Humans are outstanding in their ability to achieve mutual cooperation for direct benefits. One oftenquoted factor driving our unique cooperation skills is our ability to delay gratification for future benefits, which is supposedly linked to our advanced cognitive abilities relative to other species. The established experimental paradigm used to test delayed gratification in primates and birds is to give subjects access to a small food reward, which can be exchanged for a larger (quantitative

more »

... k) or better quality (qualitative task) food reward after a time delay. Here we used this paradigm to test primarily the cleaner fish Labroides dimidiatus for its ability to delay gratification. The specific ecology of cleaners makes them ideal candidates for this task: cleaners remove ectoparasites from visiting 'client' reef fish although they prefer to eat client mucus. This eating against preference is equivalent to delayed gratification as clients that have their ectoparasites removed (as opposed to their mucus) will both stay longer with a specific cleaner and return to it for later inspection. Cleaners performed very well in the quantitative task, i.e. similarly to monkeys. In contrast, they performed poorly if the delayed reward was of higher quality. Using the quantitative task to test other wrasse species, we found that another cleaner species and noncleaner wrasse also performed well, the latter despite no evidence that they need such ability to cooperate in nature. However, only the two cleaner species showed evidence for making decisions early on about whether to wait or to directly eat the immediate reward. We conclude that the ability to delay gratification does not warrant advanced cognitive abilities that require endotherm-specific brain morphology, although the underlying mechanisms vary between species.