Discrimination of Temporal Same-Different Relations by Pigeons
J. Gregor Fetterman Arizona State University
To what extent are nonhumans capable of learning discriminations based upon abstract relations between stimuli (e.g., a same-different rule)? This question has generated a great deal of controversy over the years (e.g., Spence, 1937), and controversy continues to surround the issue of relational learning in animals (e.g., see Carter & Werner, 1978; D'Amato, Salmon, & Colombo, 1985; Premack, 1978, 1983). The evidence indicates that pigeons have considerable difficulty learning relational concepts such as the same- different rule (see Carter & Werner, 1978, for a review), whereas monkeys trained on tasks like those used with pigeons (e.g., matching-to-sample-- MTS) appear to base discriminative responses upon a generalized conceptual rule (e.g., D'Amato & Salmon, 1984).
Apparently at odds with the literature on relational learning in pigeons are some recent findings by my colleagues and me ( Dreyfus, Fetterman, Smith, & Stubbs, 1988; Fetterman & Dreyfus, 1986, 1987; Fetterman, Dreyfus, & Stubbs, 1989; Stubbs, Dreyfus, & Fetterman, 1984) indicating that pigeons can learn discriminations that appear to be based upon abstract stimulus relations. In these experiments, pigeons were trained to make different choice responses, depending on which of two successive stimulus durations was longer, with the specific durations changing from trial- to-trial. This task was readily learned, and in light of considerations to be discussed, we believe that relational comparisons are involved. In this chapter, I will review some of these data, present the results of several recent experiments that appear perplexing in the context of our initial findings, and suggest an approach that incorporates the discrepant findings, based in part on a signal-detection framework.