Range effects and dimensional contrast: Tests of the two-dimensional allocation model

Range effects and dimensional contrast are experimental phenomena indicating that the relation between physical stimulus differences and discrimination performance is not always monotonic. A theoretical model designed to explain range and dimensional contrast effects, the allocation model, proposes that these effects result from a limited-capacity attentional resource. Three experiments were conducted with separable and integral stimulus forms in order to test a two-dimensional extension of the allocation model.;In Experiment 1, human subjects classified visual form stimuli differing in range extension and stimulus type. The allocation model predicted that range effects would be observed in all range extensions with integral stimuli, but only with relevant stimulus spreads of separable stimuli. In fact, range effects were observed whenever range was extended, regardless of the type of stimulus forms, or relevance of the extension. Experiment 2 examined the same stimulus manipulations used in Experiment 1, but with pigeon subjects. The results with pigeons in Experiment 2 were comparable to those observed with humans in Experiment 1.;In Experiment 3, separable and integral stimulus forms were used as redundant and conditional cues in a maintained generalization procedure with pigeon subjects. The allocation model predicted qualitative differences in generalization gradients based upon whether the secondary cues were separable or integral. Integral cues were generally more effective than separable cues, but the differences were not clearly qualitative as predicted.;In summary, the predicted differences in range and dimensional contrast effects with separable and integral stimuli were not observed in these experiments. Procedural modifications may be necessary to produce the predicted differences. Alternatively, separable and integral stimulus organization may not have the originally predicted impact on range and dimensional contrast effects. If so, a revision of the allocation model, in which the spread of attention is not restricted to variations of stimuli along relevant dimensions, may be able to account for the current results.