantic memory is the system that stores general knowledge. It holds facts about the world (“Tokyo is the capital of Japan”), facts about yourself (“I was born in Nagoya”), and abstract information you acquired at some point and still know. Semantic memory is timeless in its form. When you recall that Tokyo is the capital of Japan, you do not recall the classroom where you learned it or the Tuesday you first heard the word. The fact is there, clean of its origin.
Episodic memory is the system that stores specific events. It holds moments you personally lived through, stamped with a time, a place, and a self that was present. When the system works well, an episodic memory is not a fact you know. It is a moment you can re-enter: the light of the room, the chair you were sitting in, the feeling of the conversation.
In his 2002 synthesis in Annual Review of Psychology, Tulving put the difference this way. Semantic memory gives you knowing. Episodic memory gives you remembering. A healthy mind uses both constantly, and the difference stays invisible in ordinary speech. In the lab, and in certain patients, it becomes sharp.
The case of K.C.
The sharpest illustration of the distinction came from a single patient. In October 1981, a 30-year-old Canadian named Kent Cochrane was riding his motorcycle home on a highway north of Toronto when he lost control at an exit ramp. The closed head injury he sustained left him cognitively intact on most measures of intelligence, language, reasoning, and general knowledge. It also left him with a specific, near-total loss that would make him one of the most studied subjects in the history of memory research.
K.C., as he was known in the published literature, could no longer retrieve a single episodic memory from his life. Before the accident or after. Not his wedding, not his brother’s funeral, not a vacation, not a conversation from that morning. His semantic memory, including knowledge about himself, was preserved. He could tell you he had a brother, that the brother had died, that he had worked at a particular factory. He could not recall anything about his brother as a person, or produce any scene in which the two of them had interacted.
Tulving and his colleagues spent three decades studying him. A 2005 review in Neuropsychologia by Rosenbaum and colleagues assembled what his case had contributed to memory theory. K.C. demonstrated, as clearly as any single patient ever has, that the two systems are dissociable. You can lose one and keep the other. The facts of a life can remain, while every lived moment of that life becomes inaccessible.
Asked in interviews what it was like to have no episodic past, K.C. reported that his mental state when trying to remember a specific event, and his mental state when trying to imagine a specific future event, felt identical. Both produced a uniform blankness. Tulving used this to make a further claim: episodic memory and the capacity to mentally project oneself into the future draw on the same underlying system. In a 1985 paper in Canadian Psychology titled “Memory and consciousness,” he named it autonoetic consciousness, the awareness of oneself as a continuous experiencer across time. Without it, a person knows their history the same way they know Tokyo is the capital of Japan. The knowing is accurate. It is also empty.
The quiet drift from one to the other
Most people are not K.C. Their episodic memory works. The useful question is what happens to their episodic memory over ordinary time, in the absence of any specific injury.
What appears to happen, across the autobiographical memory literature, is a slow conversion. The specifics of a remembered event fade, and what remains tends to harden into semantic form. A vacation that was once a set of scenes becomes a sentence: “we went to Hokkaido that summer, it was good.” A productive month at work that was once a sequence of specific afternoons becomes a summary: “March was the month we shipped.” A friendship that was once a collection of shared evenings becomes a relationship you know you have, without being able to produce any single evening.
The person is not forgetting. They still know. The transformation runs from episodic to semantic, not from present to absent. What they have lost is the ability to re-enter the moments. What they retain is the filed version.
This is part of what the reminiscence bump in autobiographical memory captures. Adults consistently show richer episodic retrieval for their late teens and twenties than for any other decade, including the decade they are currently living in. Rubin and Schulkind’s 1997 work in Memory & Cognition documented the pattern across large samples. The explanation is not that those years mattered more. The explanation is that those years generated more distinctive, separable episodes and fewer summaries. By adulthood, the summaries start winning. Life organizes itself into categories. The specific afternoons that would once have been scenes are compressed into the category, because the category is sufficient for almost every practical purpose.
The cost is easy to miss. You still have the facts of the year. You cannot re-enter any ordinary day of it.
Why summaries feel like memories
Some of the conversion happens silently. A meaningful share of it happens through the way people talk about their past.
Each time you tell someone about a period of your life, you produce a summary. The story of the year you lived in Kyoto is not a replay. It is a compression, tuned for a listener who does not want three hundred specific afternoons. Over years of re-telling, the compression becomes the memory. You remember the version you have told, not the version you lived. The Tuesdays that never entered the story get nothing keeping them alive.
This is where diary writing starts to matter, and also where it can quietly go wrong. A diary written as a reflective summary at the end of the day (“productive morning, lunch went well, felt tired in the afternoon”) produces the same compression the mind would have produced anyway. The form of the entry is semantic. It records knowing, not remembering. Re-read a year later, it yields a fact (“I had a productive day on March 14”) and very little else.
What survives as a cue back into episodic memory is different. It is the specific. A timestamped meeting with a specific person at a specific place. A commit message written at a specific hour. A check-in at a specific restaurant. A track played at a specific time of evening. Each of these is a coordinate in time and space, attached to a self that was present. Encountered later, they carry the shape of an episode, because that is the shape they had when they were recorded.
What a diary actually preserves
deariary assembles a daily entry from the traces your tools already produce: calendar events, messages, commits, check-ins, listens, completed tasks. The entries are not written as reflective summaries, because reflective summaries are what semantic memory already provides for free. They are written as timelines of what actually happened, in the order and at the times it happened.
This matters for the distinction Tulving drew. A semantic record of your life is easy. You can write one right now, from memory, for any month of the last decade. You know where you lived, what you were working on, roughly who you were spending time with. The knowing is intact.
An episodic record of your life is not easy. It requires a specificity that human memory stops producing for ordinary time about as fast as the day ends. The only way to have one is to let something outside the mind hold the specifics while they are still specific, so that the cue-to-episode link is still narrow enough to survive the drift.
The point of a diary, from this angle, is not to remember better than your brain does. Your semantic memory is fine. You already know the facts of your life. The point is to preserve the other kind of memory, the one that gives you back a Tuesday, because the other kind is the one that quietly converts into summary if nothing holds it in place.
K.C. showed what a life looks like when only semantic memory is left. It is a life a person can describe in full sentences and cannot re-enter at any point. He was a limit case. Ordinary adults are not at his extreme. They are on a gradient that moves in his direction each year, as the specific Tuesdays fade into the month, and the months into the year, and the years into the set of things they know they lived through.
A diary assembled from the traces of a day is not a better kind of remembering. It is a way of keeping episodic what would otherwise only stay semantic.
