This is Episode 4 of Memory, Forgetting, and the Self in the Machine: how intelligence reaches across time when it cannot carry the whole past with it.
Ireneo Funes once set out to reconstruct a single day. He did not mean its outline, the way you might recall that last Tuesday involved a dentist and a late train. He meant all of it. He managed it. The job took him another full day. Funes is the invention of Jorge Luis Borges, who published his story in 1942. As the story is usually summarised, Funes remembers everything he perceives, and he cannot think in generalities.
Funes is fiction, and nothing he does is evidence about how real memory works. Fiction can still push a quantity to its limit: push memory to “everything” and the day of recall that costs a day comes into view. The question it raises matters well beyond a short story. Anyone who has used an AI assistant that forgets has wished, at least once, that it simply kept everything. The wish rests on an assumption: that keeping more of the past means having more of it. Remembering everything and being able to use what you remember are two different properties. The first does not deliver the second, and it may cost it.
The arithmetic of a perfect day
Take that day at face value. If bringing back a day means replaying it at full resolution, recall runs in real time. To put our own numbers on it, a man who wanted to consult his whole past this way would need a second life to read the first, and he would fall a day further behind for every day he spent looking. Nothing in that arithmetic is about room. It is about the trip back.
William James made the same point in 1890, without the fiction. In the chapter on memory in The Principles of Psychology, he writes that “in the practical use of our intellect, forgetting is as important a function as recollecting.” In this piece forgetting has one sense, which is James’s: a past detail that does not come back when we recall a stretch of time. Whether it was destroyed, or is still there and never fetched, is a separate question. James does not treat this kind of forgetting as a fault in the equipment. “This peculiar mixture of forgetting with our remembering,” he writes, “is but one instance of our mind’s selective activity. Selection is the very keel on which our mental ship is built.”
He then gives his reason, and it is worth reading slowly because it is exact: “If we remembered everything, we should on most occasions be as ill off as if we remembered nothing. It would take as long for us to recall a space of time as it took the original time to elapse, and we should never get ahead with our thinking.”
The claim is about time, not space. James is not saying the mind would fill up. He is saying the mind would be busy. If recalling a stretch of the past meant bringing back everything that filled it, recall would take as long as living it did. The ordinary alternative has a name in James, borrowed from someone he calls M. Ribot: “All recollected times undergo, accordingly, what M. Ribot calls foreshortening; and this foreshortening is due to the omission of an enormous number of the facts which filled them.” To take our own example, your last holiday comes back as a few scenes and a mood because almost everything that filled it has been left out. The omission is what makes the holiday available in a second.
Two cautions keep this honest. James is offering an argument, not a measurement, and he reports no experiment that timed anyone’s recall. And a line often hung on him, that the art of remembering is the art of forgetting, does not appear in his chapter. What he wrote is plainer and more useful: the cost of remembering everything is paid at the moment you try to get it back.
