I begin this column on a train from Paris to London. Opposite me are a mother and baby. I don’t know them and will probably never see them again.

The baby is nine months old and called Gabriel. A genial and relaxed child, he is grinning at me and waving his soft-toy giraffe. He’s wearing bootees, white socks polka-dotted with little red hearts, pale burgundy trousers and a grey top. He seems uninterested in northern France flashing past our window, though his mother has held him up to look; but he is taken with this new stranger, your columnist.

And I reflect: is it not very odd indeed — does it not require explanation — that Gabriel will never remember any of this? It preoccupies him now. Why does a curtain come down between our memories, some very distinct and strong, from childhood after the age of three, and what happened to us before that age? Why does our infancy go into the dark, never to be illuminated again?

This, after all, was the time when the world was fresh and new and amazing, when misery was most miserable and happiness happiest, when surprise was sharpest and disappointment most intense.

The question strikes me as such a big one, and so unanswered, so little wondered about, so taken (like that mysterious thing, ‘sleep’) for granted, that there must (I muse) be volumes of research, hypothesis and explanation on the subject. So I tried googling — and there really doesn’t seem to be much. The theories I’ve come upon appear almost wholly speculative, often starting from the premise that a weakness in the capacity to retain, to commit to memory at the time, must be the basis of enquiry: as if when we’re two we aren’t properly taking it in and filing it away, but start doing so by the time we’re four.

This strikes me as most unlikely. Those first few years are surely when we’re learning fastest — learning, for instance, to speak, to walk, to recognise, to get our way and to influence the behaviour of others. For learning, memory is indispensable. Can there be any other stage in a person’s life when they are learning more, absorbing faster, retaining better? Why, a child can learn two or three languages by the age of five. Memory must have kicked in well before that.

And so, I reckon, the research and speculation I’ve seen starts from a fundamental error. We’re trying to explain why the memories aren’t there, haven’t been laid down. Instead we should try starting from the assumption that they have indeed been laid down and are indeed still there — a huge, invaluable treasure trove of memory. Whereupon the question becomes this: why can’t we access it? Locked away somewhere in little Gabriel’s brain, that train ride, those bootees, that giraffe and — yes — that stranger opposite him will always be there. But within about three years, as he learns to speak, to walk, to see himself as an active agent in the world, as he gains what will become an adult’s conception of the universe around him, the curtain will come down on those priceless files.

And I think I’ve worked out why. The key lies in a remark of Ludwig Wittgenstein’s: ‘If a lion could speak, we could not understand him.’

Remember Betamax? This recording system was beaten by its rival, VHS, and discontinued 20 years ago. Customers worried in case recordings were lost because they could not be played. We now worry that VHS may face the same fate, the fate of our cassette tapes as well. Likewise CDs. Early laptops too (I remember mine ‘booting up’) cannot easily speak to their successors. None of this archive will in fact be lost, because we have ways of transcribing (as it were) the material. But you see what I mean. Information, though still ‘there’, can be withheld from us because the form in which it’s held is ‘unreadable’ to us.

Now back to Gabriel. Before asking what he does know, what he did retain from our Eurostar encounter, ask yourself what, aged nine months, he doesn’t know — because this is the context.

Gabriel doesn’t really know what clothes are; what heart shapes are; what a giraffe is; what a train is; what the greenery flashing past the window is. He doesn’t entirely know yet what a person is, beyond his mother and father — or even that he is himself a person. His conception of me is only of something resembling his parents but who is not his parent. All these definitions, explanations, all these shapes and templates and categories into which we later fit our experience of the world, are as yet unformed, or only partially formed, in unself knowing Gabriel’s mind.

How then — in what format, using what shapes of things known to him — does he lay down in memory his train ride and his encounter with me? It will all be there, but formatted according to a different system from the one he will soon learn to use. This Gabriel cannot ‘speak’ to his older self. This Gabriel’s memories would be gibberish to his older self: just strange shapes and movements, sensations of sight, touch, smell, hearing; no words, no names, no definitions, as invisible, subterranean, as the foundations of a house. To ask the ten-year-old Gabriel to ‘remember’ his nine-month-old self’s experiences would be like asking a VHS recorder to play a Betamax tape.

This, then, is the curtain that comes down. By the age of four or five Gabriel’s grasp of the world will not be so very different from when he is 50. The big divide comes before that. That is why 50-year-old Gabriel will be able to ‘remember’ things that happened when he was five: the two of them will be speaking the same language. But the language five-year-old Gabriel is speaking finds no words, no shapes, no templates to grasp in the memories of a nine-month-old baby. And that is what Gabriel will mean when he says, as one day he will, that he has ‘forgotten’. But he will be wrong.