Here, in an edited extract from her new book, `The Human Brain, A Guided Tour', Professor Susan Greenfield explores some of the many kinds of memory, showing how they work and what happens when things start to go wrong
In English at least, memory can serve as an umbrella term for a diverse range of processes that may well be quite distinct. Compare the memory processes of an octopus and a human. The octopus has one of the largest brains of all invertebrates, roughly equal in size to the brain of a fish, and composed of some 170 million nerve cells. Although this number seems large, it is trivial in comparison with the human neuronal count of some 100 billion. Nonetheless, the octopus has proved popular in learning and memory experiments, because it has highly developed eyes and a sophisticated system of touch. In experiments, an octopus can clearly tell the difference between certain colours and can attach different meanings to each. For example, it will readily grasp at a coloured ball it has previously learnt is associated with delivery of a prawn, but will not react to a ball of a different colour that has not been paired with anything either rewarding or aversive.
This type of memory, a simple association between a coloured ball and a prawn, may seem a far cry from the memory of a hot summer day or how to ride a bicycle. There are many distinct types of brain processes that fall under the general term memory; the most basic and familiar distinction is between short- and long-term. Short-term memory operates when we try to remember a series of numbers. Everything is fine if there are no distractions, because the strategy is usually to repeat the sequence in our mind over and over again. This process is surprisingly modest: we can only remember an average of seven digits.
But how does it relate to long-term memory? This less contrived type of memory process occurs without any need for repetition or rehearsal. Patients who remember nothing about what has happened beyond the immediate present - exhibiting an almost global amnesia - nonetheless have a short- term memory ability indistinguishable from non-amnesiacs. Clearly, the two processes can be separated, but could someone have a normal long- term memory when short-term memory is destroyed?
Long-term memory is not a one-step process but can be divided into many different aspects, for each of which there appears to be a respective form of short-term memory. Short-term and long-term memory appear to work in series. First, short-term memory comes into operation; it is a transient, highly unstable and vulnerable process where attention and rehearsal are needed in order to lead into the more permanent and dormant long-term memory. Successful rehearsal in short-term memory will eventually lead to that special phone number being retained without constant attention to it.
If an item survives in your memory for more than 30 minutes, it is probably not going to be forgotten, at least for a matter of days. Patients recovering from concussion characteristically cannot remember what happened an hour or so before the event, whereas their long-term memory remains operational. In these cases, there is presumably a disruption of only the first step in the memory process, the short-term memory stage. This early rupture in the normal course of events obliterates any chance of that hour of life being recorded more permanently in the individual's mind.
Short-term memory operates to serve long-term memory. But what do we mean by long-term memory? There is much that we learn and remember as we go through life: how to drive a car, the French for "thank you", what we did when Aunt Flo last came to visit. All these are examples of different types of long-term memory at work. However, the odd item in these examples would be how to drive a car. The memory for a fact, such as the French for "thank you", or an event, such as the recent visit of Aunt Flo, requires that we make an explicit, conscious effort. In contrast, driving a car, like many skills and habits, is performed almost on automatic pilot. This type of memory is therefore referred to as implicit, because we do not need to actively and consciously remember how to do something. In contrast, memory for events and facts can be regarded as explicit memory.
One of the most famous and intensively investigated cases of complete loss of explicit memory is that of HM, a young man who had severe epilepsy. HM's epileptic seizures became so frequent that it was impossible for him to live a normal life. In 1953, at the age of 27, HM had part of his brain removed to control his seizures. Despite its success in combating epilepsy, this operation has never been performed since due to the terrible consequence: after surgery, HM could remember only events up to about two years before his operation. Since the surgery, he has remained trapped in the present.
It is very hard to imagine HM's state of mind. He fails to recognise friends or neighbours that he got to know after the operation. Although he can give his date of birth, he cannot give his correct age, always estimating that he is younger than he is. During the night, he might ask the nurse where he is and why he is there. He cannot reconstruct the events of the previous day. He explains, "Every day is alone, by itself; whatever enjoyment I have had, whatever sorrow I have had." For HM, there are no yesterdays.
As a result of this condition, HM has only been able to carry out simple acts in the present. Therefore, he has been given monotonous jobs such as mounting cigarette lighters on cardboard displays. He could not give a description of the place in which he worked, the nature of his job, or the route along which he was driven every day.
