Mathematical Brain


New Scientist, 1999
Brain Connection, 2000
Britannica & Australasian Scientist, 1999
The Science Show, 2000
The Science Show, 2001
Educational Leadership, 2001
Plus Magazine, 2002
The Science Show, 2006
Audio Interviews  
What's New
Test and New Chapter
About Brian Butterworth
email author
Audio Welcome
Get QuickTime 
Buy Now @ Amazon, UK

Macmillan Logo


Italian Edition

Swedish Edition

Naze sugaku ga tokui na hito to nigate na hito ga irunoka?
(Why are some people good, but others bad at maths?)

Mathematical Brain
Brian Butterworth



The strange case of the former bank manager who is now able to read only number words as the result of a neurodegenerative condition.


Brian Butterworth: What's fascinating about him is that he can only read one sort of word. He can only read number words and it's never been reported before and that's why we're so interested in it. But it's theoretically very important because the standard theory of reading goes something like this: that when you see a string of letters you do two things simultaneously; you try and find the sound of that string of letters. You might do that by taking each letter and finding the sound of each letter and putting them together. You might try and recognise it as a whole word and try and find the sound of the whole word but, at the same time, you're trying to interpret the meaning of that string of letters, you're trying to find the meaning of that word and these two processes converge to give you the pronunciation of the word.

Now the problem with the theory up to now is that it's never really been tested properly. We don't know whether you can pronounce the word accurately just going via the meaning. So for example, if you see a letter string like t-r-e-e- you get the meaning of a tree and then you have to say what that meaning is. Now what we want to know is: is that route via meaning sufficient to give you just "tree", and we haven't been able to do that up to now.

This case is very interesting because this man can't use the route via sound alone. We know this because if you give him a very simple string of letters he hasn't seen before, like y-i-t- he can't say it. He used to be able to say it but he can't say it anymore. Even if the letter string sounds like a real word, like y-o-t- he still can't say it. Now the other really interesting thing is that he can't say words that he doesn't understand. So he can't say y-a-c-h-t says yacht either, he can only say words that he does understand and if he does understand them then he says them accurately. And we know that he has a very precise understanding of numerical concepts because he can calculate accurately. So he knows the difference between 1 and 2, and 3 and 4, and 13 and 15 and he can read all these words with perfect accuracy, but words he doesn't understand, and these are very common words like "take" and "give", and yet he can't read because he doesn't understand them.

The reason why he doesn't understand them is because this unfortunate neurodegenerative condition has affected the parts of the brain that deal with language and other forms of general knowledge, which is in his case in the left temporal lobe, but has spared the left parietal lobe which is where numerical ideas are.

Robyn Williams: It's very interesting you gave the example of "tree" just then. It's very close to "three' and I would have thought that one little letter might be something that could help him on his way. Are you telling me that he can read "three" but he can't read "tree"?

Brian Butterworth: That's exactly what I'm saying. Now all the previous cases of reading problems due to brain damage have been complicated by the fact that the patient either has very good word meanings in general, in which case we've now shown that this route via meanings is sufficient for reading words, or they've got some of the letter to sound route still intact, and so although they might not really understand "tree" they can maybe sound it out a bit and then get to the pronunciation. But this man unfortunately can't get any of the letter sound relationship. He can say number words OK and he can say a few other sorts of words reasonably well, but not very many. So his language is very affected too.

Robyn Williams: Now Brian, this is clearly significant because it's in the journal "Nature" this month, "Neuroscience", and you're a distinguished researcher in this field, but for innocent bystanders like me, if you're looking at one case of a person who's got a very unusual impairment how do you know that this is a window on the general brain?

Brian Butterworth: There are two answers to this. First of all, it would be impossible to do neuropsychology if you couldn't make generalisations of single cases; therefore you must be able to make generalisations of single cases. The other, perhaps more sensible account goes something like this: that we have no reason to believe that I H's brain was any different from yours or mine prior to his illness and so his brain is likely to be organised at least for things that are rather basic in a similar way to yours and mine. So we can get a clue as to the organisation of what we would call higher mental functions like calculation and language from these single cases, and it's actually proved very effective as a way of investigating brain functions since the time of Brocca in the 1860's. We can confirm some of these single case findings by looking at functional brain imaging, seeing what the brain is doing while it's doing it to see if the areas that we've identified in these single cases are the areas that seem to be important for a particular function.

