IN JORGE Luis Borges' last published short story, a Shakespeare scholar meets a man in a pub who offers him an incredible gift: The Bard's memory. Despite the previous owner's warnings that the gift is not a sinecure, the narrator believes this is his great opportunity: by possessing Shakespeare's memory, he will understand the man, he will discover the secret of his genius. It does not turn out that way. The scholar begins by whistling an unfamiliar melody and seeing unknown faces in dreams, but the memories that later come to overload his mind and threaten his sense of who "he" is are always fragmented, associative, fleeting. "A man's memory," he discovers, "is not a summation, but a chaos of vague possibilities."

As with many of Borges's stories, Shakespeare's Memory works so brilliantly because the simple daring of its premise is allied to a profound insight into the human condition. Memory truly is evanescent, shadowy and ambiguous. Try to pin it down and it just dissolves or changes shape or conjures up an equally plausible (but false) alternative. As almost every scientific experiment into human memory has shown, no two people, in the same situation, ever recall precisely the same set of events and details; and, if you ask them again weeks, months or years later, those memories are just as liable to become (illogically) more detailed and certain, as they are to fade away or mutate completely. Memory, in other words, is as untrustworthy as a politician. Even more so, indeed, as we have no understanding of its motives or methods.

And yet, memory is also the keystone of identity. Lose your memory and you lose who you are. But if memory is so "fickle and slippery" (to quote Pliny's Naturalis Historia), and we are nothing but memory, what does this say about us? That we are ever-changing, illusory creatures with no real substance beyond the cells that constitute our physical being?

What I think it says about us is this: we are the stories that we tell ourselves about ourselves. We are fictions. Based, to some degree, on fact, of course, as most fictions are, but still, essentially, made up. Memory is not a record of what has happened to us; it is not even the sum of all that we remember. It is, I believe, like a spider's web; an intricately patterned labyrinth woven together from all we remember, all we forget, and all we imagine, fear and wish to believe about our lives and our worlds. As the Czech philosopher Tomas Ryal wrote: "Memory is eight parts forgetting to two parts daydreaming."

This idea - that memory is a lie we tell ourselves every day - is one of the central themes of my second novel, The Amnesiac. It is the story of a man who breaks his leg and spends the next eight weeks alone, thinking about his past. He comes to the realisation that three years of his life are missing from his memory, and he goes back to the city where he was living during those years to discover the truth of what happened to him.

The seed of the book was autobiographical. There is a part of my life of which I have almost no memory: a period of about six months, during my first year at university. I do not have total amnesia for this period. I recall places, faces, names, and something of the general mood. But in terms of events, I have memories of only two or three, whereas for the three-month period following this, I have perhaps 20 or 30. There are doubtless pragmatic reasons why I remember so little of this time - I drank too much alcohol; I was very shy, so probably didn't leave my room or interact very much; I almost certainly did and said many embarrassing things, which I willingly blanked out - but looking back at this time from my mid-30s, I felt both attracted and disturbed by the mysteriousness of those empty spaces in my mind.

So I set out to write a detective story, in search of lost time. But the more I compared those "lost" months to other months that I did remember, the more I came to realise that my amnesia was merely a question of degree. The more I thought about memory, the more I doubted its existence. Certain mornings, particularly after a heavy night's drinking, I had the impression that my brain was scrambling to retrieve even the smallest fragments of the night before and to fit them into some kind of chronological order. In many cases, I felt like I had to decide what must have happened, based on very little evidence at all. And if that were true of a single night, whose events were separated from me only by eight hours' sleep, how much more true must it be of events a year, or five years, or 10 years, in the past? This insight had a powerful effect on me. I felt as if my own life were slipping through my fingers, vanishing second by second into oblivion.

As you may have gathered, I became somewhat obsessed. I began reading about the science of memory, and came across several strange cases of amnesia and memory. There is Sarah Hinde, a 30-year-old epileptic who lost all her childhood memories at the age of 18, and has since formed a new life, obsessively photographing and writing about all her friends because the doctors have warned her that her lost memories might return, in the process erasing all her memories of the past 12 years. There is Gordon Bell, the Californian scientist who has recorded every detail of the last 20 years of his life via a miniature camera and a computer, offering the possibility of a future technological alternative to unreliable human memory - which is already in development elsewhere in California, where research teams are working on an artificial hippocampus that will be tested on rats. And there is the famous, fascinating story of HM, the man who, in 1953, had his hippocampus removed in an attempt to cure his epilepsy. The result? He never had an epileptic fit again, but nor did his brain ever "make" a new memory. He suffered (and still suffers) from anterograde amnesia, which means that he believes Truman is still president, and that his mother is alive; he cries, every day, when he is told about her death.

