There are some things we wish we could remember better (e.g. exam answers) and embarrassing moments we would prefer to forget. But once a memory is lodged in our heads, how reliable is it? 

How we behave is influenced by things around us but also by things we have experienced in the past. Being able to learn and store information is an invaluable survival aid. So it might be assumed that our memories are faithful records of the past. But there is plenty of evidence that this is not always the case.

Many experiments have shown that our recall is easily biased in a whole host of ways. For example, mood influences memory making - terms in tune with our current mood are remembered better. Things we see will tend to be remembered better than things we hear, unusual items in a list are easier to recall, and items at the beginning and end of lists are more likely to be remembered.

We also have a number of 'cognitive biases' that affect our recollection. We have a tendency to remember things as better than they seemed at the time (the 'rosy retrospection' bias), and when we have chosen one option we tend to associate more positive things with it.

We all show such biases to some degree, but for some people they can turn into a serious problem. People with psychological disorders, such as depression or social phobia, for example, tend to focus abnormally strongly on negative emotional stimuli - an effect also seen in their memory retrieval. Someone with depression will tend to recall the bad times in a relationship rather than the good, or the negative aspects of an event rather than the positive.

With post-traumatic stress disorder (PTSD), 'flashback' memories of the trauma are triggered very easily and intrude on normal life. People regularly relive the distressing events from their past. Cognitive behavioural therapy can be used to treat PTSD (and other psychological disorders). There is also interest in using the drug propanolol, which blocks the action of the neurotransmitter noradrenaline and specifically interferes with traumatic memories.

Similarly, there has been interest in using midazolam to block memory making. One small study used it to block memory formation in children undergoing surgery. It has been suggested that midazolam interferes with explicit (conscious) memory making but not implicit (subconscious) remembering; although generally true, this may be an oversimplification.

Another possible use might be with soldiers, who are exposed to the horrors of war and suffer a high incidence of PTSD (though some query the ethics of this application).

Hidden memories

One of the most controversial areas is that of 'latent' or suppressed memories. This has come to the fore with stories of hidden memories of childhood sexual abuse being 'rediscovered'. This is a sensitive area, but it is clear that the memory can easily be manipulated and 'false memories' implanted. In one recent study, US researchers convinced a significant proportion of students that they had had a bad experience with ice cream; this was sufficiently convincing that the students were put off eating it.

The effect starts young too. In one study of pre-school age children, one group were exposed to a rumour of a particular event, rather than the event itself. Two weeks later, the number of children saying that they had experienced the event was as high in the rumour group as in the group that had actually been there.

Implanted memories may also underlie another curious phenomenon, alien abduction. It is clear that a large proportion of people who claim to have experienced alien abduction genuinely believe it. It is entirely possible that they believe it because it is an embedded memory and every bit as 'real' to them as any other memory they hold.

This raises the difficult question of whether we should always 'believe' our memories. While we need to rely heavily on their faithfulness, it is worth bearing in mind that they may be deceiving us - just as optical illusions deceive our visual perception. Indeed, being uncritical of memories may be problematic. There is some evidence in schizophrenia, for example, that auditory hallucinations - 'hearing voices' - may be irrelevant aural memories surfacing into consciousness. People with schizophrenia may be less able to see these as 'unreal' and disregard them.

Although there is much still to be learned about the mechanisms of memory encoding and retrieval, it is clear that memory is in no sense a digital recording stored away for future replay. The implications for the legal system, for example on the reliability of eye-witness testimony, are likely to be significant.