You and your sister both remember the day the dog died. You remember it clearly. You were in the back garden, holding him when he died. You can still feel his weight. You have told that story for twenty years.
Your sister says you were not there.
You were away at camp. She was the one in the garden. She telephoned you that night, and you cried down the line from two hundred miles away. She remembers the call. She remembers your voice.
One of you is remembering something that did not happen — not a small detail, but the whole scene. And here is the difficulty: from the inside, there is nothing to tell you which of you it is. Your memory of holding the dog feels exactly as real as any memory you are sure of. It has the same detail, the same certainty, the same ache. Nothing about the memory itself marks it as invented. Nor does anything about hers.
This is Philosophy for Us — philosophy for everyone, no degree required.
This is the first of eight episodes in a series called The Self Across Time. All eight circle a single question, and it is one most of us assume we have already answered: whether the "you" at the centre of your life is one thing that lasts. The same person who was once a child. The one who will be old. The one you save money for, apologise on behalf of, and make promises for. We treat that person as a single continuous thing, extended through time, with ourselves in the middle of it. Over these eight episodes we will ask whether that thing is really there. And if it turns out to be less solid than it feels, we will ask a harder question: what do you owe the person who will be living your life in thirty years — a near-stranger who happens to carry your name? That is where the series ends. We begin at the nearer end — not with the self, but with the one record the self depends on most heavily: memory.
So return to the dog. The subject today is not the dog. It is the memory.
Most of us treat memory as a kind of recording. Something happens; the mind captures it; later we play it back. The recording may be patchy — worn in places, with gaps — but what remains is what was there. This picture runs so deep that we rarely notice we hold it. And much rests on it: who we take ourselves to have been, the story we tell about our own lives, and the plain sense that we are the same person now as then, because we can remember being that person.
I am not setting out to teach a theory about this. I do not speak from some settled position above the problem. I have spent longer than you have with these arguments, and I will say where I come down — but I live inside the problem as you do. The memories I would defend most fiercely are made of the same material as yours.
So today we do one thing. We take the record most of us trust without question — our own clear memory of our own life — and we find out what it is made of. Not the self that holds the memories; not yet. Only the record. We examine the record first, because it is the one part of this we can actually test.
Let us give the recording picture its due, because it is not a foolish view. It is the most natural thing in the world to believe, and much of the time it serves us well.
Here is the ordinary experience beneath it. We each experience ourselves as one continuous person. There is a single "I" that was once small, that is sitting wherever you are sitting now, and that will grow old. And what joins those moments into one life — what makes the child in the photograph you, rather than some unrelated child — is memory. You can reach back and find those earlier moments. You remember being five. You remember a first day at work, a face, a kitchen. That capacity to reach back is what makes the line continuous. Remove it, and the moments no longer hang together as one person's life.
So memory is doing a lot of work here. It is not merely one feature of the mind among others. It is what holds the self together across time. And the reason it can do that work — the reason it feels equal to it — is that we trust it to be a record, and a faithful one. You do not remember deciding what your kitchen looked like. You simply bring the memory up and it is there, the way a photograph is there when you turn it over. It feels found, not made — given to you, not built by you.
And in fairness, most days it earns that trust. You remember where you live. You remember the faces of the people you love. You remember how to do your work. If memory were not broadly reliable, you could not cross a road. So the claim here is not that memory is worthless. It is not. It is good enough to run a life on, and that is precisely why the recording picture is so hard to give up.
But consider what the recording picture asserts. A recording has one defining property: it is fixed at the moment it is made. It captures the sound in the room, and afterwards nothing that happens can reach back and alter what was captured. The recording can be lost, or allowed to degrade, but a later conversation cannot change the concert it holds. On this view, the past, once recorded, is closed.
That is the strong claim contained in "memory is a recording" — not merely that memory is usually accurate, but that a memory of an event was set by the event itself, and is therefore protected from everything that came afterwards: every later conversation, every photograph seen, every retelling. On this view the memory is something that happened once and now sits in storage, unchanged, waiting to be retrieved.
