The correspondence theory of truth holds that a belief or sentence is true when it corresponds to a fact — when the way it describes the world matches the way the world actually is. Rorty argues this account is empty: we have no way of stepping outside our beliefs and comparing them to bare unconceptualised reality. What we call "checking a belief against reality" always turns out to be checking a belief against other beliefs, or seeing whether acting on a belief gets the results we hoped for. The correspondence theory adds nothing to our practices of inquiry.
Rorty endorses the deflationary account of truth: to say a belief is true is not to ascribe a property to it (the property of corresponding to reality) but simply to commend it — to endorse it, to say it is what one ought to believe given current evidence. This is consistent with taking truth seriously — false beliefs are dangerous and careful inquiry is important — but it decouples inquiry from the metaphysical project of mirroring the intrinsic nature of reality. We aim at beliefs that are well-supported and practically effective, not at beliefs that reflect the Deep Structure of Things.
The anti-representationalist picture replaces the image of the mind as mirror with the image of the organism as tool-user. Language, thought, and inquiry are instruments for coping with the environment and achieving our purposes — biological and social tools that evolve and are refined in use. Successful inquiry is not convergence on the Truth but the production of beliefs that serve us better than the ones they replace. Rorty insists this is not a counsel of irrationality: the pragmatist is as committed as anyone to evidence, argument, and the norms of inquiry — she just refuses to cash these norms in terms of mirror-accuracy.
Anti-representationalism is the central thread of Consequences of Pragmatism (1982) and is most fully developed in Objectivity, Relativism, and Truth (1991).
