ABSTRACT
As we have seen, Hanson was among the first historian-philosophers of science to suggest that Wittgenstein’s later thinking pointed a way beyond the various dilemmas which – according to critics like Quine – beset the discourse of logical empiricism.1 These included, most famously, the problem of distinguishing synthetic (or empirical) statements from analytic or logical ‘truths of reason’, and the impossibility – as Quine saw it – of testing empirical statements one-by-one against items of observational evidence.2 Hence his verdict that no statement should be held immune from revision, whether those at the empirical edge of our belief-fabric (where perceptions might be subject to some distorting influence) or those at its very centre (where even the logical ‘laws of thought’ might need to be revised under pressure from certain physical discoveries such as those of quantum mechanics). Rather, we should think of the fabric as ‘underdetermined by its boundary-conditions’ and of logical axioms as open to suspension or pragmatic adjustment whenever this helps to facilitate the conduct of scientific enquiry. For there is always the option of redistributing truth-values over the fabric as a whole so as to save some cherished theory by discarding certain anomalous empirical results or, conversely, to save those results by revising some hitherto sacrosanct ‘law’ of logic in the interests of empirical adequacy. Thus science faces the ‘tribunal of experience’ as a whole and cannot be thought of – in logical-empiricist terms – as comprising on the one hand observation-statements that meet the criteria of straightforward empirical warrant and on the other hand theories or covering-law statements whose validity consists in their properly conforming to the ground-rules of logical thought. In which case one has to let go of the idea that a theory or hypothesis might be falsified through some crucial experiment that throws up discrepant results. For the theory can always be conserved by invoking different auxiliary hypotheses or again – if all else fails – by revising one of the logical ground-rules (e.g., bivalence or excluded middle) in order to avoid any conflict with the range of well-attested empirical data.
