Abstract: The paper considers how one might assess the adequacy of art-historical accounts that exemplify what Michael Baxandall terms the 'inferential criticism' of paintings, where the latter is an instance of what Wollheim called 'criticism as retrieval'. The problem is to reconcile the interpretive nature of such accounts, and their claims to provide us with trans-cultural and trans-historical understandings, with more general 'scientific' constraints on explanation. I draw a parallel with a problem in interpretive ethnography considered by Clifford Geertz, and sketch the latter's claim that the former involves 'generalising within cases' rather than 'generalising across cases'. I offer a 'pragmatic' reading of what Geertz means by 'generalising within cases': on this reading, the goal is to provide an 'intelligible frame' in which we can locate a set of presumptive signifiers, one whose justification resides in its serving the cognitive interests that ground our interpretive practice. I then argue that Baxandall's defence of 'inferential criticism' should be seen as adopting a similar 'pragmatic' approach to defending 'inferential criticism'. In arguing for this, the paper gives an analysis of Baxandall's account in terms of a range of heuristic principles that are justified through serving our interests in art history.

Key words: art history, interpretation, Michael Baxandall, inferential criticism, Clifford Geertz, generalising within cases