Visual search by chimpanzees (Pan): assessment of controlling relations.
Different chimps used different rules in the same visual search task, so always test what controls your client's choices.
01Research in Context
What this study did
Researchers watched three chimps hunt for matching pictures on a screen. The task looked like a simple matching game, but the team tweaked the pictures to see what the chimps really used to pick the answer.
Some trials hid identity cues. Other trials pushed odd-one-out rules. By changing the set size, they could tell which cues each chimp followed.
What they found
One chimp followed an oddity rule: pick the picture that does not match the others. Two chimps followed identity rules: pick the picture that matches the sample. Accuracy dropped when the display held more pictures, but the style of control stayed the same for each animal.
How this fits with other research
Beurms et al. (2017) later showed that humans quickly form symmetry relations even when timing shifts. People seem flexible, while the chimps locked into one rule.
Belisle et al. (2022) extended the idea to kids with autism. After equivalence training, the children found trained words faster in a word search. Both studies show that prior relational learning guides later visual search, but the 1995 paper warns that the guiding rule may differ across subjects.
Duker et al. (1991) and Taylor et al. (1993) already proved that context can switch which stimulus controls a choice. The chimp data add a cross-species note: even without added context, individual animals can settle on different controlling relations in the same task.
Why it matters
Before you run matching-to-sample with any learner, probe what actually controls their choices. Mix in oddity trials, vary display size, or test without the usual sample. If the child shifts rules, you will spot it early and avoid false negatives in equivalence tests. One quick probe can save weeks of puzzling data.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add two oddity trials to your next matching-to-sample set and see if the learner still picks by identity.
02At a glance
03Original abstract
Three experimentally sophisticated chimpanzees (Pan), Akira, Chloe, and Ai, were trained on visual search performance using a modified multiple-alternative matching-to-sample task in which a sample stimulus was followed by the search display containing one target identical to the sample and several uniform distractors (i.e., negative comparison stimuli were identical to each other). After they acquired this task, they were tested for transfer of visual search performance to trials in which the sample was not followed by the uniform search display (odd-item search). Akira showed positive transfer of visual search performance to odd-item search even when the display size (the number of stimulus items in the search display) was small, whereas Chloe and Ai showed a transfer only when the display size was large. Chloe and Ai used some nonrelational cues such as perceptual isolation of the target among uniform distractors (so-called pop-out). In addition to the odd-item search test, various types of probe trials were presented to clarify the controlling relations in multiple-alternative matching to sample. Akira showed a decrement of accuracy as a function of the display size when the search display was nonuniform (i.e., each "distractor" stimulus was not the same), whereas Chloe and Ai showed perfect performance. Furthermore, when the sample was identical to the uniform distractors in the search display, Chloe and Ai never selected an odd-item target, but Akira selected it when the display size was large. These results indicated that Akira's behavior was controlled mainly by relational cues of target-distractor oddity, whereas an identity relation between the sample and the target strongly controlled the performance of Chloe and Ai.
Journal of the experimental analysis of behavior, 1995 · doi:10.1901/jeab.1995.63-175