Link Search Menu Expand Document

We Can Have Retrieval-Inference Synthesis

Seong-Young Her

Subject matter of art criticism (i.e. "what is art?")MethodologyPurpose
Croce-Collingwood ThesisInternal state of the artist expressed through artefactsStrategic empathy to help readers accurately "re-enact" the artist's inner state expressed by the artworkReproduce the (internal) work of art (cf. the art object)
Wollheim's retrieval criticismA type-token pair consisting of an artwork and its constituent art objects"Retrieval" (cf. "scrutiny") of the creative process of the artist using relevant "cognitive stock"Perceive the meaning of the (external) work of art more accurately
Baxandall's inferential criticism"A concrete solution" to an aesthetic problem addressed by the artist"Inference" of the "intentionality" (cf. "narrative") which led to the production of the art objectExplain the design of the artwork as a solution to a particular problem

Inferential criticism is an art-critical methodology introduced by Baxandall (1985). The purpose of art criticism, according to inferential criticism, is to explain the design of the artwork in light of the problems addressed by the artist through the work. Although Baxandall explicitly denies that he is addressing the question “What is art?” (Baxandall, 1985, viii), his account of inferential criticism is nonetheless based on his ontology of artworks as embodied solutions to particular aesthetic problems.

Wollheim’s (1968) attempt at answering the same question, in Art and Its Objects, and his own theory of “Criticism as Retrieval” (Wollheim, 1980), based on his ontology of art as type-token pairs, provide an ideal point of comparison. Wollheim characterises inference as diametrically opposed to retrieval, and criticises it on methodological grounds as well as purposive grounds (1984). Wollheim’s attack on inference is devastating in that it threatens to render it either tautological or totally detached from truthfulness.

Despite Wollheim’s strong opposition to inference (Wollheim, 1984), I find that Baxandall’s sense of inference escapes the criticisms, and instead a powerful synthesis of the two methodologies is possible. If art is problem solving, then explaining the solutions would be perhaps the most efficient means of encouraging further production of art. Inferential criticism promises to be the most productive method of art criticism.

“[Criticism] is the process of coming to understand a particular work of art.” (Wollheim, 1980, p. 185). By this, Wollheim means specifically to reject what he calls the Croce-Collingwood thesis, which claims that criticism is the process of reproducing the internal state of the artist such that the critic’s readers may also share the correct experience (Wollheim, 1968, p. 56). On this point, both inferentialists and retrievalists agree.

Yet, Baxandall does so by following Collingwood (and Popper) in characterising “purposive” human behaviour, including art, as taking the form of “procedural pattern of problems and situations and solutions”. Therefore, the inferential critic, in understanding a work of art, must “reconstruct both the specific problem it was designed to solve and the specific circumstances out of which [the artist] was addressing it.” (Baxandall, 1985, pp. 14-15). Inference, as Baxandall demonstrates through the example of Baker’s Forth Bridge, is a kind of reverse engineering.

Against the criticism that artists do not generally think of their artworks as solutions to problems, Baxandall responds by modifying the concept of “intentionality” as “not a reconstituted historical state of mind,” but “a relation between the object and its circumstances” and “referred to pictures rather more than to painters” (Baxandall, 1985, p. 42). Baxandall is thus able to analogise art with logic: “in logic one technical sense of ‘problem’ is the question implicit in a syllogism… one could discourse in the propositional style of the syllogism without ever actually formulating the problem as a question” (Baxandall, 1985, p. 70). Given Baxandall’s concession that the creative process involves “a numberless sequence of developing moments of intention” (ibid., p. 63), this methodology resembles Peirce’s abductive reasoning, which begins with some conclusion (namely, the artwork) and attempts to infer the most likely set of true premises (namely, the design) which led to that conclusion.

Wollheim, on the other hand, claims that “understanding a work of art is not an inferential activity or a matter of arguing from certain premisses to a conclusion: it is an activity of perception… critically relevant information serving as cognitive stock, and perception of meaning.” (Wollheim, 1984, pp. 251-252). But because Baxandall’s inferential criticism is Peircian in form, it begins with the conclusion and works its way backwards towards premises, thereby escaping Wollheim’s attack. Whereas inductive and deductive logics argue from premises to conclusions in a way that narrows the parameters for the possible range of next steps in the inference, abductive logic is open-ended and can continually generate alternative hypotheses about the premises that led to the accepted conclusion. The inferential critic therefore begins with the art object as the conclusion and the intentionality as the possible set of premises which best explain the artwork. The first premise is always known to the inferential critic: purposive human behaviour takes the form of problems, situations, solutions.