Here is the difficulty. That is not how the process works. When we remember something, we are not opening a file and pressing play. Each time a memory is brought up, the brain reconstructs it — assembles it again, on the spot, out of fragments, habits, and expectations. And when we are finished, it is laid down again, slightly altered by the act of having handled it. Remembering is less like playing a recording and more like retelling a story told many times before: a little smoother with each telling, a little more confident, and reworked, each time, by everything learned since.
If that is right, the recording picture predicts something false. It predicts that the past is closed. So there is a test, and it is a clean one. If memory really is a recording, then information arriving after the event should not be able to enter the memory and change what is remembered. The concert is already recorded. Talking about it later should not move it.
So let us see whether it moves.
In the early nineteen-seventies, the psychologist Elizabeth Loftus, working with John Palmer, ran a study that has troubled people ever since. It is simple, and that is part of why it is so hard to escape.
They showed people short films of car accidents — ordinary footage, a few seconds long, two cars. Afterwards they asked the viewers a question: about how fast were the cars going when they hit each other?
But not everyone was given the word "hit." They varied it from person to person. Some were asked how fast the cars were going when they contacted each other. Some were given hit. Others were given bumped, or collided, and some were asked how fast the cars were going when they smashed into each other. The same film, the same accident — one different verb.
The estimates varied with the verb. Those asked about the cars that smashed gave figures around forty miles an hour. Those asked about the cars that hit gave around thirty-four. Those asked when the cars contacted gave lower figures still. No one had seen a speedometer. They were reaching for a number, and the word in the question shaped the number they produced. Smashed is a more violent word, and the accident they remembered became more violent to match.
Now, one might set that aside. Perhaps people simply give higher figures when a dramatic word is used — an adjustment in the answer, not a change in the memory itself. The footage, after all, is still the footage. So Loftus and Palmer ran the study again, and this time they waited.
They showed the film, asked the speed question with the loaded verb, and then sent everyone away. A week later they brought the same people back and asked a new and innocent-looking question: did you see any broken glass?
There was no broken glass in the film. None. The accident had broken none.
Those who had been asked, a week earlier, about the cars that smashed were more than twice as likely to say yes — yes, they had seen broken glass. Nearly a third of them remembered glass that had never been there. Among those given the milder verb, far fewer did.
This is more than a mistaken estimate. A week had passed. They were no longer guessing at a speed; they were reporting what they had seen. And what some of them now saw, in their own memory of their own experience, was broken glass on the road. The memory had acquired a detail. And the only possible source of that detail was the word smashed — a single word, placed in a question, a week earlier. Something that occurred after the event had entered the memory of the event and added something that had never been in it.
Return to the test from a moment ago. A recording is fixed when it is made, and nothing later can change what it holds. Here, what it held was changed. It was changed with one word, after the fact, and the people whose memories had been altered had no idea it had happened. To them the glass was simply there — part of what they had seen, as real as the cars.
So the first claim falls. Whatever memory is, it is not a fixed recording. The past, as we hold it, is not protected from the present. Later information can reach it and revise it.
But perhaps you are thinking: all right, a single detail. A piece of glass. A stray fact can be pushed into the edge of a real event. That is a long way from inventing the whole thing. No one talked these people into a crash that never occurred. They did see a crash; only the edges were disturbed.
Fair enough. So let us find out how far in this goes.
A single detail is one thing; a whole event is another. So Loftus asked the harder question: could a person be given the memory of something that had never happened to them at all? Not a colour, not a piece of glass — an entire afternoon.
Here is the method. No one is hypnotised. No one is pressured. You go to the person's family — a parent, an older sibling — and collect a few true stories from their childhood, real ones, with real detail. You write those stories up in a short booklet, and among them you place one more that the family confirms never took place. In the original study the invented story was always the same: at about five years old, the person had become lost in a shopping mall, had been frightened, had cried, and had eventually been found by an older person and returned to the family.
