Another talk on Simon

Posted on April 7, 2008


I have been accepted to give a talk at the Pragmatism and Naturalism workshop in Tilburg next month. It’ll be closely related to the “Good habits, better heuristics” paper that I’m to give in several places but, due to the focus of the meeting, will also deal with Herbert Simon’s relation to pragmatism:

Simon’s heuristics, reliabilism and habits

It is not an original claim to say that Herbert Simon, the American polymath scientist, has also made a highly significant contribution to philosophical pragmatism and naturalism. Much the same has been claimed by Stephen Stich. However, rather than pursue Stich’s somewhat superficial analysis of what Simon achieved in his empirical work, I would rather take a different tack and examine Simon’s central concept – heuristics. My contention is that with heuristics, Simon has formulated an account of rationality that is profoundly naturalist and pragmatist. That this is the case can be understood once some of the characteristics of heuristics are considered.

The first of their simplicity. Heuristics are simple rules of thumb in order that limited cognisers such as human beings may use them to make decisions quickly enough to react to their changing environment. This care for the reality that human (or indeed any imaginable) rationality is bounded should clearly be a part of a naturalist account.

The second is their context-dependence. The simplifying assumptions that heuristics make about the context they will be used in can turn out false, in which case the heuristics will not provide even approximately correct answers. This, in effect, means that heuristics can only ever be justified a posteriori – in line with the rejection of the a priori that is common to many naturalist positions.

The third is their systematic bias. Even within their appropriate contexts-of-use heuristics, again due to their simplifying assumptions, make systematic errors. However, this bias is not a problem so long as the results provided by a heuristic are good enough for the particular purpose to which that heuristic is being employed. Indeed, the whole point of using heuristics is not to get the best possible result but one that is satisfactory. This choice of satisficing various (not necessarily epistemic) values over optimising a single value such as truth seems be to deeply in tune with many pragmatist accounts of reason.

Something of the way that Simon’s heuristics help to fill out other naturalist and pragmatist accounts can be seen by comparing them to just two important examples: Goldman and Hume.

In the case of Goldman, Simon’s heuristics can be understood to be the reliable methods he advises us to use. Importantly, where Goldman to a certain degree eschews too much talk about what the precise methods are, his reticence can be understood very well in contrast to Simon’s empirical project of identifying and inventing heuristics. The reticence can be seen as appropriate to a philosophical account that aims to work hand-in-hand with the empirical approach – as such objections to Goldman that he ‘leaves out’ what is most important come to be seen as misplaced.

In the case of Hume, Simon can be seen as spelling out the habits that Hume recognises as part and parcel of the normal way people go about thinking. Given that Quine has expressed the naturalist project as trying to explain how we manage to come to grips with our world without solving Hume’s problem, this is perhaps the best context in which to understand what Simon has achieved.