ABS 2023
Search
Experimental evidence for compositional processing of visual stimuli in rhesus macaques
Angelle Antoun1,2, Rohini Murugan1,2, Kathleen J. Bostick2, Benjamin Wilson1,2. 1Emory University, Atlanta, GA, United States; 2Emory National Primate Research Center, Atlanta, GA, United States

Compositionality, the ability to combine meaningful words into single representations, is a critical feature of language. While there is evidence for call combinations in non-human primates, these combinations are minimal, with limited expressivity compared to language, suggesting that compositionality may be uniquely human. In language, negative clauses (e.g., ‘not red’) are necessarily compositional; the meaning of the phrase cannot be derived from either word alone. To assess compositionality in non-human primates, we developed a series of non-linguistic negation experiments. We presented six rhesus macaques with cue stimuli paired with either a ‘positive’ or ‘negation’ label, and two choice stimuli. On positive trials, monkeys selected the stimulus that matched the cue, and on negation trials selected the non-matching stimulus. Critically, the cue and choice stimuli changed each trial to avoid learning about individual stimuli. Macaques solved this task, combining the cue and label on a trial-by-trial basis to select the correct stimulus. Our results provide strong evidence for compositionality in non-human primates, suggesting the core of this ability may not be human unique.