There is some rich history and sociology in the study of the evolution of cooperation. Some of the biggest names in biology have debated this issue. In academia, using or ignoring the terms as defined by a previous scholar is akin to confirming or supporting their view. So scientific controversies can be influenced by decisions of semantics more than one might expect or want to admit.
The biological study of cooperation is plagued by much confusion and debate regarding the definition of terms like “altruism”, “cooperation”, and “reciprocity” (reviewed here), even ignoring that these terms also have completely different meanings in the social sciences and in everyday speech.
The term reciprocity (or reciprocal altruism) was at first a very broad idea describing cooperation in which partners exchange and enforce mutual direct fitness benefits through contingent acts of helping. During much of the 70s, 80s, and 90s, it was the default alternative to kin selection for explaining the evolution of cooperation. Reciprocal “altruism” was actually a way of enforcing mutualism, whether within a species or between species. In the original paper, Trivers used the example of the cleaner fish and client fish mutualism as an example of reciprocity, and further models were even broader in scope, hoping to provide insight to microbial cooperation, cancer, and intracellular cooperation.
However, over time, the term somehow gained a more specific and more controversial connotation. Reciprocity apparently required advanced cognition, intention, calculation of cooperative services, planned social investments, and even numerical discrimination. And reciprocity had to fit, in a literal way, all the ideal unrealistic assumptions of the iterated prisoner’s dilemma game. One evolutionary psychologist even told me that reciprocity requires that an organism “expect” a return on it’s cooperative investment. For these reasons, recent reviews suggest that reciprocity is likely only important in humans.
As a consequence, many behavioral ecologists now avoid the term “reciprocity” altogether. For example, in some excellent studies showing that mutualists preferentially invest in (reward) cooperators and do not invest in (punish) freeloading cheats, the mechanisms are typically called “sanctions” or “reciprocal rewards” rather than reciprocity. To further deepen the semantic rift, the term “reciprocity” is favored primarily by game-theoreticians who don’t always play nice with biologists. For example, the controversial Martin Nowak, ever seeking to fundamentally revolutionize the study of social evolution, has stated that pretty much all the previous work of biologists studying social evolution over more than 45 years “must be considered meager”. He has also annoyed many in the field by reinventing, rediscovering, and re-naming many previously described forms of cooperation as: “direct reciprocity”, “indirect reciprocity”, “spatial reciprocity”, and “network reciprocity”.
As a consequence of these “terminology wars”, there have been several papers (including this one, this one, this one and this one) about how terms should be defined and classified. I have spoken with a few biologists who felt they received negative reviews on studies for using the “wrong” terminology (i.e. reciprocity or reciprocal altruism). This is a shame because such semantic arguments can overshadow some fascinating science. For example, there is an interesting study on cooperation in mobbing birds that I feel was overlooked due to its use of the outdated term “reciprocal altruism”.
Birds join together in loud aggressive “mobs” to intimidate and chase away predators. Mobbing birds were originally considered anonymous mobs rather than more individualized groups of familiar birds. However, Krams and co-authors (2008) showed that pairs of pied flycatchers joined those mobbing neighbors that previously helped them mob, and withheld support from neighbors who did not. To my knowledge, this is the only study that simulated cheating in order to test the enforcement of cooperation using natural avian helping behavior. Normally, pied flycatcher pairs will join together to mob a perceived predator such as an owl (or in this case, a fake owl). Krams and others tested 44 trios of mated pairs at equidistant nestboxes in the following experiment: Pair A was exposed to a fake owl near their nestbox. Pair B was captured and hence could not participate in the cooperative mobbing. Pair C was left alone. Consequently, pair C then helped A, but pair B did not and could not (they were secretly held captive). Next, the test: the authors presented both pairs B and C with owls. Who did A help? In 30 of 32 trials, pair A helped C (the past helpers).
In a followup experiment, pair B was presented with an owl. Amazingly, in 8 out of the 9 cases, pair A (the birds that were not helped by pair B) did not join to help B in mobbing. But pair C (the control pair) did. This study demonstrates that pied flycatchers reward helpers by joining them and punish non-helpers by not joining them. Rather then being upheld as strong evidence of reciprocity in a natural system. This study has attracted controversy and criticism claiming that this form of behavior cannot be called “reciprocity”. Why? Because the reviewers thought: how could these birds be smart enough for reciprocity?
Russell and Wright (2008) implied that reciprocity is too cognitively difficult for this species. They then argued that helping only occurs because pair A likely thinks that C is more willing to help them mob at that moment than pair B. So the birds will think it’s more efficient to mob with C. Yet this hypothesis make little sense because pair A could easily see that pair B is mobbing in both test phases of the experiment, so decisions to not help pair B must have been based on past, not present, experience. Connor (2010) offered a second alternative explanation: that A did not help B to avoid potential parasite infestation. Both these hypotheses appear to be based on the logic that reciprocity is very unlikely a priori and another “simpler” explanation is therefore better. Discounting reciprocity in this way often stems from thinking of reciprocity as planned and calculated, rather than being an evolved social response triggered by the appropriate cues. Frankly, the notion of calculated reciprocity as a general mechanism is ridiculous– akin to imagining that autumn caching of seeds for the coming winter requires that squirrels be able to value future rewards using appropriate “temporal discounting”– putting off their desire to eat the seed now so that they can enjoy it later when they need it more.
Thinking of reciprocity as “calculated” leads to the view of reciprocity as something rare and spectacular, which is not what Trivers was imagining in his original paper. The resulting debate about whether or not a situation semantically “counts” as reciprocity is unhelpful, because it moves attention away from actual research questions such as: What behaviors prevent cheating in a mutualism? And what cues do animals use to make helping decisions? To what extent to cooperators prevent cheating actively through reward and punishment versus passively through partner choice?
There is excellent evidence that animals enforce cooperation by basing their own cognitive decisions about helping on assessments of their partner’s ability to provide a cooperative return. This is an important phenomenon that should not be ignored whether or not people choose to call it “reciprocity”.
Further reading: Influential uses of very restrictive definitions of reciprocity are provided by Clutton-Brock’s review in Nature, Stevens & Hauser’s review. And a response is given here.