It never happened. The family confirms it never happened. But there it sits in the booklet, beside three events that did happen, recorded in the same plain language of fact and vouched for by the same people who would know.
Then the participants are simply asked to remember. Read these over. Tell us what you recall of each. Take your time; we will talk again in a few days.
About a quarter of them come to remember being lost in the mall.
Not "I suppose that might have happened." They remember it. They fill it in. They add details never supplied to them — what the place looked like, what the person who found them was wearing, how frightened they were, what their mother said on their return. They construct the false afternoon out of their own minds. And when they are later told that one of the four stories was invented, some cannot identify which. The planted afternoon does not stand out as the false one.
These were not credulous people, and they were not lying. They were doing what any of us does in remembering — reaching back, locating the event, reporting what is there. The reaching back worked as it always works. It returned a memory: a place, a feeling, a small sequence of events. And at the far end of it there was nothing.
I should be exact about what the study found, because it also cuts the other way. On average, these planted memories came back somewhat thinner than the real ones — reported with rather less detail, and held with rather less confidence. And you may want to seize on that as an escape. Perhaps a false memory is fainter. Perhaps, with attention, one could feel the difference: the true memories vivid, the invented ones dim. That would be reassuring. If certainty were a reliable signal — if the vivid memories were the true ones — then the vivid ones could be trusted, and the problem would quietly resolve.
So that is the question on which everything turns. Return for a moment to the garden: you, holding the dog, feeling his weight. That memory is not faint. It is as vivid and as certain as anything you possess. Does the vividness of a memory actually indicate whether it happened? If vividness were a mark of truth, you would be safe. If it is not, you have nothing.
We have already come a long way. Not a stray detail added to a real event, but a whole event, with nothing behind it, that a person will furnish, believe, and defend as their own.
And here is the objection you ought to be raising, because it is the right one. This was a laboratory. Someone came in and deliberately interfered — handed people a planted story and asked them to remember it. Of course memory bends when an experimenter is bending it. But that is not ordinary life. No one is sitting you down with a forged booklet about your childhood. Your real memories, out in the world, with no one tampering with them — surely those are secure.
So let us take the experimenter out of the room. And let us go directly to the most vivid, most certain memories people have — the ones no one would think to question — and see how they hold up.
There is a kind of memory we all treat as unshakeable: where you were when you heard the news. Not ordinary news, but the news that halts the day — an attack, a death, a disaster. People who have lived through such a moment will tell you they remember it with complete clarity: where they were standing, who told them, what they were holding. These are the memories we trust most. They feel permanent. Psychologists gave them a name — flashbulb memories — on the idea that the moment goes off like a camera flash and fixes the whole scene at once.
No laboratory is involved here. Nothing is planted. There is only a person, their own life, and the most vivid memory they hold.
In nineteen-eighty-six the space shuttle Challenger broke apart shortly after launch, on live television — the kind of event that fixes itself in people's minds. The morning after, the psychologist Ulric Neisser, working with Nicole Harsch, did something rarely possible. He had a room of students write down, that day, how they had heard the news: where they were, what they were doing, who told them. The accounts were fresh — one day old.
Then he set those accounts aside and waited.
Two and a half years later he brought the same students back and asked the same question: how did you hear about Challenger? And they told him — confidently, in detail, with feeling. Many were certain. This was, they felt, one of those memories one does not lose.
A large number of them were wrong. Not vague — wrong. Entirely different accounts. One student who had written, the morning after, that she heard the news in class from another student now recalled, with complete conviction, that she had been in her dormitory room watching television when the news broke. A different place, different people, a different way of learning it. And she had no sense that anything had changed. To her, this was simply the memory — vivid, certain, hers.
And here is the finding that does the real damage. Neisser showed some of them their own accounts from the day after — their own handwriting, written when the event was one day old. Some looked at it and could not make it feel true. One student looked at her own writing and said, in effect: that is my handwriting, so I must have written it, but that is not how it happened. She trusted her present memory over a document she had written herself when the event was a day old. The present memory prevailed over the written evidence, in her own hand.
