Transitive inference formation in pigeons.

Abstract

Pigeons were trained with 4 pairs of visual stimuli in a 5-term series-A+ B-, B+ C-, C+ D-. and D+ E-(in which plus[+] denotes reward and minus(-] denotes nonreward)-before the unreinforced test pair B D was presented. All pigeons chose Item 8, demonstrating inferential choice. A novel theory (value transfer theory), based on reinforcement mechanisms, is proposed. In Experiment 2, the series was extended to 7 terms. Performance on test pairs was transitive, and performance on training pairs accorded with the theory. The 7-term series was closed in Experiment 3 by training the flrst and last items together. In accordance with the theory, the Ss could not solve the circular series. The authors suggest that primates, including humans, also solve these problems using the value transfer mechanism, without resorting to the symbolic processes usually assumed. 1. Edith is fairer than Suzanne. 2. Edith is darker than Lili. 3. Who is the darkest, Edith, Suzanne, or Lili? Here the competent subject concludes that Suzanne is the darkest, although no direct information about the relationship between Lili and Suzanne was given. In this purely linguistic Juan D. Delius and Lorenzo von Fe~n were supported by the Deutsche Forschungsgemeinschaft, Bonn; C. D. L. Wynne was SUJ>ported by the Science and Engineering Research Council, London, and the Alexander von Humboldt.Stiftung, Bonn; and J. E. R. Staddon was supported by the National Science Foundation and the Alexander von Humboldt-Stiftung. We thank W. Mathaus for initial motivation, F. von Mi.inchow-Pohl for assistance in programming, and A. Lohmann for help in running the experiments. We are grateful to A. Elepfandt, B. Mc-Gonigle, J. Pearce, H. Terrace, and T. Trabasso for helpful comments on drafts of this article.

The hippocampus and memory for orderly stimulus relations

“…Previous studies have demonstrated that several animal species can learn an overlapping series of discrimination problems and demonstrate a capacity for transitive inference (15,(26)(27)(28)(29)(30)(31)(32). The present results identify the hippocampal region as critical to transitive inference and indicate that the hippocampus plays a critical role in the development or flexible expres- sion of a representation of orderly relations among stimulus items.…”

The discovery and comparison of symbolic magnitudes

“…An additional phenomenon, typically observed for comparisons involving a closed-set series for which only ordinal information is available (e.g., an arbitrary ordering of elements for which magnitude information is not provided), is a bow-shaped serial position curve: accuracy and decision time indicate greater difficulty for pairs drawn from near the center of the list than for pairs closer to the ends. A bowed serial position curve is not observed for magnitude continua such as those on which we have focused in the present paper, but it is found for arbitrary orderings, both for humans (e.g., Potts, 1974;Trabasso & Riley, 1975;Woocher et al, 1978) and many animal species, including squirrel monkeys (McGonigle & Chalmers, 1977), rats (Davis (1992) and pigeons (von Fersen et al, 1991; for a review see Merritt & Terrace, 2011).…”

“…For example, under certain conditions the Rescorla-Wagner model of associative learning (Rescorla & Wagner, 1972;see Wynne, 1995) can model qualitative aspects of animals' ability to infer transitivity of choice (e.g., after being trained on only adjacent pairs of stimuli exhibiting the reward pattern A > B, B > C, C > D, D > E, an animal will tend to choose B over D). Other associative models can account for learning of orderings across a broader range of conditions (von Fersen, Wynne, Delius, & Staddon, 1991).…”

“…As we have seen, a great variety of species can make comparative judgments based on linear orderings. By contrast, animals have considerably more difficulty learning circular orderings, or rings (von Fersen et al, 1991). The special status of linear orderings is a natural consequence for BARTlet and other models that base comparisons on magnitudes, or some similar unidimensional quantity.…”

Categorical perception and conceptual judgments by nonhuman primates: the paleological monkey and the analogical ape

“…Alternative associative models of transitive inference in both primates and nonprimates have received increased attention in recent years (e.g., Delius & Siemann, 1998;Wynne, 1997). Some such models are based either on direct reinforcement histories of test stimuli (e.g., Couvillon & Bitterman, 1992) or associative value transfer in which the rewarded member of a stimulus pair transfers some of its associative strength or value to the nonrewarded pair item (von Fersen, Wynne, Delius, & Staddon, 1991). Studies of social concepts, which are addressed elsewhere in this issue by Seyfarth & Cheney, provide additional evidence of associative processes being the representation "glue" in monkeys rather than the more abstract propositional encoding of the human investigators (cf., Thompson, 1995;Seyfarth & Cheney, 1997).…”