Numbers And Brains - Numberphile

BRIAN BUTTERWORTH: The first patient that we tested formally and in detail was a woman who had kept the accounts of the hotel that her family ran. But after a stroke which affected her left hemisphere, she was unable to deal with numbers above four. Any numbers in any way we presented them. I mean, if we gave her a word for a number above four, she wouldn't know whether it was even a real word or not, let alone what it meant. What was really interesting for us was that everything else seemed to be fine. I mean, her language was fine, she could understand what we were saying perfectly well. She could remember what she'd had for breakfast. She could reason. We gave her lots of reasoning tests. She was absolutely fine. She understood symbols. So we gave her lots of different sorts of symbols, like traffic signs, political party signs and so on. Absolutely fine. It was just numbers that seemed to be selectively affected. So we thought, well, why is it that numbers are selectively affected? Because the current theory at that time including, the theory that came from Chomsky was really, numbers were just an aspect of language. That all the capacities that you need for language kind of automatically gave you numbers as well, if you bothered to learn them. But here you had somebody whose language was still fine, but her numbers were really, disastrously badly affected. So of course she couldn't do her work anymore. Then we found another patient who seemed to have the opposite problem. Different kind of neurological condition. A kind of degenerative disease that affected a different part of the brain. This affected the temporal lobes rather than the parietal lobes, which seem to be the main areas for numbers. This patient was-- language was disastrous, completely disastrous. If you gave him what we call a picture pointer task, like point to the orange, he basically was at chance at that. And his speech was confined to very stereotype phrases, just a few stereotype phrases. His memory for other sorts of things seemed to be fine, except what we call semantic memory. So he didn't, for example, know that an orange was a fruit. However, his calculation was superb. If you gave it to him in writing, he could do multi-digit, multiplication and division absolutely fine. And what was even more surprising to us was that although his language was terrible, he couldn't read anymore, he couldn't write anymore, if you asked him to read number words like T-W-O or E-I-G-H-T, he could do that perfectly. If you asked him to write these words, remember, they're very irregularly spelt. There isn't another word in English that gets consonant W-O, it's pronounced "ooh." He was absolutely perfect. But on equivalent words which weren't in the number domain like flat or give, or take, he couldn't read them and he couldn't write them. So we thought, well, there's actually a separate system in the brain for numbers. And maybe for arithmetic as well. And so that was really how we got started in this area, looking at these kind of dissociations between the number domain and other domains of cognition. Our first paper in this area was published in 1991. It wasn't that long ago, no. BRADY HARAN: Neuroscience is quite an old science, though. It's not like, brand new. It seems extraordinary to me that in the 1990s, people still had no grasp of how we were dealing with something as fundamental as words versus numbers. BRIAN BUTTERWORTH: That's absolutely right. People were not really aware of the difference between the number domain of memory and other domains of memory, except that in the very early part of the 20th century, there was a Swedish neurologist called Henshin, who actually came to the same conclusion that we did. But although his work is often referred to, it didn't seem to generate a lot of new studies in this area. The left parietal lobe seemed to be the key area for numbers. And if it was damaged, then you've got these selective deficits. If it was preserved, you've got preserved numerical capacities. And if you did the imaging, what you saw was that in number tasks, this was the area that was active when you were doing the number tasks. These are the parietal lobes. These bits around here. And although you do very simple number tasks, like estimate the number of objects in an array, both in the left and the right parietal, when you're doing calculations, it's mostly in the left parietal. So that area there seems to be the part that's critical for numbers. The language areas, they're here. This is Wernicke's area, and this is Broca's area around about there. So you can see that anatomically, they're rather separate. BRADY HARAN: The thing that you've said to me so far that really strikes me is that my brain could be damaged in such a way that I can no longer function very well with words, but I will still recognize the words for numbers, which makes me think the words for numbers, to me, aren't really words at all. Is that right? Does my brain not process E-I-G-H-T as a word? Does it process it as a number? BRIAN BUTTERWORTH: Well, it interprets it as a number. So if you still know number meanings, then you can process E-I-G-H-T. So the neurological patient who could read and write E-I-G-H-T, but not a word like G-A-V-E, gave-- although he had this word in his brain. We know he had the word in his brain because we asked him to say whether G-A-V-E was a real word. He would say yes, it is. What does it mean? He didn't know. How do you pronounce it? He couldn't do it. Can you write it? He couldn't write it. It was in his brain, but he didn't have an interpretation for it. So he only had interpretations for number words. So they were the words that he could pronounce. They were the words he could understand. The other words, although they were still in his brain , he couldn't do anything with them. BRADY HARAN: What do I do for big numbers? BRIAN BUTTERWORTH: Well, you make an estimate which is based on extracting various visual properties from the stimulus. And there's now some brilliant work done by Marco Zorzi's lab in Italy, where they've modeled how this might work.