Tuesday, March 11, 2014

Wittgenstein among the Sciences

Although Margaret Thatcher claimed that there is no such thing as society I don't think anyone denies that there can be such things as social studies. The question is whether there can be a social science and, if so, what form this ought to take. In Matters of the Mind (pp. 220-230) William Lyons argues for replacing the conception of science that regards physics as the paradigm with the older idea of scientia. Lyons follows Samuel Johnson's 1755 definition of the word 'science' to come up with this:
any rigorous method which produce[s] certainty via demonstration (p. 226),
which means that you arrive at
exact knowledge simply by employing valid deductive reasoning on given premises or axioms [...] or [...] via both observation and reasoning on the basis of the observation (p. 227)
If this is our idea of social science then there can (it seems to be universally agreed) be a social science. What Rupert Read denies is that there can be a social science if we take the narrower notion of science instead. Like Lyons, Rupert recognizes that "our paradigms of science" (p. ix) are the natural sciences.

I should note that Rupert intends his book to be therapeutic, to be used by the reader to help clarify his/her own thoughts, not dogmatic. His opinions, he says, are irrelevant (see pp. xi-xii). I think it is pretty clear, though, what his opinions are, and it is these that I will tend to focus on. This is somewhat unfair, but I think it will be more productive than proceeding otherwise. And I won't attribute any opinion to him without good reason to do so.

The part of Rupert's book that most interests me is the second one (the first is about Kuhn), in which he goes through a set of cases, starting with schizophrenia. It seems to me at least possible in principle that conditions like schizophrenia might have a biological cause or, at least, be susceptible to biological treatment. We might find, for instance, that electric shock treatment or some pill improves the lives of people with the condition. Such improvement is surely a noble goal for psychology or psychiatry to have, and seems like just the kind of thing that science is good for. On the other hand, schizophrenia is not just a natural phenomenon. It is something from which people suffer. And these people can talk. Their speech and behavior is, at least in principle, intelligible, and it is as it affects their speech and behavior that schizophrenia shows up as a problem. So a purely physical response is not necessarily best. Perhaps therapy might help in some cases. This takes us away from the physics paradigm and into the arena of understanding other people, which is closer to philosophy. In some cases, Rupert suggests, a schizophrenic person will not be intelligible at all. Recognizing this means telling sense from nonsense, and this is a philosophical, conceptual task, not a scientific, empirical one.

In the next chapter, similarly, he takes on extreme aversive emotions, such as dread. A scientific approach might help us to treat such emotions, but if we want to understand them then we need to get to grips with what people mean when they describe the reality of their surroundings fading, or the feeling of being walled off from other people, or feeling unable to believe that the world is real. This involves understanding ideas (or perhaps seeing them to be nonsense, not ideas at all), which again is not in the province of science but of the humanities and, arguably, philosophy in particular.

He turns to economics next. One model of economics involves making predictions about human behavior on the basis of game theory or rational choice theory plus the assumption that people tend to maximize their utility. Rupert (p. 148 note 6) suggests that this assumption is pseudo-scientific and/or morally corrupting. Another problem, though, is that predictions reached in this way might turn out to be false. If they turn out to be true then I don't see anything very pseudo-scientific about the approach. If they turn out to be false and the approach is not altered or abandoned then that looks like pseudo-science. As for the idea of people as utility-maximizers as corrupting, I don't know. If the idea is actually useful then it seems dishonest to refuse to use it because of moral concerns. It's only meant to be an approximation, after all, and if it does approximate to how we behave then why deny this? It if doesn't, of course, then it would be perverse to persist. The relevant questions seem to be about what works. Very sensibly, Rupert recommends that we study economic history and the actual practices of businesses and the people who work in them. I suppose, though, that economists are interested in the economic behavior of everyone, not just those who work for private corporations, and this might be too large a field for anyone to study. One way to break down the field of human behavior into more manageable chunks would be to generate hypotheses first and then find out whether they are true. This is roughly what economists do, I take it, although it's not clear how responsive they all are to empirical findings. Some, I suspect, cling to their favored models, assumptions, and hypotheses. Nor is it clear what to do with such findings given the complexity of real life and the seeming impossibility of conducting controlled experiments in economics. Behavior in the lab might well differ from behavior in the real world, and any real world finding that deviates from the hypothesis can be attributed to irrelevant factors. "True, increasing the minimum wage in this case did not lead to more unemployment, but that's because...", etc. Which allows people to keep clinging to their pet theories. Does this actually happen? I need to look into that.

I have only summarized a few points from the book, but I think they are good points.


  1. I'm not sure it will help but some of Read’s concerns vaguely remind me of some concerns Hannah Arendt has about the social sciences. (see mostly in her Essays in Understanding: 1930-1954). I wonder if Read mentions her.

  2. And also relevant there is Jon Elster's "Functional Explanation," in Explaining Technical Change.

  3. Thanks! I don't think he mentions Arendt or Elster. Neither appears in the index.