INTERVIEW

Stanislas Dehaene

Interviewed by Laura Spinney

This article also appears in BioMedNet's Conference Reporter.

Posted June 22, 2001 · Issue 105

Background

Biography
Stanislas Dehaene started his career as a mathematician at the Ecole Normale Supérieure in Paris, but quickly turned to neuroscience. He combined studies in cognitive psychology, neural modeling, and brain imaging, making him unusually qualified to study the neural substrates of numerical knowledge and reasoning. He is a research scientist at the French medical research organization INSERM, and directs research at the imaging center at the Frédéric Joliot Hospital in Orsay, France. Most recently he and his colleagues have been working on the interdependence of calculation and language, as well as the pathways that are unique to each.

What event led you into research?

As a teenager, I was intrigued by math and by the brain. I was intrigued not just by the mathematics themselves, or how you do mathematics, but also why they take the form they do - the foundation problem that is at the heart of mathematics. And as I was learning high-level mathematics, I became interested in the kind of mental activity that was behind it. So I stopped at the master's [in applied mathematics and computer science] and went into a Ph.D. in cognitive psychology.

Who has most inspired and/or influenced your work?

I studied under two important people. One is Jacques Mehler, who is the editor of Cognition and a very bright psycholinguist. The other is Jean-Pierre Changeux, the well-known molecular neurobiologist. I learned my neurobiology from one and my cognitive psychology from the other, and the combination of the two interested me. The main stance in cognitive psychology at the time, in the early 1990s, was a functional one, whereby it was deemed not important to look at the brain. It was thought sufficient to look at the mental processes without caring about their brain representation, and I always objected to that. Now we have these wonderful brain imaging tools that actually allow us to combine the two approaches.

Who awarded you your first grant and what was it for?

The French system doesn't work that much with grants. But in 1999 I was stunned to receive a James S. McDonnell centennial grant [the James S. McDonnell Foundation was set up in 1950 by the aerospace pioneer and provides grants to support science that "improves the quality of life"], which was a million dollars to pursue the work on mental representation of mathematics.

What was your best idea or theory?

An intriguing idea that I am working with just now is that we can access the meaning of a number without consciousness. So we look at behavior and brain activation to subliminal digits and we can show that there is a whole stream of processing that is taking place, even for a digit that you are not aware of seeing because it has been masked - presented very briefly on a screen. We have been able to look at the entire stream of processing, decomposing the different steps. Now the big question, of course, is what is different when you are aware. It seems that when you become aware you have access to a much broader system of information broadcasting, what we call a neural workspace, which involves largely the frontal lobes but other areas as well. Once the information is in that system it is available to all sorts of different processes, breaking the modularity of the brain as we know it.

Which scientific idea (yours or others) do you regret the most?

In neurology for a long time there was the idea that even neuropsychology, the study of brain lesion patients, could be done not by looking at the site of the lesions, but just by looking at the pattern of behavior, the impairments, and trying to derive from those the impaired representations. There was this idea that it was not useful to look at the lesions because they were very broad, and the brain was not very well organized, so impairments could not be localized easily. All this turned out to be quite wrong. There is a lot of correlation between the site of the lesion and the nature of the impairment. It is very useful to look at multiple lesions and see how they cross-correlate, how they intersect. That way you can learn about the anatomical substrate of a behavior or impairment with a great deal of precision. We now even have wonderful tools for studying the connections between brain regions. There is a new technique called diffusion tensor imaging which allows you to trace the connections anatomically in a living person. With this the discipline is coming together and there is no need for the distinction between cognitive studies on the one hand and neuroimaging on the other. They are all one discipline.

What are your scientific plans for the next five years?

I have two big plans. One is to study the development of calculation, to try to see its roots and how the brain is changed by learning to calculate - looking at education from a cognitive perspective. The second is to try to understand better this transition between unconscious and conscious processing.

What are the qualities of a successful researcher?

Being bald! Seriously, though, in my case it may be creativity. I didn't want to follow the well-trodden path. But there are many different styles. You can also dig into a subject, and continue to dig and to dig, and reach the nuggets in the end. What is for sure is that you need to read a lot. The field has become enormous now, and because there is this cross-disciplinary aspect, lots of things become relevant. You cannot encapsulate your knowledge nicely in a small area. More people are needed from mathematics and physics. We are in a field which is moving into very big science now. I used to work with just pencil and paper, then reaction time measurements, and now we are all working with imaging machines that cost tens of millions of francs [millions of U.S. dollars]. So more and more we will have to work in consortiums which bring together very good technical people with psychologists and neurologists.

If you could work with any scientist (historical or current), who would it be?

As far as neurology is concerned, I think Joseph Jules Dejerine would have been a very interesting man. He discovered alexia [the inability to read]. In 1891 he wrote the first paper on it. That was over a century ago, and yet it has all the accurate features that people are rediscovering now.