The term was popularised by Michael Dummett, who introduced it in his paper Realism to re-examine several classical philosophical disputes involving such doctrines as nominalism, conceptual realism, idealism and phenomenalism. The novelty of Dummett's approach consisted in seeing these disputes as analogous to the dispute between intuitionism and platonism in the philosophy of mathematics.
According to intuitionists (anti-realists with respect to mathematical objects), the truth of a mathematical statement consists in our ability to prove it. According to platonists (realists), the truth of a statement consists in its correspondence to objective reality. Thus, intuitionists are ready to accept a statement of the form "P or Q" as true only if we can prove P or if we can prove Q. In particular, we cannot in general claim that "P or not P" is true, since in some cases we may be able neither to prove nor to disprove the statement P.
Dummett argues that the intuitionistic notion of truth lies at the bottom of various classical forms of anti-realism. He uses this notion to re-interpret phenomenalism, claiming that it need not take the form of a reductionism (often considered untenable).
In philosophy of science, anti-realism applies chiefly to claims about the non-reality of "unobservable" entities, entities, such as electrons, which are not detectable with our normal human senses but which many nonetheless claim are real. For a brief discussion comparing such anti-realism to its opposite, realism, see (Okasha 2002, ch. 4). Ian Hacking (1999, p. 84) also uses the same definition.
In discussions of art (including [[Visual arts and design|visual art]], writing, music, and lyrics), anti-realism and anti-realist may be used in one of the philosophical senses described above, or may simply be used in contrast to realism, in whatever sense the latter is meant. Thus surrealism in visual art is an "anti-realist" tendency, and the psychedelic bands common in the United States in the 1960s were "anti-realist," etc. Thus these terms may not be as precise when applied to art as when applied to philosophical matters.