|Top: Interior decorating skills - a highly desirable |
attribute in male bowerbirds
Bottom: One operationalization of the preference
for "nest fanciness" in bowerbirds, r = .54
(Borgia, 1995). x-axis = # of decorations;
y-axis = # of copulations.
Now, imagine you wish to test the hypothesis that females have a preference for males who are able to construct fancier nests. You might examine the extent to which (a) the nest fanciness of several male bowerbirds predicts (b) the extent to which female bowerbirds want to mate with them. In fact, this association is fairly strong (see graph at right), suggesting that bowerbird females have a strong preference for the attribute “nest fanciness” in their mates.
Meanwhile, if you study humans rather than bowerbirds, you might want to test the hypothesis that women have a parallel mate preference for fanciness of abode. You could examine the extent to which (a) the apartment fanciness of several men predicts (b) the extent to which women want to date each of them.
I could be wrong, but I don’t think anyone has conducted this study with humans. And that might be because there is a second, completely different way of studying preferences for attributes when working with human participants: You just ask them. After all, asking is much easier: Humans (unlike bowerbirds) can simply rate the appeal of the attribute “fancy apartment” on a 1-9 scale. 
But here’s the problem: Asking humans about their mate preferences assesses their ideas about the attributes they like and dislike, rather than the extent to which an attribute actually drives their mate preferences in real life. In other words, these two mate preferences are not the same construct. Rather than providing a quick-and-easy measurement shortcut to the human analogue of bowerbird mate preferences, rating scales provide a tool for measuring a different—and perhaps uniquely human—type of mate preference.
As this new paper discusses, these two types of preferences are distinct enough that they deserve different names. As a nod to the animal behavior literature, we call the first one—the association between (a) the level of an attribute in each of a series of targets and (b) liking for each of those targets—a functional preference.  We call the second one a summarized preference because it reflects a person’s evaluative summary of the attribute as an overall concept. A functional preference for an attribute is the extent to which the attribute drives liking for a set of targets (e.g., the extent to which the intelligence of a potential partner drives the extent to which you like them). A summarized preference for an attribute is the extent to which a person likes the attribute itself, as a concept (e.g., your evaluation of the attribute “intelligence in a romantic partner”).
The paper linked above reviews how functional and summarized preferences fundamentally differ in a number of ways. For one, they have different evolutionary origins. Functional preferences should exist in any species that possesses an evaluative mechanism (e.g., these food sources are good, and these are bad). But a summarized preference requires an organism to be able to evaluate an attribute as an abstraction—an evolutionarily much more recent ability—and it seems plausible that only humans can do this. What’s more, functional and summarized preferences tend to be biased by different sources of information, and they may have different downstream consequences.
In the existing mate preferences literature, the summarized preference is the construct you will see most often. That’s okay if you intend to study people’s ideas about the attributes they like. But it’s not a shortcut to studying functional preferences: Depending on the context, the correlation between a summarized and functional preference for the same attribute ranges from r = ~.00 (if people are evaluating live interaction partners) to r = ~.20 (if people are evaluating photographs). In other words, although both constructs are interesting, they’re not the same thing.
And if we want to understand mate preferences in humans, we have to stop conflating summarized and functional preferences; overlooking this distinction creates a construct validity nightmare. For example, many scholars take inspiration from the animal literature on mating to generate new predictions about human mating. But they then test these predictions with summarized preferences—preferences that have no conceptual parallel in nonhuman animals and would not have been subject to the same evolutionary pressures. If you are generating predictions about human mate preferences with evolutionary relevance, your construct—nine times out of ten—is the functional preference.
So if you work with humans, you can choose to study how strongly attributes predict evaluative outcomes (i.e., functional preferences), or you can study people’s ideas about the attributes they like (i.e., summarized preferences), or you can study both and the relationship between them. Just be mindful of and clear about which construct(s) you are studying. If you work with nonhuman animals, you are almost surely studying functional preferences...unless you have figured out how to get bowerbirds to fill out a pen-and-paper survey.
Borgia, G. (1995). Complex male display and female choice in the spotted bowerbird: Specialized functions for different bower decorations. Animal Behaviour, 49, 1291-1301.
Fletcher, G. J., Simpson, J. A., Thomas, G., & Giles, L. (1999). Ideals in intimate relationships. Journal of personality and social psychology, 76, 72-89.
Ledgerwood, A., Eastwick, P. W., & Smith, L. K. (in press). Toward an integrative framework for studying human evaluation: Attitudes towards objects and attributes. Personality and Social Psychology Review.
Wood, D., & Brumbaugh, C. C. (2009). Using revealed mate preferences to evaluate market force and differential preference explanations for mate selection. Journal of personality and social psychology, 96, 1226-1244.
 “Nice house or apartment” is, in fact, one item on a very popular mate preference scale (Fletcher et al., 1999) – a scale that I myself use all the time.
 Wood and Brumbaugh (2009) offer the most comprehensive prior treatment of this construct, which they called a “revealed preference.” We shied away from the “revealed preference” label primarily because behavioral economists use this term to refer to an observable behavior. We mean something far more specific (i.e., the association between an attribute and liking within a set of targets).