After the damage caused by a stroke has been assessed, the most important question for anyone facing this situation is whether it can be repaired. The answer to this question depends in part on the sheer extent of the damage, and in part on the particular networks and sets of connections that were damaged. The essential question is whether an alternate route--a 'detour'--can be found that will permit the person to perform some important acts in cases where the usual route is now 'out of order.' If there are already connections in place that can serve as the basis for such a detour, even though they are weak because they were not used much before the stroke, they can often be strengthened and used in place of the damaged connections.
ilda's stroke, in which the animal-name network itself was damaged, input from several other networks at the same time will probably be necessary to activate the names of animals, since individual neurons do not regenerate. However, Matilda can be encouraged to try to produce such additional input by herself, instead of waiting for someone else to provide her with cues. For example, she can be taught to try to think of as many facts as she can about the animal in the picture. Since we have been assuming for simplicity in this case that Matilda's animal-name network was the only one damaged in the stroke, she should be able to learn to do this. Then, when faced with the picture of a donkey, she will be able to activate sentences like 'This animal eats hay' or 'This animal carries loads' in her animal-fact network. The activation in this network could then work together with that in the animal-picture network to activate Matilda's animal-name network sufficiently to allow her to say 'donkey.' In some cases the stroke may damage the connections between two networks rather than those within a particular network. Such cases may be treatable by a very similar method, even though the processes of activation between the networks are somewhat different. Let us say that the connection between the animal-picture network and the animal-name network has been damaged in Carlton's brain as the result of a stroke. Then he will not be able to say 'donkey' when he sees a picture of a donkey, and the letter 'd' will not help him either, because there are so many different words beginning with that letter. In fact, the 'd' is more likely to activate 'dog,' since this word is more familiar, and the word 'donkey' is not being activated by its picture because of the broken connection.
But here too the method of encouraging the patient to think of as many facts as he can about the animal in the picture can be useful, although for a different reason. In Carlton's case the animal-name network itself is intact. Thus, when he thinks of some facts about the donkey in the picture, his animal-fact network is activated, and this in turn activates his animal-name network. That is, thoughts about the animal in the picture eating hay and carrying loads may allow Carlton to think of the word 'donkey' even though he cannot think of the word when he just looks at the picture. In this case the activation takes a detour around the damaged set of connections by using the intact connections going from the animal-picture network to the animal-fact network, and from there to the animal-name network.
These examples illustrate the way rehabilitation works after a stroke. In those cases where the rehabilitation is successful, its success is due to the strengthening of weak connections so as to form a detour around the damaged area. This also explains why so much practice is needed, and why the process is aptly described as 'learning how to walk/talk/write all over again.'
How is temporary memory different?
One mental function which has turned out to be practically impossible to recover after damage is the ability to store information in temporary memory. There are various ways in which temporary memory differs from permanent memory. Networks which allow us to remember what we did five minutes ago and what we need to do five minutes from now are differently structured from the networks that let us name animals and think of many facts about them. Not enough is known yet about how the networks that serve temporary memory operate to be able to say why this function is so difficult to restore. But since the main mechanism for restoring function after damage in other areas of the brain seems to be the strengthening of connections that can constitute a detour, it seems plausible to suggest that there may be only one set of connections available for storing information in temporary memory, so that if these connections are damaged, no other connections can be recruited to detour around them. Let us look at what actually happens in the case of damage to temporary memory. Let us say that Jane has had a stroke which also impaired her ability to move her left arm and leg, so that she is in the hospital for evaluation. She asks the doctor why she can't use her arm, and he explains the situation carefully. Jane seems to understand what the doctor is saying, so he goes on to evaluate the patient in the next bed. As he passes by Jane's bed again on his way out of the room five minutes later, she says, 'Doctor, could you please tell me why I can't move my arm?' If the doctor is alert to the possibility of damage to temporary memory, he will realize that Jane may not have simply failed to understand his explanation; she may have totally forgotten that she just asked him this question and he just answered it. He will therefore order tests to examine the possibility that Jane's hippocampus, where her temporary memories should be stored, has been damaged.
What seems to have happened here is that the doctor's explanation was not encoded at all in Jane's brain. But let us take the story a bit farther. Jane's daughter, who is sitting with her in the hospital and has heard the doctor's explanation as well, tells it to her again and again, each time she asks. The next day, when the doctor comes back, he decides to test Jane's memory himself and asks her if she knows what is wrong with her arm. She answers that she had a stroke. He then asks her how she knows, but she cannot answer.
The reason that Jane can answer the first question but not the second is that her permanent memory has not been damaged. When her daughter answers her question about her disability again and again, the answer can be stored in whatever network is responsible for storing her knowledge about herself. This takes a long time, with many repetitions, because her temporary memory, which normally does the job of getting important facts into permanent memory networks, has been damaged. Eventually, however, as her daughter takes over the function of the temporary memory network by reminding her again and again of what has happened to her, this fact is stored in Jane's permanent memory. The memory about how she learned it, in contrast, is the sort of information that is generally stored only in temporary memory and not moved to permanent memory. Thus it does not find a place in permanent memory, and so is forgotten.
Trying to get information directly into the patient's permanent memory when her temporary memory is damaged can thus be seen as a sort of detour as well, since one form of memory is being used in place of another. The difference between this sort of detour and the one discussed earlier is that the first sort is a detour within the structures of the permanent memory networks, while the second involves the use of permanent memory in place of temporary memory, because of the impossibility of forming detours within temporary memory.
