Theories of transposition are evaluated in terms of the available empirical data. The most often cited model, that of Spence (1937), is found to be adequate in its explanation of simple transposition and the distance effect. It encounters difficulty in explaining or incorporating the findings of: intermediate size, simultaneous versus successive discrimination, contrast and background effects, and mediated transposition. Other models are considered, with adaptation-level approaches providing the most adequate alternative to the Spence position.
RETROSPECTIn the typical transposition demonstration, subjects (5s) are first trained to respond differentially to two stimuli, and are then tested with one or more new stimuli lying on the same physical dimension. For example, 5s may be reinforced for a response to a stimulus square 256 cm. 2 , nonreinforced for a response to stimulus square 160 cm. 2 , and then tested on stimulus squares 256 cm. 2 and 409 cm. 2 . If S chooses the larger (409 cm. 2 ), he is said to have transposed on the basis of the relative properties of the stimulus situation (relational responding); if S chooses the smaller (256 cm. 2 )-the one to which he has been trained -he is said to have acted on the basis of the absolute properties of the stimulus situation (absolute responding).A number of early investigators demonstrated that 5s do reliably perform in a rela-