Wollheim’s own Peircian influence provides the point of possible convergence for the two theories. In Art and Its Objects, Wollheim advances the ontology of art as Peircian types and art objects as Peircian tokens. He clarifies that types are not classes (with members) or universals (with instances) (Wollheim, 1968, pp. 90-91). This clarification is vital, as Wollheim claims that “identifiable works of art constitute a historical, not an ideal, set.” (ibid., p. 160).

The same kind of debate has persisted in the philosophy of biology over the ontological status of species. Wollheim’s ontology of art closely resembles David Hull’s ontology of species as individuals (1978), as does his characterisation of art as a Wittgenstinian form of life. According to Hull, species are not classes (such as planets) but “spatiotemporally localized cohesive and continuous entities” (such as Mars). (ibid., p. 335). In retrieval criticism, the artist and the artwork constitute precisely such a historical entity throughout the creative process; the creative process, while not an organ of the artwork, is nonetheless continuous with it in a way that makes it relevant to the art critical process.

For Wollheim, this suggests “a recursive method of identifying works of art”: that we “seek, not a definition, but a general method for identifying works of art… that we should, first, pick out certain objects as original or primary works of art; and that we should then set up some rules which, successively applied to the original works of art, will give us (within certain rough limits) a subsequent or derivative works of art.” (1968, p. 159). This method is identical in form to Baxandall’s inferential criticism, only recursive: beginning with a particular art object, infer the design through which the work of art may have been produced. Then, apply the solution in order to produce derivative works. It also resembles the use of a biological type, which is a particular specimen of an organism formally selected to serve as the exemplar for the species.

Inferential criticism is, in turn, identical in form to adaptationism in the philosophy of biology. Adaptationists infer causal explanations for heritable traits in an organism based on the assumption that the persistence of those traits throughout generations must have contributed to the reproductive advantage of the ancestral organisms, since they would have been selected out of existence otherwise. While not without challenges (see Lewens (2004) for a comprehensive treatment of the debate), the success and productivity of this methodology in biology and the philosophy of biology are indisputable.

All of this is to suggest that retrieval and inference are compatible, and that methodologies resembling the two have proven wildly successful in at least one other domain which deals with entities that can reasonably be characterised as ontologically equivalent (and equally ill-defined). Robin Haack, considering the influence of pragmatism on Wittgenstein, suggests that we could “view Wittgenstein’s naturalism as standing to the pragmatists’ as Kepler’s descriptive work on planetory [sic] motion stands to Newton’s explanatory theory of gravitation.” (1982, p. 163). Following this, I suggest that we view Wollheim’s retrieval as standing to Baxandall’s inference as Mendel’s descriptive work on genes stands to Darwin’s explanatory theory of natural selection (that each were both descriptive and explanatory is only more fitting). Wollheim mused that “if we could lay down the rules in accordance with which the historical derivations have been made, we should have a theory which not merely was comprehensive of all works of art, it would also give us some insight into their formation.” (1968, p. 160). A synthesis of retrieval and inference may provide a path towards this goal.


Baxandall, M. (1985). Patterns of Intention. Yale University Press.

Haack, R. (1982). Wittgenstein’s Pragmatism.

Hull, D.L. (1978). A Matter of Individuality. Philosophy of Science, 45(3), 335-360.

Lewens, T. (2004). Organisms and Artifacts: Design In Nature and Elsewhere. The MIT Press.

Wollheim, R. (1975). Art and Its Objects. New York: Peregrine Books.

Wollheim, R. (1980). “Criticism as retrieval”. Art and Its Objects (2nd ed.). Cambridge University Press.

Wollheim, R. (1984). Art, Interpretation, and the Creative Process. New Literary History. Vol. 15, No. 2, Interrelation of Interpretation and Creation, 241-253.