The fact that I have chosen to employ a single graphic symbol, an arrow, could be taken to imply that 'agency' has some quintessential, generic form, of which the various types of agency so far mentioned are species. This inference would be incorrect; the agency arrow implies no particular kind of agency, only the polarity of agent/patient relations. I set no limit whatsoever to the type of 'action' involved. Sometimes this action is psychological; for example, the 'action' of an index in impressing a spectator with its technical excellence, or arousing the spectator erotically; while at other times the action may be physical, as happens, for instance, if the index is a holy icon which cures the rheumatism of the one who kisses it, rather than merely looks at it. Conventional 'theories of art' are mostly predicated on one, or a limited selection, of 'kinds of agency'. Thus, aesthetic theories of art are predicated on the idea that artists are exclusively aesthetic agents, who produce works of art which manifest their aesthetic intentions, and that these intentions are communicated to the public which views their works in the light of approximately the same set of aesthetic intentions, vicariously entertained. In an ideal art world, such might indeed be the case, and nobody would have recourse to works of art with anything in mind except the garnering of aesthetic experiences, and certainly not in the hope of being cured of rheumatism. Semiologic or interpretative theories of art assume that works of art are vehicles of meaning (signs, symbols) which spectators have to decode on the basis of their familiarity with the semiological system used by the artist to encode the meanings they contain. I do not deny that works of art are sometimes intended and received as objects of aesthetic appreciation, and that it is sometimes the case that works of art function serniotically, but I specifically reject the notion that they always do.
The kind of agency exercised in the vicinity of works of art varies considerably, depending on a number of contextual factors. In gross terms, it may be supposed that whatever type of action a person may perform vis-U+00EO-vis another person, may be performed also by a work of art, in the realms of the imagination if not in reality -- not that we are always in a position to decide what is 'real' and what is not. The anthropology of art, to reiterate, is just anthropology itself, except that it deals with those situations in which there is an 'index of agency' which is normally some kind of artefact.
Gell's main concepts are agency, index, prototype, artist and recipients. Agency is mediated by indexes, that is material objects which motivate responses, inferences or interpretations. Indexes can stand in a variety of relations with their prototypes, artists and recipients. Prototypes are the objects or persons that indexes represent or stand for, mimetically or non-mimetically, visually or non-visually. Recipients are those who are (or are intended to be) affected by the indexes. Artists are those persons considered to be the immediate cause or author of the existence and properties of the index.