Now set that against what the students felt. They were not less certain of the wrong memories. If anything, they were more certain of the vivid ones. Confidence was high throughout, and confidence did not track accuracy. The students who were certain were not the students who were correct. The feeling of remembering something perfectly turned out to be exactly that — a feeling — and it rested just as firmly on the false memories as on the true ones.
So much for that hope. A short while ago we allowed ourselves to hope that a false memory might at least feel fainter — that the vivid ones could be trusted because they are vivid. These are the most vivid memories people have: where they were when they learned of a disaster. And the vividness guaranteed nothing. A memory can be full of detail and certainty and still be of a room one was never in. Vividness is not a sign of truth. It is only vividness.
So we take the experimenter out of the room, and nothing improves. The experimenter was never the source of the trouble. The laboratory merely accelerated, and made visible, something memory does on its own, quietly and continually — reconstructing the past, smoothing it, updating it to fit the person one now is, and returning it marked certain. The booklet in the mall study did not introduce a flaw. It borrowed one already present.
So consider where we have arrived. Put plainly:
You hold within you a great many memories. Some are records of things that happened. Some — as we now know — could be records of nothing, built in a few days out of a suggestion, or reworked over years while no one interfered, and they would feel no different. And there is no internal mark that distinguishes the two. No stamp, no watermark, no inner voice that speaks only for the true ones. The true memory and the manufactured memory arrive in the same form.
Try it yourself, with the most solid thing you have. Choose a memory you would stake your life on — a morning, a person, a room you can still walk through. Now ask: how do I know this one is real? And notice what you reach for. You reach for how clear it is, how much detail it holds, how certain you feel. Those are the three things you would offer as proof — and those are exactly the three things that failed today. People built whole afternoons that never occurred and furnished them with their own detail. And the most vivid, most certain memories anyone holds — where they were when they heard the news — proved wrong as often as right, the certainty resting just as firmly on the false ones. Clarity, detail, the feeling of certainty: not one of them indicates whether the event happened. So none of them can serve as your proof. They appear either way.
What would actually settle it? Nothing inside the memory. You would have to go outside it — find a photograph, ask someone who was present, locate a document. Something in the world, not in your head. And notice what that concedes. For a memory to be checked, it must be checked against something that is not the memory — which means the memory, on its own, was never the witness you took it to be. It was a claim awaiting evidence all along, and you had been treating it as the evidence.
That is the conclusion to carry away, and it is the central finding of today. The recording is not a recording. The record most of us trust above all others — our memory of our own life — is not a sealed capture of what happened. It is something the mind builds, and rebuilds, and can be made to build out of nothing, and from the building itself there is no telling which is which.
Be careful, though, about what this does and does not take away. It does not say your memories are mostly false. They are not; you could not live if they were. It does not license doubting everything. What it removes is narrower and harder to set down. It removes your right to be certain of any particular memory — your right to say "I remember it clearly, therefore it happened." That step is no longer available. You may keep your memories. What you cannot keep is the sense that they vouch for themselves.
We have established one thing today, and one only: the record. We have left something entirely untouched — the one who keeps it. There is still, you will have noticed, a you in all this: the one who holds these memories, true or false, the one doing the remembering. Today that self stood to one side, untouched, watching its own record prove unreliable.
Next time we turn to it. There is a famous and elegant answer to what that self is. It comes from John Locke, writing at the end of the seventeenth century, and the answer is this: you simply are your memories. What makes you the same person as the child in the photograph is neither your body nor your soul, but that you can remember being her. What links you to her is memory. It is one of the most liberating ideas anyone has proposed about the self. And you already know the question we will carry in with us. If the link is memory — and memory is the thing we have spent today taking apart — then what is holding the self together?
So the next time you hear yourself say I remember it perfectly — about an argument, about your childhood, about who said what — stop on the word perfectly. Because a false memory says it too, and says it just as loudly. And from where you stand, on the inside, looking at the memory and nothing else, there is no way to tell which of the two you are holding.
Thanks for listening. I'll see you next time.