Contrasting two versions of the 4-cup 2-item disjunctive syllogism task in great apes.

Chimpanzees excel at inference tasks which require that they search for a single food item from partial information. Yet, when presented with 2-item tasks which test the same inference operation, chimpanzees show a consistent breakdown in performance. Here we test a diverse zoo-housed cohort (n = 24) comprising all 4 great ape species under the classic 4-cup 2-item task, previously administered to children and chimpanzees, and a modified task administered to baboons. The aim of this study is to delineate whether the divergent results reported from the literature are taxonomic differences or artefacts of their methodologies, while extending the literature to cover the remaining great ape species. We find that apes adaptively adjust their choice behaviour in both variants of the task, but that they perform better in trials where the information provided rules out a location rather than removes one of the food items. In a second experiment involving those subjects who passed the first, along with a group of naïve subjects, we test whether subjects were able to apply the logical operation selectively by including control trials where the correct response is reversed. Performance in standard trials breaks down with the addition of control trials, meaning that if apes did solve the first experiment logically, they are not capable of applying that logic flexibly. Considering this finding, we conclude that a 4-cup 2-item task may not be a suitable test of logical reasoning in great apes.