Some thoughts on John Duncan’s “How intelligence happens”
Recently I read John Duncan’s "How intelligence happens". It was an interesting book, written from one of the world leading experts in the field.
I a few words, John Duncan believes that the root of intelligence lies mainly in the frontal and parietal complex, in a system he calls the "Multiple Demand System". This system is responsible for organizing the logical steps required in getting an action done.
For example, let’s say that you have to solve the simple equation x+5-42=3+5. The multiple demand system would create a sequence of actions like this:
1) x+5-42=8
2) x+5=50
3) x=50-5
4) x=45
This looks a lot like the approach taken by classic AI in solving problems, that is, by analyzing logically a problem into discrete steps. I like this theory for the following reasons
1) It makes sense, since almost every person will agree that for solving logical problems, we follow discrete steps.
2) I think it can be linked quite easily to the findings that link working memory to intelligence. For those of you who don’t know what working memory is, it is the cognitive system responsible for keeping all relevant information for a task in our minds. Ever tried to calculate, for example, 13234/4349384923384? The reason you can’t complete the calculation is due to your limited working memory capacity.
3) It defines a specific brain area as the core area of intelligence, which gives directions for future research.
What I don’t like about the theory:
1) I believe that the theory is focused too much on problem solving, leaving other areas of what we call intelligence outside. What about idea creation? Could, for example, all great scientific theories be simply explained by the workings of the multiple demand circuit?
2) What about understanding? Could understanding, for example, a difficult mathematical concept be explained by this theory?
3) Further understanding would be required for the cognitive underpinnings of the MD circuit. What is its relation with long-term memory for example? I believe that we would hardly call intelligent someone with poor memory, but problem solving an memory capacity are not directly linked. They are, of course, linked in the sense that crystallized knowledge definitely helps in problem solving, but what I am talking here is the ability to absorb new information.
4) While the theory designates a specific brain area as the root of intelligence, it is highly likely that there are specific neural mechanisms (such as speed of signal transmission) that take active part in determining a person’s general intelligence.
Finally, I have to make a comment on the nature of this theory. It is a purely psychological theory and like all psychological theories, it focuses mainly on human intelligence. On the other hand, I prefer more holistic views on intelligence. Therefore, I prefer speaking of intelligent systems. Swarms, ants, humans, chimpanzees, maybe even the stock-market, show intelligent and adaptive behavior. This concept is a root concept in intelligence research that has its roots in computer science, research in fields such as: computational intelligence, AI, machine learning, evolutionary computation.
However, these two areas of research (the one that focuses on human intelligence and the other that has its roots in computer science), rarely communicate, if at all. John Duncan, with his MD theory, makes an important step in recognizing some common elements between classic AI problems and the human intelligence. I think that this is the one element of the theory that I found most intriguing. I believe that John Duncan’s view is through the lens of a psychologist, but I would be interested in seeing how this research could advanced research in AI.
In any case, John Duncan’s "How intelligence happens" is a good read. I would also recommend it to non-psychologists, since Duncan makes a very well-written review of some of the most important researches ever done in cognitive psychology. This book will not only give you a good opinion on intelligence, but, also, it can be a good introduction to cognitive psychology for people outside the field.