HM can still remember strings of up to seven digits, thus demonstrating that short-term memory is a separate process from the subsequent stage of long-term memory. Moreover, although HM appears to have lost his ability to remember in the long term, his brain has retained a different type of memory. He can perform quite well at certain motor skills, such as tracing a star shape reflected in a mirror. This is a demanding exercise in sensory motor co-ordination that improves with practice, like driving. Every day, HM did indeed improve, showing that this other type of memory - implicit memory - was not processed in the same part of the brain as memories for events. Interestingly, HM was not conscious of remembering the event of learning to draw the star (an example of explicit memory), although his brain was quite happily getting better at doing so - implicit memory.
Although HM cannot remember events occurring after, and two years prior to, his surgery, past memories from long ago are still intact. These memories are obviously not dependent on the brain area that has been removed. It must be the case that no one brain region can assume entire responsibility for the whole memory process of facts and events. Rather, memories must be somehow processed through one region but consolidated elsewhere. In HM's case, the damage must have intervened at the stage where a new memory is first processed. Hence, all the memories that had already been consolidated were safe. This is confirmation that different brain regions are responsible for different aspects of a function.
The area that HM had removed was the middle part of his temporal lobe, which lies on either side of the brain, as its name suggests, by the temples. This area also includes a region underneath the cortex called the hippocampus. A considerable amount of clinical and experimental evidence has shown, subsequent to HM's case, that damage to this region results in an impairment in the laying down of memories.
Even for this more specific aspect of memory, its initial consolidation, there is another area that appears to be important: the medial thalamus, which is vital for the relaying of incoming sensory information to the cortex. We know that the medial thalamus contributes to memory because of one or two unfortunate and bizarre accidents in which people have ended up with either fencing foils or snooker cues up one nostril, destroying the medial thalamus. The victims of these accidents displayed amnesia for events. Unlike the examples of amnesia we have looked at so far, the problem was temporary. Despite this, there is a permanent inability for memory of events that occurred while the amnesia lasted, presumably while the medial thalamus was malfunctioning. Hence the medial thalamus can be seen as important in the consolidation of memories.
Another type of memory malfunction is source amnesia, a loss of when and where an event occurred. If there is no space or time reference, events cannot be differentiated, and there is no personal involvement of the individual with what has happened. As events are unique and personal, while facts are generic and free of time and space frames of reference, it follows that source amnesia will primarily affect memory for events rather than for facts. Whereas memory for both facts and events appears to rely on the integrity of the hippocampus and the medial temporal lobe, memory for events seems affected by damage to a third area, the prefrontal cortex.
Damage to the medial thalamus, which has connections with the prefrontal cortex, can also result in errors in the time-space allocation of memories. Memories can pop up out of context, irrelevant to the speech and ideas of that moment. The prefrontal cortex presumably has some influence not just in the way events are recalled but also in how they are associated with related events.
Facts, as in semantic memory, need differ only from the events of episodic memory in that they are removed from a specific moment and place. Once the pink elephant is displaced from the jungle hideaway in which you saw him one night last summer, he is reduced to the generic thought that elephants can be pink. Damage to the area where facts have been personalised into events by time and space referencing would not actually destroy memory itself but rather would uncouple facts from the contexts in which they occurred. Specific events would be reduced to mere generic facts in that they would have no unique features in time and space.
If the prefrontal cortex is needed for this type of time-space allocation of events, it follows that this type of memory for events would be particularly pronounced in humans, with our disproportionately large prefrontal cortex. For other animals, perhaps memory of an event is more generic, less anchored by unique time and place co-ordinates. A cat may not remember a specific spring day when it caught a mouse in the back garden just after drinking a saucer of milk and before climbing a tree, but it may have a more general recall of catching mice.
However, in the mid-l900s, Wilder Penfield, a surgeon in Canada, contrived a situation where human memories can seem to be more like this generic type of memory. Penfield's pioneering studies involved 500 patients who were undergoing neurosurgery. As there are no sensors for pain within the brain itself, it is possible for the brain to be exposed in conscious patients without them feeling any pain. With the patients' consent, Penfield used the operations, which had to be performed in any case, to investigate the storage of memory in the brain. As the surface of the brain was exposed and the patients were fully conscious, he was able to stimulate different parts of the cortex electrically while documenting the reports of the patients as to what they were experiencing.
