Individual differences in sociocognitive traits in semi-free-ranging rhesus monkeys (Macaca mulatta)

Abstract

Characterizing individual differences in cognition is crucial for understanding the evolution of cognition as well as to test the biological consequences of different cognitive traits. Here, we harnessed the strengths of a uniquely large, naturally-living primate population at the Cayo Santiago Biological Field Station to characterized individual differences in rhesus monkey performance across two social cognitive tasks. A total of n = 204 semi-free-ranging adult rhesus monkeys participated in a data collection procedure, where we aimed to test individuals on both tasks at two time-points that were one year apart. In the socioemotional responses task, we assessed monkeys' attention to conspecific photographs with neutral versus negative emotional expressions. We found that monkeys showed overall declines in interest in conspecific photographs with age, but relative increases in attention to threat stimuli specifically, and further that these responses exhibited long-term stability across repeated testing. In the gaze following task we assessed monkeys' propensity to co-orient with an experimenter. Here, we found no evidence for age-related change in responses, and responses showed only limited repeatability over time. Finally, we found some evidence for common individual variation for performance across the tasks: monkeys that showed greater interest in conspecific photographs were more likely to follow a human's gaze. These results show how studies of comparative cognitive development and aging can provide insights into the evolution of cognition, and identify core primate social cognitive traits that may be related across and within individuals.

Keywords: comparative development; emotional processing; gaze following; primate cognition.

Figure 1

Experimental setup for the sociocognitive tasks. (a) The socioemotional responses task measured preferences for viewing photographs of conspecifics producing emotional versus neutral expressions. Monkeys were presented with up to four photos, each for 10 s: a neutral female face (trial 1), followed by a threat face of the same female (trial 2), and a neutral male face (trial 3), followed by a threat face of the same male (trial 4). Before each trial the apparatus had a purple cover over the window, and the experimenter would remove the cover so the monkey could see the photo to start each trial. (b) The gaze following task measured co‐orienting responses with a demonstrator. Monkeys completed up to four trials where the experimenter captured the monkey's attention and then looked directly upward for 10 s; we measured whether the monkey also looked up during this period.

Figure 2

Individual variation in the socioemotional responses task. (a) Mean looking time to conspecific neutral and threat photographs. Graph shows mean duration of looking time towards neutral versus threat stimuli for male and female photographs, broken down by subjects' age cohort and sex. Error bars indicate standard error. (b) Predicted changes in difference scores (negativity bias) with age. Each subject was assigned a difference score indexing their negativity bias (threat looking time–neutral looking time) for female and male photographs, which increased with age. Estimated values are derived from mixed models also accounting for subjects' identity, sex, age, and experimental variables. Ribbons indicate 95% confidence intervals. (c) Predicted performance over repeated testing. We estimated how performance in year two of testing was related to year one performance, for individuals who completed both years of testing. Estimated values are derived from mixed models also accounting for subjects' identity, sex, age, and experimental variables. Ribbons indicate 95% confidence intervals. (d) Longitudinal changes in looking responses. We used repeated measures correlations to isolate within‐individual associations in looking time. The panel depicts responses to the first trial (female neutral photos). The solid gray line indicates the overall linear relationship between age and looking time in the data set incorporating both within‐ and between‐individual age effects (estimated from a mixed model), whereas the dashed lines indicate the common within‐individual age‐related change estimated from repeated measures correlation.

Figure 3

Individual variation in the gaze following task. (a) Propensity to follow gaze. Graph shows mean proportion of individuals looking up across the four trials, broken down by subjects' age and sex. Error bars indicate standard error. (b) Predicted propensity to follow gaze did not change with age. Estimated values are derived from generalized linear mixed‐effects models (GLMMs) of propensity to look up also accounting for subjects' identity, sex, age, and trial number. Ribbons indicate 95% confidence intervals. (c) Predicted performance over repeated testing. We tested if performance in year two was related to year one performance. There was no relationship between performance in year one and likelihood of looking up in year two for individuals who completed both years of testing. Estimated values are derived from GLMMS of propensity to look up accounting for subjects' identity, sex, age, and trial number. Ribbons indicate 95% confidence intervals. (d) Longitudinal changes in duration of looking responses. We used repeated measures correlations to isolate within‐individual associations in time looking up. The panel depicts responses to the first trial. The solid gray line indicates the (lack of) overall linear relationship between age and time spent looking up, whereas the dashed lines indicate the (lack of) within‐individual changes with age.

Figure 4

Inter‐task relationships. (a) Predicted performance in the socioemotional responses task from the gaze following task. Responses to conspecific photos for individuals who completed both tasks in a given year are depicted, broken down by their responses in the first two trials of the gaze following task (which all individuals completed; they therefore could have gaze followed on up to two trials). Gaze following did not significantly predict interest in the conspecific photographs. Estimated values are derived from linear mixed models also accounting for subjects' identity, sex, and age, and experimental variables. Error bars indicate 95% confidence intervals. (b) Predicted performance in gaze following task from the socioemotional responses task. Responses to gaze following demonstrations for individuals who completed both tasks in a given year are depicted, according to their average duration of looking at conspecific photos in the two trials of the socioemotional responses task (which all individuals completed). Individuals who looked longer at conspecific photos were more likely to look up in the gaze following task. Estimated values are derived from generalized linear mixed‐effects models of propensity to look up also accounting for subjects' identity, sex, age, and trial number. Ribbons indicate 95% confidence intervals.