The discrimination of magnitude

A number of theories of conditioning are based on the assumption that generalisation gradients either side of a stimulus are symmetrical. From this assumption, the prediction follows that the rate at which a discrimination is acquired between two stimuli will be unaffected by which of them signals the reinforcer (S+). In contrast to this prediction, when S+ and a non-reinforced stimulus (S-) are different in terms of their relative position on a magnitude continuum, this symmetry does not always hold. In many cases, discriminations with a high-magnitude S+ and low-magnitude S- are acquired more readily than when S+ is low-magnitude and S- is high-magnitude. The primary aim of this thesis was therefore to offer an account for this asymmetry. Chapter 2 presents evidence for such an asymmetry with clickers differing in intensity. Chapters 3 and 4 present an asymmetry with arrays differing in the number of black squares on a white background, although it emerged that the mechanisms responsible for this effect were different than for auditory intensity. One possibility, argued in Chapter 5, is that these modalities present contrasting results due to differences in how the stimuli are represented at the receptor level. With this considered, in an attempt to account for both sets of data in terms of theories of conditioning, an adaptation to Pearce´s (1987) configural theory is proposed whereby similarity between stimuli is based on the proportion of common elements, rather than the number of common elements. It is further argued that this amendment provides a more satisfactory account of the reported results than the original theory of Pearce. In addition, the amended theory is shown to account for a wider range of results than the Rescorla-Wagner (1972) model