BEYOND VERISIMILITUDE - A LINGUISTICALLY INVARIANT BASIS FOR SCIENTIFIC PROGRESS

This paper proposes a solution to David Miller's 'Minnesotan-Arizonan' demonstration of the language dependence of truthlikeness (Miller 1974), along with Miller's first-order demonstration of the same (Miller 1978). It is assumed, with Peter Urbach, that the implication of these demonstrations is that the very notion of 'truthlikeness' is intrinsically language dependent and thus non-objective. As such, 'truthlikeness' cannot supply a basis for an objective account of scientific progress. I argue that, while Miller is correct in arguing that the number of true atomic sentences of a false theory is language dependent, the number of known sentences (under certain straightforward assumptions) is conserved by translation; 'degree of knowledge', unlike 'truthlikeness', is thus a linguistically invariant notion. It is concluded that the objectivity of scientific progress must be grounded on the fact (noted in Cohen 1980) that knowledge, not mere truth, is the aim of science. © 1991 Kluwer Academic Publishers.