Bshary & Grutter (2006) Image scoring and cooperation in cleaner fish mutualism

Ref: Bshary, R., & Grutter, A. S. (2006). Image scoring an cooperation in a cleaner fish mutualism. Nature, 441, 975-978.

Evidence that client fish use observations (image scoring) to decide whether to invite cleaner fish into their mouths. The client fish are not piscivorous. The cleaner fish can eat client ectoparasites, but also prefer client mucus - thus, there is a temptation to cheat.

Client fish almost always invite a cleaner's inspection if they witness that the cleaner's last interaction ended without conflict, invite less if they do not have such knowledge, and invite the least if the last interaction ended with conflict. Furthermore, cleaner fish are more cooperative in the presence of bystanders than when they are alone with one client. These results are consistent with the hypothesis that image scoring by client fish is a selfish act used to find a cooperative cleaner or to avoid a cheating cleaner, resulting in (more) cooperative behaviour in cleaners in the presence of bystanders.


Experiment 1: They allowed a client fish access to two cleaner fish, one it had observed cooperating, one with an "unknown" cooperative level. Both times this was tested, the client fish spent more time with the cooperative cleaner.

Experiment 2:  They showed that the cleaner fish could basically learn to "cooperate" more (by choosing a less preferred food) if it meant more access to food in general and occasional preferred food. They presented two dishes, each of which had preferred prawns and less-preferred fish flakes on them. In the control non-image scoring condition, a plate was removed as soon as a prawn was eaten of it. In the image scoring condition, both plates were removed as soon as a prawn was eaten off of either plate. The results show that in the image scoring condition, cleaner fish ate a much high proportion of fish flakes off of the first plate their touched - in this condition, all cleaners sometimes are only flakes from the first plate (median 7/20 trials), which allowed them access to the second plate. This happened way less in the no-image-scoring trials (median 1.5/20 trials). This was also compared to a case in which only a single plate was presented.