Anterograde amnesia is in fact by far the most common form of memory loss. Yet it is rarely used in films or novels (the excellent Memento is the only example I've ever seen) because it does not easily lend itself to drama. By contrast, retrograde amnesia (the loss of memory for a specific period in the past), despite its rarity in real life, is an everyday occurrence in fiction because of the possibilities it offers for mystery, for symbolism, and for explorations of identity. As a plot type it became so popular in the 20th century that it can almost be considered a genre in its own right. There is even a literary anthology dedicated to it: The Vintage Book Of Amnesia, edited by Jonathan Lethem.

ODDLY, the Borges story chosen for this anthology is not about amnesia at all, but its absolute opposite: Funes, His Memory is the story of a young man who, having been bedridden after falling off a horse, discovers that he can remember everything. Before the accident, he says, he "had been what every man was - blind, deaf, befuddled, and virtually devoid of memory". Afterwards, he is able to reconstruct, perfectly, every dream and daydream he had ever had. Twice he reconstructs an entire day - "each reconstruction had itself taken an entire day".

Can we imagine how it would feel to have such a prodigious memory? Funes finds it difficult to sleep because he "could picture every crack in the wall, every moulding of the precise houses that surrounded him". He finds it difficult to generalise because he does not recognise that "the dog' of three-fourteen in the afternoon, seen in profile, should be indicated by the same noun as the dog' of three-fifteen, seen frontally". All of this sounds like a parable, like an exaggerated fantasy, yet Borges's insights regarding "memoriousness" were echoed almost precisely by Russian psychologist Alexander Luria's book about a real-life case of a man with a limitless memory, The Mind Of A Mnemonist, published some 20 years later.

Aside from this uncanny second-guessing of science, what is striking about the story of Funes is the hellishness, the unbearable weight of this imagined memory's burden. To remember everything is not to be able to live in the present, to relax, to think abstractly, to create a narrative of your own life. In other words, forgetting is an essential function of a healthy working memory.

This is a recurring theme in the books on the science of memory. We need to forget in order to avoid the "garbage heap" scenario of our brain being overwhelmed by trivial details. We also sometimes wish to forget major events, yet are unable to do so; particularly in the case of traumatic memories, which recur frequently and involuntarily, colouring the present with the dark emotions of one specific moment in the past. This is obviously true of catastrophes, such as the Holocaust, but equally so for smaller, more personal events - a broken heart being the most common. In all such cases, there is the fantasy of a drug or treatment that will wash the bad memories away, like the waters of Lethe.

This is the idea underpinning the plot of the Oscar-winning movie, Eternal Sunshine Of The Spotless Mind, in which the two protagonists, former lovers, each have the other electronically erased from their brains, with tragicomic effects. Such technology at the moment is still in the notional stage, but there is a firm in New Jersey called Memory Pharmaceuticals run by Nobel-winning neuroscientist Eric Kandel, which is researching a drug that does something similar; while Tim Tully, a genetics expert who experiments on fruit flies, apparently has such a drug already in the works. According to Lauren Slater, author of the popular science book, Opening Skinner's Box: "If marketed, this drug could be used within 24 hours of a trauma, and it would delete your memory of the trauma, along with whatever else happened that day."

Such drugs, of course, bring with them serious ethical and emotional questions - could the pill be "force-fed" to a victim by a rapist, for example? How would it feel to have such a "hole" in your life? And that is why none are currently available. On the other hand, memory-enhancing drugs, aimed at preventing and treating the effects of diseases such as Alzheimer's, are expected to flood the market within the next few years. Once available, of course, there is no reason why they may not also, illegally, be taken by students cramming for exams or office workers hoping to impress the boss. Memory-enhancing pills could, in theory, alter the whole nature of what it means and how it feels to be human. They could make us superhuman memory machines, as prodigious as Funes or as Gordon Bell's surrogate computer memory.

But would we truly be better off? Personally, after working through my own horror at how little of my life I truly remembered, I have now learned to stop worrying and love my forgetfulness. In the words of Bertrand Russell: "Happiness is nothing more than good health and a bad memory."

Sam Taylor's novel, The Amnesiac, is published by Faber and Faber, £12.99