Do chimpanzees (Pan troglodytes) mentally represent collaboration?: Action-learning and communication in a partnered task

Abstract

Non-human primates engage in complex collective behaviours, but existing research does not paint a clear picture of what individuals cognitively represent when they act together. This study investigates chimpanzees' capacity for co-representation. If individuals represent others' actions as they relate to their own during a collaborative task, they should more easily learn to reproduce that action when their roles are switched. In a between-subjects design, we trained ten chimpanzees (Pan troglodytes) on a sequential task, in which the first action is performed by either a human partner or a non-social object, and the second action is performed by the subject. We then imposed a breakdown in the action sequence, in which subjects could perform both actions themselves, but received no help from the experimenter or object. We measured subjects' success in reproducing the first action in the sequence, as well as their attempts to recruit the experimenter's help using requesting gestures. We found no overall difference in subjects' ability to perform the first action in the sequence, but we observed significant qualitative differences in their solutions: individuals in the partnered condition replicated the experimenter's action, while those in the non-social condition achieved the same end using alternative methods. This difference in solution style could indicate that only those chimpanzees in the partnered condition mentally represented the experimenter's action during the collaborative task. We caution, however, that given the small number of subjects who solved the task, this result could also be driven by individual differences. We also found that subjects consistently produced communicative gestures toward the experimenter, but were more likely to do so after exhausting all actions they could take alone. We suggest that these patterns of behaviour highlight a number of key empirical considerations for the study of coordination in non-human primates.

Fig 1. Depiction of the see-saw apparatus flipped toward the grooved edge.

Dashed lines indicate the path of the grape as it rolls from the ridged edge to the grooved edge and is pulled toward the subject using the paper.

Fig 2. Top left (A): See-saw apparatus in the starting position, about to be flipped by the experimenter.

Next to the experimenter’s hand is the lever used in the object condition. Top right (B): graphic of see-saw apparatus at the start of a trial, before it has been flipped. Bottom left (C): After the see-saw has been flipped, the subject (Velu) is pulling the paper to retrieve the grape. Bottom right (D): overhead view of the see-saw apparatus at the start of a trial, before it has been flipped. X indicates the approximate location of the experimenter’s hand or object (lever) when flipping the apparatus.

Fig 3. Total number of flips (as opposed to alternative methods) by experimental condition (partnered, object) including only unblocked phase trials in which a success occurred (out of a possible 12). Points depict individual subjects’ data (n = 3 from each learning condition). The central line of the boxplot shows the median, the coloured boxes show the IQR, and the whiskers indicate the minimum and maximum values of the data within 1.5 x IQR. ***p < .001.

Fig 4. Median number of communication behaviours per trial in the unblocked phase, separated by the state of the apparatus (pre- or post-pull).

The central line of the boxplot shows the median, the coloured boxes show the IQR, and the whiskers indicate the minimum and maximum values of the data within 1.5 x IQR. Data outside of this range are plotted as separate outlying points. **p < .05.