Cognitive Science and the bad logic student

This article is an opener for a conversation I've been wanting to have for a while. Hutchins suggests that the failure of some otherwise bright, hardworking students to do well in logic is due to a heretofore unknown learning disability she calls "inference blindness." I am quite convince that there is a class of students who do poorly in logic classes which cannot be attributed to general problems in learning, like a lack of effort or general study skills. Normally the poor performance of these students is attributed to some sort of math-phobia. Hutchins doesn't consider this possibility, which I'm ok with, because I don't think math-phobia can account for all of the phenomenon we are looking at. This leaves us with a substantial body of students whose poor performance we don't have a good explanation of. I would really, really like an understanding of these students that is grounded in some real cognitive science.

Hutchins' proposal doesn't go far at all in this direction. She doesn't so much ground her hypothetical in the cognitive science as offer one way a learning disability like inference blindness might be realized in one model of the neurology of inferential reasoning. She does not say that the model predicts the existence of inference blindness, nor does she say that putting inference blindness in the context of this model leads to some other prediction. The model in question is Vinod Goel's dual mechanism theory, which says that we reason about logical connections using a combination of two different systems. One is a visual-spacial system that is used in unfamiliar or highly abstract situations. In terms of hardware, this systems is based on a bilateral occipital-parietal-frontal network. The other system is a linguistic system based in the left frontal and temporal lobes and comes on line when logical problems are placed in concrete, familiar situations. Part of the advantage of this model is that it explains performance on the Wason Selection Task. Subjects do better when reasoning about the conditional "If you are drinking alcohol then you must be over 21" because the second system is brought online. Hutchins thinks inference blindness, if it exists, would be cause by a neurological deficit, perhaps marked by real physical lesions, which prevents the linguistic system from coming to the aid of the spatial system. I'm not sure why she wants the deficit to be specifically located in the connection between the two systems. Why not say that the students in question have a weakness in just the abstract spatial system? We know this system is weak in everyone, but it is still possible for their to be lots of individual variation. Alternatively, why not say that the deficit is just in the verbal system? Perhaps there are many distinct kinds of deficits out there?

Another problem with the proposal is the use of the language of disability and talk of specific physical lesions. For starters, it creates the risk that the class of students we wind up identifying is too narrow to be of any use to us. In my fourteen years of teaching logic, I have had precisely one student whom I could confidently say struggled with logic because of a neurological deficit. He had been the victim of a gunshot wound to the head. The doctors had initially said he would never walk or talk again. As it turned out, he could do both, and was one course away from graduating college. Unfortunately, that course was not going to be symbolic logic.

Of course, not all learning disabilities wind up picking out a very narrow class of people. Experience with disabilities like ADHD and dyslexia show that once a learning disability gets picked up by the media, its definition can be blurred to the point of complete meaninglessness. But this is also a bad outcome, if what we are seeking to discover is a substantial body of students that can be helped with a concrete intervention. When I taught at an extremely elite private school (which we will call Stuffwhitepeople Like University) huge numbers of students, sometimes as much as a third of the class, would come to me with documentation of a learning disability that would give them accommodations like time and half on tests. None of these students ever needed the accommodation. As near as I can tell, they had just learned how to game the system to get every advantage for themselves. Their training in this probably began in infancy, while their parents were intently reading New York Times articles on how to get your kids into the "right" preschool. These days I teach at a community college which serves a lot of severely underprivileged students and I am also seeing a lot of students who deserve and would benefit from the kinds of accommodations that come with a diagnosis of ADHD. Only very rarely do I get students who will ask for accommodation, can document their disability, and be able to get the help they need.

Rather than talking about disability here, I think it would be more productive to talk about a subpopulation of normal students who do poorly in logic for reasons not related to general scholastic ability or some kind of math-phobia. This will allow us to look for a class of students that is larger than the topic of a single case study of a neurological deficit, while avoiding the extensive political complications that come the language of learning disabilities. To be clear, its not that I doubt reality of learning disabilities. I just think that the way the American educational system addresses learning disabilities is hopelessly dysfunctional.

There are other issues that come up in deciding what kind of class of student we should be looking for. I've already said I'm interested in a broader set of students than just the handful that can be categorized as having a physical brain lesion. Another question is whether we want to specifically target students who don't improve their performance when familiar contextual information is added. Hutchins' idea for Inference Blindness was that it involved the failure of the verbal system to help the visual-spatial system. This means that we are looking at people who continue to do badly on the Wason Selection Task even after concrete details are added. This is definitely an interesting class of people, but are they the ones we want to intervene with in our classes? If the alternative is simply looking at students who do badly at the abstract version fo the WST and improve on the concrete version, then the answer is "yes," because the latter category is basically everyone.

The next step in this conversation is to look for, or develop, correlation studies. It would be interesting to see, for instance, whether doing well in logic without extra help correlates with doing well in the abstract version of the Wason Selection Task. Similarly, it would be interesting to see if students who do poorly even with extra help also fail to respond to the addition of concrete detail in the WST. It may be hard to identify correlations here, though, simply because success in a symbolic logic course takes you into realms of abstraction far beyond recognizing the truth conditions for a single if...then statement. Also, even though undergraduates are typical subjects for psychological experiments, actual empirical work on the process of education always seems to run into many more difficulties. It is easy enough to set up a single psych test for undergraduates to take, but tracking them through a full course introduces all sorts of other confounds.