Proximate versus ultimate causation
Social behaviour is best understood by differentiating its proximate cause (that is, how the behaviour arises in animals) from its ultimate cause (that is, the evolutionary history and functional utility of the behaviour). Proximate causes include hereditary, developmental, structural, cognitive, psychological, and physiological aspects of behaviour. In other words, proximate causes are the mechanisms directly underlying the behaviour. For example, an animal separated from the herd may exhibit behaviours associated with fear reactions (such as elevated heart rate, shaking, and hypersensitivity to sounds), which cause it to behave in ways that increase its chances of reuniting with the group. The underlying hormonal response, which is triggered by separation from the herd, is a proximate cause of these fear-based behaviours. In contrast, the ultimate causes of social behaviours include their evolutionary or historical origins and the selective processes that have shaped their past and current functions. In the case of the isolated herd animal, the development of a better defense against predators that results in increased survival of individuals remaining in groups would be an ultimate cause for the tendency to reunite with the herd.
Dutch-born British zoologist and ethologist Nikolaas Tinbergen was first to clarify these levels of explanation, naming four which he referred to as “survival value,” “causation,” “development,” and “evolutionary history.” Tinbergen also emphasized the importance of addressing questions at the appropriate level of explanation. For example, determining the underlying mechanism (causation) of a behaviour does not address the hypotheses regarding its historical origin (evolutionary history) or current survival value. This still causes confusion among evolutionary biologists interested in adaptation, and many examples of unproductive arguments across levels of explanation can be found in the scientific literature.
Strong inference and the scientific study of social behaviour
Use of the scientific method to study social behaviour permits biologists to deduce the proximate and ultimate functions by using strong inference based on a set of critical predictions. If experiments to test these predictions indicate that the predictions are not met, then the hypothesis is falsified and discarded. If the predictions are met, the hypothesis is supported, but that does not prove it is true.
This is illustrated by examining a question: Why do male birds sometimes adopt and feed offspring of widowed females? One possible explanation is that they have mated with the female and have genetic offspring in the female’s nest (current benefits hypothesis). An alternative hypothesis is that the adoptive male gains future benefits because his foster-parenting increases the likelihood that the female will mate with him during her next breeding attempt (future benefits hypothesis). The current benefits hypothesis predicts that some of the female’s nestlings were sired by the adoptive father, whereas the future benefits hypothesis predicts that the adoptive male will mate sooner, usually with the widowed female, and produce more offspring in the future than an unpaired male that fails to adopt. While mutually exclusive hypotheses are ideal, in many cases behaviours have more than one current function and, as in the example of adoption, one or both hypotheses may be true.
Strong inference relies on critical predictions that are capable of distinguishing between alternative hypotheses, whether proximate or ultimate. It also relies on devising clear tests in which each alternative can be falsified by using one or more predictions. In general, predictions can be tested either with data collected from field observations or with experiments. Experiments are considered preferable to field observations because confounding factors are more easily controlled. Unfortunately, manipulations involved in experiments may alter other factors beyond those which the scientist intended, especially where social behaviour is concerned. In order to minimize such problems, researchers take great pains to avoid biases in their experimental procedures and to test their hypotheses by using multiple lines of evidence.
For example, consider the question of why offspring of some species of birds and mammals delay dispersal and remain on their natal territory where they may help raise younger siblings. One of the many basic questions raised by such “helpers-at-the-nest” is the importance of genetic relatedness and kinship to the evolution of the behaviour. Experimentally, cross-fostering young so as to eliminate any genetic relatedness between nestlings and helpers does not typically alter or reduce helping behaviour, but does this demonstrate that kinship is not important? The current thinking on this matter is that cross-fostering leads to a situation where totally unrelated young occur in the nest, a situation that has never been found in the wild. Other studies, meanwhile, have shown that the vast majority of helpers normally feed closely related young. When given the choice, helpers whose own nests have failed preferentially choose to aid closely related young over more distantly related or unrelated young. This behaviour was demonstrated even when the latter were closer to the helper’s own failed nesting site. Such results indicate that kin selection plays a key role in the evolution of helping behaviour, despite the experiments suggesting otherwise.