So for example, I believe in the case if I H that the temporal lobes are involved in language and that the parietal lobes are involved in number. We've done experiments and other people have done experiments using functional imaging of normal people, people who do speak and calculate normally. And we find that the temporal lobes are very active during speech production and speech comprehension and the parietal lobes are active during calculation, so we do have independent grounds for thinking that the brain of I H is not at all unrepresentative of humanity in general. I mean, one point that I should make is that before his illness I H was a bank manager. Now we assume that bank managers have normally organised brains.

Robyn Williams: Sometimes too organised.

Brian Butterworth: Sometimes. He was also and indeed still is an addicted gambler, so he's obviously had a lot of experience with numbers. So it may be that his number areas are larger and better developed than mine are for example. However, one of the consequences of his illness is that he's had to change his preferred mode of gambling. It used to be horses, but in order to bet on horses you have to be able to read the name of the horse. On the other hand, if he goes to the dogs there are only six dogs in every race and you usually identify each dog but its number. So you can go up and say $10 on No.3 and that he does perfectly well and he still continues to lose money in the normal way.

Robyn Williams: Poor man. However, here is a very interesting example just to recap. He knows numbers, therefore he is able to read something he understands and therefore he has this facility, but he can't read non-number words. What does this tell you about how we might approach reading in the first place? What does this tell you that might be of assistance to treating people who've got reading difficulties at all?

Brian Butterworth: As I said, the theory says and the theory now seems better supported on the basis of this case, that when we see a letter string we approach it in two separate ways; at least normal people are trying to find the sounds of the individual letters and trying to find the meaning of the letter string. Now some people, either through brain injury or through some genetic abnormality known as dyslexia, find it difficult to read by one or other system. And I think that one of the implications of this is that you try and identify which system it is that's affected in the individual and then try and exploit the intact system.

Robyn Williams: And so you've got a test, have you now?

Brian Butterworth: We've got tests and these tests have been available for quite a long time. It's just that you very rarely get cases as clear as this, so this tells us that we've been on the right track.

Robyn Williams: And what in the practical sense - how do you apply this; having done the diagnosis on say a dyslexic patient, how do you actually say right, now try this?

Brian Butterworth: Well, what you do is you try and encourage the patient to, instead of sounding out letters if he's the type of patient that I H is, to try and sound out the whole word. And you might give them meaningful hints as to what those words are. So if it's "tree" you might say, well think of what you see in the country.

Robyn Williams: Cows!

Brian Butterworth: Right.

Robyn Williams: Except in England, there's not many left.

Brian Butterworth: Yes, that's not a surefire way of doing it, but certainly most children with dyslexia are going to have intact semantics so they can use this route via meaning if that's the route that's preserved or that's developed reasonably normally in their case. But the case of I H is interesting from another point of view, which is that we've been interested for quite a long time in how language and numerical abilities are organised in the brain and it's these very clear cases that can somehow tell you about that. For example, we reported patients who have got very good language but very impaired numerical abilities. So there was a patient that we reported some years ago who, if you spoke to her, you wouldn't know there was anything wrong with her, but because of her brain injury she wasn't able to do calculations anymore. She couldn't in fact deal with numbers above 4 at all. Before her stroke she'd had care of the books of the hotel that her family ran, so she was pretty good a numbers prior to her injury.

Now people might say, well of course, if you've got brain damage what's going to go first; Is it the stuff that you found hard? And so we find language easy because we use it all the time but numbers are a bit more difficult because you have to learn them at school. Cases like I H say, well that can't be quite right, because here we've got somebody whose language is gone, language of course which he's used everyday, in his dreams even, but his numbers - he's very practised with numbers but nevertheless he's not using numbers as much as language. So numbers are still going to be, on those grounds, slightly more difficult than language, yet it's his numerical abilities that are preserved.

And so we now have evidence what neuropsychologists like to call a double dissociation between language and number. One type of patient has good language and terrible numbers and this other type of patient, I H is the best reported example so far, has got terrible language but very good numbers. So we can see that in the brain these two processes are independent. And in fact, we can look to the evolution of these two processes and we can say, well maybe they have evolved independently as well, because we know that infants, even in the first week of life, are sensitive to the number of things that they see. So if you show a baby of one week two objects, another two objects, another two objects, another two objects, the baby loses interest. But then when you show the baby three objects the baby suddenly starts to look.

Robyn Williams: Numbers, numbers.

Brian Butterworth: Numbers, numbers, numbers, that's right. So you're born with this numerical ability and it seems that the brain circuits that you're born with are in the parietal lobes and in the case of I H those seem to have survived whereas unfortunately his neurodegenerative condition has affected his language very severely.




Preface Reviews What's New email author Author
Test and New Chapter

Macmillan Logo