Liz Brannon — Seminar on Numerical Cognition
Professor Liz Brannon of Duke spoke at our last Uni seminar of the semester on the very cool topic of numerical cognition in humans and other animals. Her talk focused on research in her lab and others on how human infants develop “number sense” — essentially, a primitive ability to distinguish between different numbers of objects that precedes knowledge of formal arithmetic — and how numerical cognition in human infants compares to that in other mammals, including monkeys, mongeese, and lemurs.
I’ll first describe a bit about how these experiments work in Monkeys. Basically, the monkey sits in front of a touch screen and is given tasks to do. The tasks involve pictures of some number of objects. For example, the monkey might be shown two pictures on the screen at the same time and have to touch the one with more objects. When she gets it right, the monkey gets a sip of juice. After some amount of training, monkeys can do this quite rapidly and accurately. They can also be trained to “add” and “subtract.” In these arithmetic experiments, the monkeys see some dots go behind an opaque section of the screen. They either see two sets of dots going in, or one set going in and one set going out. They are then shown two pictures and they have to choose the one with the correct number of dots. Amazingly, monkeys can be trained to do this too.
The accuracy and speed with which monkeys — and, it turns out, other animals — can perform this primitive sort of arithmetic is a function of the “Weber fraction.” This is essentially the ratio of the number of objects in one set to the number of objects in the other. For example, if one set has four objects and the other two, then the fraction would be 1/2. The same is true for a set with ten objects and one with five. The closer the fraction is to 1, the slower and less accurate the monkey becomes at determining which set has fewer. This by itself isn’t too shocking, but the really crazy thing is that the effect on reaction time and accuracy is relatively conserved across species, including for humans!
The Brannon lab has also done a lot of work with numerical cognition in human infants. It’s quite hard to do this research with very young children, but they have found that even for six month old babies the existence of number sense is measurable. To address this, they show babies a stream of periodically changing pictures. The pictures are, again, just pictures of some number of objects. When the number of objects doesn’t vary much from picture to picture, the babies just sit there and look bored. But when a new picture is shown with a substantial difference in the number of objects then the baby perks up and gets pretty excited. He also stares longer at the images when the object number changes a lot vs. not at all or only a little.
Finally, there has been some interesting research from the Brannon lab and others suggesting that this primitive numerical acuity in infants is correlated with performance on actual arithmetic tests when the children begin to learn formal mathematics. This effect seems to persist later in life as well. This suggests that training this ability may actually help to improve performance in arithmetic tasks.
All in all a very interesting body of work and a great way to close our semester!
