Professor of Behavioural Neuroscience, Dept. of Experimental Psychology, University College London
Memory Is a Labile Fabrication

We used to think of memory as a veridical record of events past, like a videotape in our heads that is always on hand to be re-played. Of course, we knew this memory to be far more fragile and incomplete than a real videotape: we forget things, and many events don’t even get stored in the first place. But when we replay our memories, we feel sure that what we do recall really happened. Indeed, our entire legal system is built on this belief.

Three scientific discoveries in the past century have changed that picture: two some time ago, and one—the “news”—that is very recent. Some time ago, we learned that memory is not a record so much as a re-construction. We don’t recall events so much as reassemble them, and in doing so, crucial aspects of the original event may get substituted—it wasn’t Georgina you ran into that day, it was Julia; it wasn’t Monte Carlo, it was Cannes; it wasn’t sunny, it was actually overcast—it rained later, remember? Videotapes never do that—they get ragged and perhaps skip sections or lose information but they don’t make things up.

It has also been known for some time now—since the 1960s in fact—that the act of re-activating a memory renders it temporarily fragile, or “labile.” In its labile state, a memory is vulnerable to disruption, and might be stored again in an altered form. In the laboratory, this alteration is usually a degradation, induced by some memory-unfriendly agent like a protein-synthesis inhibitor. We knew such drugs could affect the formation of memories, but it is more surprising that they can also disrupt a memory after it has been formed, albeit only when it has been re-activated.

The story doesn’t end there. Very recently, it has been shown that memories aren’t just fragile when they have been re-activated, they can actually be altered. Using some of the amazing new molecular genetic techniques that have been developed in the past three decades, it has become possible to identify which subset of neurons participated in the encoding of an event, and later experimentally re-activate only these specific neurons, so that the animal is forced (we believe) to recall the event. During this re-activation, scientists have been able to tinker with these memories so that they are different from the original ones. So far, these tinkerings have just involved changing emotional content, such that, for example, a memory for a place that was neutral becomes positive, or for one that was positive becomes negative, so that the animal subsequently seeks out or avoids those places. However, we are not far from trying to actually write new events into these memories, and it seems likely that this will be achievable.

It may seem odd that memories are so plastic and vulnerable to change. Why would we evolve a strange, disconcerting, system like this? Why can’t memory be more like a videotape, so that we can trust it more? We don’t know the answer for sure yet, but evolution doesn’t care about veracity, it only cares about survival, and it usually has a good reason for what may seem like odd design features.

The advantages of the constructive nature of memory seem obvious: rather than the enormous storage capacity required to remember every “pixel” of a life experience, it is a far more economical use of our synapses to stockpile a collection of potential memory ingredients, and then to simply record each event in the form of a recipe: take a pinch of a Southern French beach, add a dash of an old school friend, mix in some summer weather… etc. The labile nature of memory is more curious, but many theoretical neuroscientists think it may allow construction of super-memories, called semantic memories, which are agglomerations of individual event-memories that combine to form a more general, overarching piece of knowledge about the world. After a few visits to the Mediterranean you learn that it is usually sunny, and so the odd incidence of overcast gloom gets washed out, and fades from recollection. Your behavior thus becomes adapted not to a specific past event but to the general situation, and you know on holiday to pack sunscreen and not umbrellas.

The fabricated, labile nature of memory is at once a reason for amazement and concern. It is amazing to think that the brain in constantly and busily re-assembling our past, such that that past is not really the one we think it is. It is also concerning, because this constructed past seems extraordinarily real—almost as real as our present—and we base our behavior on it trustingly.

Thus, an eye-witness will make confident assertions based solely on recollection that lead to the lifelong incarceration of another person, and nobody worries about this except neuroscientists. It is also amazing/concerning that as scientists and doctors, we are now on the threshold of memory editing: being able to alter a person’s life memories in a selective manner.

The therapeutic potential of this is exciting—imagine being able to surgically reduce the pain of a traumatic memory? But we could also reach into the brain and change a person’s past, and in doing so change who they are. These are technologies to use with care. However, one could argue that the fabricated and labile nature of our memories means that perhaps we aren’t really who we think we are, anyway.