Productive Knowledge and the Social Sciences — Some Pragmatic Ruminations
A monthly Hayek seminar at the London School of Economics on risk and uncertainty sets a reliably high standard for intellectual stimulation, with quite the diverse constellation of luminaries (and just a few stragglers like me).
Most regard themselves as social scientists; I don’t see myself as a scientist, in the modern sense: I like to study social phenomena from a pragmatic perspective (in the William James sense, of what’s useful rather than what’s “true.”)
This week’s session further clarified what these two perspectives mean for thinking about economic models.
The ‘classical’ economic scientist’s approach, going back to J.S. Mill, was to start with axioms, and deduce what “tendencies” followed. Mill for instance defined economics as the deductive science of inferring tendencies that result from seeking wealth. He emphatically did not assume that wealth seeking was the only – or even dominant or uniformly distributed — motivation. Therefore, the tendencies deduced might not actually be observed in specific instances or even in the aggregate.
We have of course come a long way since then, both in the sophistication of deductions and efforts for empirical validation. This has prompted new deductive models — and backward induction from observation. This effort would include behavioral and evolutionary economists (of various stripes), the Santa Fe complexity folk, and George Soros’s ‘reflexivity.’
For all that, I think it’s fair to say that the gap between models and observations remains vast and most everything – as far as I can tell — is over determined (with many plausible explanations for the same thing).
The most that ‘scientific consensus’ can realistically expect is some kind of ‘abductive’ generalization: the “best” explanation for the widest possible phenomena. And I’m skeptical that without some kind of intellectual bullying such a consensus is possible.
The alternative ‘pragmatic’ enterprise (in the William James sense) looks for whats useful rather than what’s universally true — and thus (as John Kay and Mervyn Kay argue in Radical Uncertainty) to ‘situational’ abduction: the best possible inference about a particular circumstance. They further argue that such abduction requires a kind of “plumber’s tool kit” of models. Charlie Munger (of Berkshire Hathaway fame) says this as well in arguing for a diverse ‘latticework’ of models.
This prompts the further pragmatic question: which tools, used under which circumstances? Plumbers learn this in trade school, through apprenticeships, learning by doing, learning by watching (again to borrow from Mervyn King) etc. Yet Knightian uncertainty (perhaps short of ‘radical’) about tool selection and use seems unavoidable.
I think social scientists have something to bring to the pragmatist’s table: more suggestive tools and heuristics for their selection and use. It would be a pity if this were lost in a dogmatic striving for the “best” model and approach.