In behavioural ecology, we face a number of limitations in trying to ferret out the relationship between behaviour and evolutionary forces. These range from the philosophical and theoretical (e.g. what makes a behaviour adaptive or an adaptation?) to the mundane and methodological (is that experimental set up really measuring aggressive behaviour?), and solving these problems is one of the most pressing tasks facing a behavioural ecologist attempting to make useful statements about a behaviour’s evolution. However, while some of these issues are recurrent and obvious, others are more subtle and can sometimes slip under the radar. One such problem is the topic of a recent paper by Véronique Martel and Guy Bovin, published recently in the Journal of Insect Behaviour and entitled “Do choice tests really test choice?” (DOI: 10.1007/s10905-011-9257-9).
The thrust of their argument is that there is a difference between “apparent choice”, and “true choice”, which is driven largely by the fact that we can’t ask animals what they would have done under different circumstances. As Martel and Bovin point out, animals may make one choice when presented with a particular set of stimuli, or resources as they call it (which may mimic natural conditions!), but express a different preference when presented with a larger set of resources, or when the conditions of the choice are changed. They distinguish three characteristics of a true choice, only one of which is met by an apparent choice:
- The choice must be non-random, i.e. that individuals must choose one resource more often than the others; testing only this criteria means that researchers are measuring apparent choice, while this is a necessary but not sufficient criteria for true choice. (I would add to this that the choice probability should be fairly stable if the animal is made to choose under exactly the same conditions).
- The choice should be the same even in the “absence of a differential response by the resource” (p. 332). The authors state this to avoid situations in which the resource (e.g. a potential mate) is manipulating the choice of the focal animal, a problem which reminds me very much of the literature on animal signalling.
- It should be demonstrated that every resource is perceived, to avoid issues of sensory bias and the like. It strikes me that this criterion will be hard to meet; for example, if while testing mate choice the researcher tries to demonstrate a lack of bias by showing responses by the focal individual to each of the potential mates in isolation, how does that prove that one or more of the potential mates aren’t being ignored when the focal individual is given the choice between all of them?
The importance of distinguishing between apparent and true choices depends on the objective of a study. If the objective is to establish which resources will be exploited under natural conditions, then the apparent choice is appropriate. If the experimenter wants to know which female will be mated by a male in a natural situation, then the results of this test (the apparent choice) will provide the answer. However, if the objective of the experiment is to establish the mechanisms of this choice, then it becomes important to look more closely at the results. If a male does not perceive a mated female as a resource because she does not produce sex pheromone, the male is thus inseminating virgin females as they are the only resource perceived. In this case, an apparent choice (the virgin female) is expressed, but this choice is the result of the non-perception of the mated female, which prevents this apparent choice from being a true choice. Measuring an apparent rather than a true choice does not remove the relevance of the test, but only modifies its interpretation. Consequently, it is important for the experimenter to state a clear question before identifying the adequate experimental setup to use.
I think that it’s important to mention here that the ideas expressed in this paper aren’t terribly groundbreaking; a number of people ranging from economics to psychology to behavioural ecology have, at one time or another, made largely the same argument or a variation thereof (one example of a related problem is raised by a really smart guy, Jeffrey Stevens, in this book chapter here). In fact, I’m a co-author on a paper currently in press at Behavioural Ecology talking about this issue from the opposite direction, wherein we argue that the mechanisms that underlie behaviour may be constrained and that these constraints need to be taken into account when assessing the evolution of behavioural outcomes. I even made an argument very much like the one in this paper during my Ph.D. synthesis exam!
Having said that, I like the paper for its laser-like focus on raising awareness about a very specific part of animal behaviour and cognition that can seriously undercut the conclusions drawn from experimental or field work if the appropriate test isn’t matched to the hypothesis the researcher wishes to explore. I suspect that their definition of apparent and true choices is incomplete and leaves out issues that will be hashed out in future papers, but if the journey of a thousand steps has to start somewhere, it’s not a terrible first stride.
. I’ll write more about this here when the paper is published.