Anything for a cheerio: Brown capuchins (Sapajus [Cebus] apella) consistently coordinate in an Assurance Game for unequal payoffs

Abstract

Unequal outcomes disrupt cooperation in some situations, but this has not been tested in the context of coordination in economic games. To explore this, we tested brown capuchins (Sapajus [Cebus] apella) on a manual version of the Stag Hunt (or Assurance) Game, in which individuals sequentially chose between two options, Stag or Hare, and were rewarded according to their choices and that of their partner. Typically, coordination on Stag results in an equal highest payout, whereas coordinating on Hare results in a guaranteed equal but lower payoff and uncoordinated play results in the lowest payoff when playing Stag. We varied this structure such that one capuchin received double the rewards for the coordinated Stag outcome; thus, it was still both animals' best option, but no longer equally rewarding. Despite the inequality, capuchins coordinated on Stag in 78% of trials, and neither payoff structure nor their partner's choice impacted their decision. Additionally, there was no relationship between self‐scratching, a measure of stress in capuchins, and choices. After completing the study, we discovered our reward, cheerios, was sufficiently valuable that in another study, capuchins never refused it, so post hoc we repeated the study using a lower value reward, banana flavored pellets. Capuchins completed only 26% of the pellet trials (compared to 98% with cheerios), constraining our ability to interpret the results, but nonetheless the monkeys showed a decrease in preference for Stag, particularly when they received fewer rewards for the coordinated Stag outcome. These results reinforce capuchins' ability to find coordinated outcomes in the Stag Hunt game, but more work is needed to determine whether the monkeys did not mind the inequality or were unwilling to sacrifice a highly preferred food to rectify it. In either case, researchers should carefully consider the impact of their chosen rewards on subjects' choices.

Some dogs can find the payoff-dominant outcome in the Assurance game

“…With more exposure to the game, the same capuchin monkeys coordinated in the payoff-dominant NE. 21 , 31 , 67 Thus, even when dogs in our study were tested with a considerably lower number of sessions in comparison with other species, four dyads (36% of our sample) were able to find the payoff-dominant NE.…”

Comparative economics: how studying other primates helps us better understand the evolution of our own economic decision making

“…Of course there are exceptions, for instance if the modality of the task changed owing to differences in sensory systems across species, but even small procedural changes, such as in reward value or how a choice is made, may influence responses (i.e. [55,56]). Finally, since the goal is not to understand humans, broadly, but to compare the results to previous ones, which typically used college students as the sample, it is usually reasonable to use college students.…”

“…As mentioned earlier, the computerized version differs by having more trials per session, less time between trials, and a shorter delay between choice and reward, all of which promote learning, which may account for the difference (anecdotally, the capuchins’ performance improved when we increased the trial count of sessions, although this was confounded with experience; [58,62]). However, in later versions of the task, capuchins did equally well on the manual version [55], suggesting that with experience there is little difference between the two modalities and, even initially, there is no difference in choices among those who did learn the contingencies.…”

“…We became curious how inequity would impact coordination in capuchins, who are known to respond negatively to receiving a less valuable reward than a partner [79], particularly in the context of cooperation [46,47]. To do so, we used the manual token trading version of the Assurance game, with the exception that we altered it to be an explicitly sequential task (one player was always given their choice first), and changed the pay-offs slightly so that one player received a lower pay-off (two rewards) for coordinating on stag than the other (who still received four; [55]). When the rewards were a highly preferred food, the level of coordination stayed the same, but when we switched to a less preferred reward, the rate of stag choices dropped off, with both the first and, in particular, the second mover more often choosing hare .…”

“…It is important to recognize that for many of the reported studies (including all of the studies on which we are authors), the subjects are never deprived of food, water, treats or anything else to compel motivation to participate in testing, which means that they may simply not be interested in playing if the rewards are not good enough! Given that reward value [55,56] and even feeding regimen [80] affects choices, this becomes a significant influence on our results.…”