Nelson Goodman’s claim that the method of geometric perspective is a convention is contrasted with E. H. Gombrich’s view that the method possesses objective validity. The status of those compensations made when viewing perspective pictures from ‘incorrect’ positions is discussed in relation to Gombrich’s distinction between the ‘how’ and ‘what’ of visual perception. After arguing that accounts of compensation cannot distinguish between perspective pictures and nonperspective pictures or verbal descriptions of the depicted scenes, the author analyzes the concept of convention. He notes that conventions may be thought of in two different ways. Either a convention is arbitrary or it is the result of innate human characteristics and, therefore, in a sense, objective. He suggests that Gombrich’s writings may support the claim that perspective is an objective method of representation. He places this particular debate within the context of Gombrich’s general view of art history.