The White Elephant

Picture a white elephant.

You did it instantly. Whatever detail you bothered with - the skin, the light on it, the ground it is standing on - arrived without effort. And you have almost certainly never seen one. There was no source. Your brain produced a complete image out of nothing and handed it over without a single mark on it saying this was invented.

That is the part worth sitting with. Not that you can imagine things. That the imagined thing does not feel any different in the hand.

The Same Machinery Runs Your Past

We talk about memory as though there is a file somewhere and remembering is opening it. Nothing like that is happening. There is no footage. What is stored is closer to a sparse set of facts, feelings and fragments, and when you go looking, the brain builds a scene out of them on the spot - filling the gaps with whatever is plausible, exactly the way it filled in the elephant.

The gaps get filled from everything else you know. Photographs you have seen since. A story an aunt tells at every gathering. What that room looked like in a later year. What you have decided, in the meantime, that the episode meant about you. All of it goes into the render, and none of it arrives labelled.

You are not playing back a recording. You are generating a plausible scene and experiencing the output as evidence.

The Camera Angle Problem

If you want proof that runs on your own hardware, go and fetch an early memory. Something from before you were ten. Now check one thing: where are you in the picture?

Most people, most of the time, can see themselves in it. Small, from across the room, from slightly above. From an angle their eyes have never once occupied.

That shot does not exist. It was never recorded, because you were behind your own face the entire time. The scene has been staged after the fact by something that needed a picture and built one. Psychologists have a name for it - observer memories, as opposed to the first-person kind - and the further back the memory goes, the more likely it is to have quietly switched to the outside camera.

One Afternoon That Was Probably Eleven

I have a childhood memory I trust more than almost anything else I know. A specific afternoon: the rain coming down hard enough to be a sound rather than a sight, a particular doorway, a particular person standing in it. It has a temperature. It has a smell. It is one of maybe six things I would have called certain.

It cannot have been an afternoon. Too many things in it belong to different years to sit in the same hour. What I have is not a day. It is eleven or thirty similar days, compressed into one representative scene with the details borrowed from across the whole range - and then the compression was so convincing that I filed it as an event.

This turns out to be the normal case rather than the failure case. The tells are easy to spot once you look for them:

• The memory has no beginning and no end. Real episodes come with dull edges - the walk there, the boredom after. Composites start on the good frame.
• The weather, the clothes and the ages don't quite line up with each other, or with the year the event has to have happened in.
• It has a narrative shape. It arrives already meaning something, which is a sign it has been retold rather than retrieved.
• There is a photograph of it. If a photo exists, that is very likely what you are actually remembering - the image, not the day.
• It is a story your family tells. Heard often enough, someone else's account installs itself as your own footage, and there is no internal marker to tell the two apart.

Every Recall Is a Rewrite

Here is the part that reverses the usual intuition. A memory does not sit untouched between visits. Bringing it up appears to make it briefly editable, and what gets put back down is the version you just constructed, including whatever today's mood and today's knowledge added to it.

Which means the memories you have returned to most often - the formative ones, the ones you would swear to - have been through the most edits. The ones you have not thought about in twenty years are, in a strange way, the better originals.

The memories you trust most are the ones you have rewritten the most. Confidence is a measure of rehearsal, not of accuracy.

This has been tested in the least comfortable way. Ask people where they were when they heard about some enormous public event, record the answer, come back years later and ask again. The answers change. People produce a different room, different company, a different way of finding out - and they are certain. When shown their own earlier account in their own handwriting, some of them do not believe it. Confidence and accuracy simply do not travel together, and we use confidence as the only gauge we have.

It runs the other way too. Imagine an event in enough detail, often enough, and it starts acquiring the texture of something that happened. The elephant, given time, gets a location and a date.

Why This Isn't Bad News

The first reaction to all this is usually a small panic, because if the archive is unreliable then the person assembled out of it is on shaky ground. I had that reaction for a while and I think it points the wrong way.

Consider what it actually buys you. The memory that has been quietly shaping how you behave since you were nine - the one that proves what you are like, what people do, what you should not attempt - is not evidence. It is a reconstruction, assembled by a child, re-rendered a few hundred times since by someone whose mood was not always neutral. It has the authority of a fact and the reliability of a rumour.

That is very close to something I have written about before, watching a thought long enough to notice it is a thought. The move is the same one. The moment a memory is held as a reconstruction rather than a recording, its grip loosens, and it stops being able to testify about you.

And there is a gentler reading of the composite. That afternoon that was really eleven afternoons is not a lie. It is a summary. The brain kept what those days had in common and threw away the rest, which is roughly what it is for. The specific rain was never the point. The specific doorway was never the point.

So: hold the visuals loosely, because the visuals are the part your brain is most willing to invent. Hold the meaning a little more loosely too, since you assigned it later and you can assign it again. What is left after both of those is small, and probably closer to true than the film you have been screening for yourself all these years.

Then picture the elephant one more time and notice that nothing in the experience tells you which of your pictures came from the world.