Comparative Perception and Attention Laboratory

Do animals perceive the relations between or among stimuli?  Perhaps the most famous chapter in the story of relational discrimination learning in animals comes from the case of transposition, first studied by Wolfgang Köhler

A figure above shows a typical transposition task. A choice of a previously reinforced circle in the test indicates control by exact properties of the stimuli and is termed absolute response. In contrast, ignoring a previously reinforced circle and selecting a new circle suggests learning to attend to relations among the stimuli (“smaller is correct”) and is termed relational response. 

Although Spence (1937) has shown that the choice of relationally correct stimulus can be explained by interaction of absolute response tendencies, my prior research conducted in collaboration with Ed Wasserman (University of Iowa) and Mike Young (Kansas State University; Lazareva, Miner, Wasserman, & Young, 2008; Lazareva, Wasserman, & Young, 2005) have demonstrated that it is possible to design an experiment in such a way that interaction of absolute response tendencies cannot explain subjects’ choices. In other words, transposition paradigm is appropriate for studying relational responding in nonverbal subjects. 

Together with Mike Young and Ed Wasserman, I have recently developed a new mathematical model of relational learning in transposition task (Lazareva, Young, & Wasserman, 2014). Drake students Emily Leiker and Kaitlyn Kandray have done a lot of work replicating some of our earlier findings that was crucial for model evaluation.

This model uses three predictors that jointly determine a likelihood of relational responding:

1) Generalized excitation/inhibition: This predictor assumes that a stimulus will be chosen merely based on its physical similarity to previously reinforced and nonreinforced stimuli.

2) Relational disparity: This predictor assumes that a stimulus will be chosen to the degree that it is larger than its comparison stimulus if the larger stimulus was reinforced during training (or smaller, if the smaller stimulus was reinforced).

3) Familiarity: This predictor assumes that a stimulus will more be reliably chosen if a testing pair is similar to the training pair(s); if the testing pair is very different from any of the training pairs, then its novelty is likely to lead to chance levels of responding. 

​We have also shown that some procedural manipulations, such as the increase in the number of training pairs, lead to an increased control by relational disparity (i.e., increased sensitivity to the relationship between the stimuli). In future, we will be using this model to assess the relative contributions of different factors on an organism’s behavior, to compare the tendency to respond relationally across sensory domains or species, and to create experimental designs and to select testing pairs that should maximize (or minimize) the influence of the relevant factors.

In collaboration with Onur Güntürkün (Ruhr-Universität Bochum), we have recently found that birds learning relational transposition task have higher levels of expression of immediate early genes in certain subareas of hippocampus than birds learning associative transposition task (Panfil et al., in preparation). This interesting result implies that even simple relational tasks that do not require manipulation of relational information can still involve hippocampus. We are currently planning to conduct a lesion study to verify these findings.

Published papers:

Lazareva, O. F., Young, M. E., & Wasserman, E. A. (2014). A three-component model of relational learning in a transposition paradigm. Journal of Experimental Psychology: Animal Learning and Cognition, 40, 63-80. doi:10.1037/xan0000004

​Lazareva, O. F. (2012). Relational learning in a context of transposition: A review. Journal of Experimental Analysis of Behavior, 97, 231-248. doi: 10/1901.jeab.2012.97-231

Lazareva, O. F., Miner, M., Young, M. E., & Wasserman, E. A. (2008).  Multiple-pair training increases transposition in pigeons. Learning & Behavior, 36, 174-187. doi: 10.3758/LB.36.3.174

Lazareva, O. F., Wasserman, E. A., Young, M. E. (2005). Transposition in pigeons:  Reassessing Spence (1937) with multiple discrimination training. Learning and Behavior, 33, 22-46. doi: 10.3758/BF03196048

Relational learning in transposition task