Most of the time, perhaps not surprisingly, the patients did not report any new experience. Sometimes, however, they claimed they could remember vivid scenes. They often said that memories were like a dream, more generalised experiences that did not have particular time and space points of reference. Perhaps in this highly artificial situation, the electrical stimulation was locally kick-starting the medial temporal lobe, without recruiting other requisite but more remote regions. Of these remote brain regions, the prefrontal cortex in particular would normally be operative during memory of an event. Without the prefrontal cortex our memories exist but are vaguer and less specific, perhaps resembling the dream-like memories of Penfield's patients, or even normal dreams. If a reduction in the role of the prefrontal cortex did indeed induce a more dream-like state of mind, it would follow that animals with a less pronounced prefrontal cortex do not have the precise memories we have. Instead, their memories could be disembodied facts that lack a time-space context: "episodic" memory for an event would have almost become "semantic" memory for a fact.
So for explicit memory of events and facts, clinical cases suggest that the hippocampus and medial thalamus play a role in laying down memories for about two years; these long-term memories are somehow "stored" in the temporal lobe. Meanwhile, the prefrontal cortex co-ordinates facts with an appropriate time and space context to ensure that an event is remembered as a unique happening.
But how do the memories actually become consolidated in the cortex? We have seen that all types of memory first enter the highly transient and dissociable phase of short-term memory, but short-term memory only lasts at most for half an hour. In contrast is the striking case of HM, where, although he had perfect recall of all that had happened early in his life, he could not remember anything from the period of two years preceding his operation. For the hippocampus and medial thalamus to consolidate memories, it is not just a matter of a few minutes but of a substantial period of time.
No one really knows exactly how the hippocampus and medial thalamus might be working over a period of years, in conjunction with the cortex, to lay down memories that will eventually no longer depend on the integrity of these subcortical structures. One attractive idea draws on a memory being composed of otherwise arbitrary elements, brought together for the first time in the event or the fact to be remembered. The role of the hippocampus and medial thalamus would be to ensure that these disparate, previously unassociated elements are now associated and thus somehow bound into a cohesive memory.
If explicit memory for events and facts depends on an initial dialogue between the cortex and certain subcortical structures, perhaps this same arrangement could also apply to the laying down of skills and habits: implicit memory. Certain habits, such as remembering sequences or making a certain type of movement in an appropriate context without needing to think about it, can be performed adequately in amnesiac patients with medial temporal lobe damage. However, patients suffering from disorders of the basal ganglia, such as Parkinson's disease and Huntington's chorea, have seemingly no problem remembering facts and events. Instead, their problem is that they are no longer able to perform the habit of an appropriate sequence of movement, or of recognising the next item in a sequence that had been shown to them over and over again and normally would have been implicitly remembered.
We have seen that memory can be subdivided into different processes and that each process will be served by different combinations of brain regions. But common to all these processes is perhaps the most mysterious issue of all: some people can remember what happened to them 90 years ago, but by then every molecule in their body will have been turned over many times. If long-term changes mediating memories are occurring continuously in the brain, how are they sustained? Irrespective of brain region, how do neurons register more or less permanent change as a result of experience?
It is well known that most people cannot remember events that occurred before they were about three years old. This phenomenon cannot be accounted for by simple length of time, since we are able subsequently to remember events for some 90 years. Moreover, young children are able to remember habits and skills from an early age - it is only explicit memory that is the problem. On the other hand, babies as young as five months are able, arguably, to show explicit memory by looking at a new item more than one they have previously seen, when the two items are presented together. Children under a year old may copy games they saw someone else playing, even only once, on a previous day.
It seems then that some simple form of explicit memory is available to young children, which in turn would mean that their hippocampus and medial thalamus must be operational. With regard to maturity, more in doubt is the cortex. If neurons in the cortex were unable to form many associations, then the explicit memory of children would not be, as indeed it is not, very robust. After the age of three, the ability to associate items with a richer repertoire garnered from experience, aided in turn by an increased number of neuronal connections in the cortex, would make memory, as we know it, possible.
As yet we cannot establish a causal relation between the physical and the phenomenological in the human brain; however, for the moment, it is sufficient to be aware of the correlation between these two levels of operation. Memory is multi-faceted and multi-staged. It is more than a mere function of the brain, as it encapsulates an individual's inner resources for interpreting, in a unique fashion, the world around them. !
Independent on Sunday readers can purchase a copy of Susan Green- field's book The Human Brain for the special price of pounds 9.99 p&p free (recommended retail price pounds 11.99). Orders can be made by phone on 01903 736736 or by sending a cheque to Littlehampton Book Services, PO Box 53, Littlehampton, West Sussex BN17 7HE. Please quote reference HBReuse content