Canny corvids curb cooperation after they’ve been cheated

Excuse the alliteration, as I’m tired this morning. But not too tired to report on a new paper in Nature Scientific Reports by Jorg Massen, Caroline Ritter, and Thomas Bugnyar (free access and pdf, reference below, popular summary at IFL Science). It shows something heretofore unknown in birds: a recognition that they’ve been cheated and a resulting reluctance to cooperate with anyone after that.

The paper begins with a useful literature survey of animals that cooperate with others (killer whales, Harris’s hawks, chimpanzees), animals that wait for another individual to arrive to help them perform a cooperative task (capuchins, bonobos, Asian elephants), and animals known to discriminate with respect to partners, choosing to work with those that have been more helpful (chimpanzees and coral trout [!]).

But that kind of partner discrimination hadn’t been described in birds. Now it is, thanks to a clever experiment in which ravens (Corvus corax), already known to be wicked smart, were forced into a situation where they had to cooperate to get food.

The experimental setup is shown below: ravens are put in a cage, and on the other side of the wire is a wooden platform containing two pieces of cheese. There are two strings threaded through grommets on the platform, and to pull the cheese toward the cage and get noms, two ravens have to pull at once; if only one pulls, the string simply slips out of the grommets and birds no can haz noms.

(From paper): Figure 1. Experimental set-up. Two birds have to pull the two ends of the string simultaneously to move the feeding platform in reach. If only one bird pulls, the string will just go through the two metal loops anchored to the feeding platform and become unthreaded, while the platform remains stationary. Picture drawn by Nadja Kavcik-Graumann.

There were two sets of experiments, involving in total 9 juvenile hand-reared ravens. The first conclusion is that ravens can’t figure out the task on their own; that is, they can’t spontaneously realize that there have to be two birds pulling together to get the noms. This was discerned from the failure of ravens to wait for the arrival of a partner before pulling the strings, and from their tendency to pull the string when there were no other birds present. As the authors note:

With regard to the control trials, in 243 out of 288 solitary trials (84.38%) individuals pulled the string even though this had no effect. Similarly, only 2 birds waited in 5 (out of 288: 1.73%) waiting trials for their partner to arrive at the apparatus and subsequently cooperated successfully. These data suggest that the ravens did not understand the need for a partner to solve the task in this experimental set-up.

So they’re not THAT smart!

However, the two experiments did give some intriguing results:

• The ravens eventually did learn that joint pulling did get them noms. In the first study, using groups of ravens in the cage, there was spontaneous cooperation in 397 out of 600 trials, and every bird was successful at least 32 times.

• Using knowledge about the age of the birds, their rank in the raven dominance hierarchy, sex, and kinship (relatives or not?), as well as “inter-individual tolerance” (the distance at which birds could tolerate each other’s presence when both were pulling on individual strings connected to cheese), the only predictor of cooperation was inter-individual tolerance (IIT). Here are the data, with the left graph showing the correlation between IIT and successful joint pulling when groups of birds were tested, and the right the same graph, but when only pairs of birds were tested. The correlation between IIT and joint success is significant, and shows that birds recognize each other, preferring to cooperate with those they can tolerate better:

• The experiment was repeated using pairs of birds in all possible combinations, except for those ravens at the top of the dominance hierarchy, who simply didn’t let any other ravens get the cheese. This experiment showed two things. First, some ravens cheated in the dyadic cooperation: after two of them pulled in the cheese bar, one would shove the other out of the way and eat both pieces of cheese. These were the CHEATERS. When the authors analyzed the results, they found that a raven was less likely to cooperate in subsequent trials if it had been victimized by a cheater. Here are those results with the caption:

(a) Proportion of trials in which two birds cooperated successfully subsequent to an equal or unequal reward devision after the previous successful cooperation trial

As you see, the proportion of trials in which two birds worked together successfully was lower when the previous trial gave an unequal division of rewards compared to trials where both birds got the cheese (left bar compared to right). There was also an effect of kinship: kin cooperated significantly more than non-kin, but the p value wasn’t impressive (0.02), and the kinship effect wasn’t seen in the group experiment.

• A final analysis tested the notion that, given the above results, cheaters who got two pieces of cheese would be even more likely to cooperate the next time. But that didn’t pan out: ravens who were cheaters, and got two rewards in a trial, were just as likely to successfully cooperate as those who got only one reward, as shown in the following plot. But again we find that if a raven gets no cheese, it’ll be less likely to cooperate in future trials:

Proportion of trials in which a bird pulled the string after it had received zero, one or two rewards in the previous successful cooperation trial.

• Finally, the authors tested whether cheaters became more fair—that is, allowed their pulling partner to get a piece of cheese—in subsequent trials. One might expect this if, because ravens cotton on to cheaters and don’t cooperate thereafter, the cheaters have to reinstate themselves in the esteem of their comrades by being fair. For if you’re marked as a cheater, nobody will cooperate with you and you might get less cheese in the future. Sadly, cheaters kept on cheating. I’m not sure if this is disadvantageous to them: after all, some birds get duped into cooperating with cheaters, even after they’ve been duped, so it may be generally advantageous to cheat.

Overall, then, this experiment shows that ravens can recognize individuals, have a tendency to cooperate with those whose presence they tolerate more easily, and can recognize cheaters and avoid cooperating with them in the future.

This is intriguing, but of course these are hand-reared ravens tested in a highly rarified experimental situation, so the relevance of the study to the behavior of these birds in nature is unclear. How often do ravens cooperate in the wild? Are flocks semi-permanent, so birds have a chance to recognize and avoid cheaters? Do flocks include relatives? These are questions that may already have an answer, but you’ll have to ask Bernd Heinrich.

Oh, I forgot: the experiment also shows that ravens love cheese. In a serious scientific omission, the authors don’t identify the type of cheese they used.

h/t: Scott G.

________

Massen, J. G. M., C. Ritter, and T. Bugnyar. 2015. Tolerance and reward equity predict cooperation in ravens (Corvus corax). Nature Scientific Reports, 15021, doi:10.1038/srep15021