The New Logic

The purpose of this paper is to communicate some developments in what we call the new logic. In a nutshell the new logic is a model of the behaviour of a logical agent. By these lights, logical theory has two principal tasks. The first is an account of what a logical agent is. The second is a description of how this behaviour is to be modelled. Before getting on with these tasks we offer a disclaimer and a warning. The disclaimer is that although the new logic is significantly different from it, we have no inclination to see the new logic as a rival of mathematical logic. We do not advocate the displacement of, e.g. model theory, but rather its supplementation or adaptation. The warning is that, whereas mathematical logic must eschew psychologism, the new logic cannot do without it. The fuller story of the new logic is part of our book, The Reach of Abduction, scheduled to appear in 2001 or early 2002. 1 The Sort of Thing a Logical Agent Is 1.1 Psychologism and the New Logic Is there a logic of discovery? Some say not. Others are not so sure. We ourselves are in a third camp: In work underway we are actually trying to codify such a logic. Critics of the logic of discovery, those who think it a misbegotten enterprise as such, are frequently drawn to the idea that accounts of how people entertain and select hypotheses, form and deploy conjectures, and more generally how they think things up, are a matter for psychology. Underlying this view is something like the following argument. Let K be a class of cognitive actions. Then if K possesses an etiology (i.e. is subject to causal forces), this precludes the question of the performing or disperforming the K-action for good or bad reasons. If there were a logic of K-action it would be an enquiry into when K-actions are performed rationally, that is, for the right reasons. Hence there can be no logic of K. Against this, Donald Davidson is widely taken as having shown that far from reasons for actions precluding their having causes, reasons are causes, or more strictly, having a reason for an action is construable as a cause of it. (Davidson [14]. See also Piestroski [54].) We ourselves are inclined to emphasize a substantial body of work in reliabilist and other forms of causal epistemology. In its most basic sense, a subject performs a cognitive task rationally when his performance of it is induced by causal mechanisms that are functioning reliably, that are functioning as they should. If a logic can be seen as a theory of rational cognitive performance, the present argument fails to demonstrate the impossibility of a logic of discovery. Even so, the idea of logic as a 141 L. J. of the IGPL, Vol. 9 No. 2, pp. 141–174 2001 c ©Oxford University Press