Evolutionary Psychology & Embodied CognitionPublished 28 January, 2013
A recent paper published in the journal that hosts this blog, Evolutionary Psychology, explores an idea drawn from a set of ideas that travel under the label “embodied cognition.” Storey and Workman report some research in which subjects played a Prisoner’s Dilemma game while holding something hot (chemical hand-warmers) or cold (freezer packs). They predict and report that holding something warmer rather than colder leads to more cooperation because there is, in their words, “a link between real world temperature sensation and feelings of psychological warmth.”
The “link” is built on two ideas, and I apologize to the authors if I’m not rendering either or both properly. From what I can tell, the idea is that close relationships during development are literally “warm” because of the close physical contact involved with hugs and such. This history of feeling warm during such encounters causes a psychological association between literal warmth and close relationships. The second source of the “link” is neurophysiological, and the authors draw on findings that suggest that trusting someone leads to activation in the same brain area as one implicated in processing temperature.
The most important thing that strikes me about this application of the notion of “embodied cognition” from the standpoint of an evolutionary/functional analysis is that the proposal here is that, whatever the cause for the “link,” the effect of the link is to cause people to make mistakes. In the context of relationships – or decisions to cooperate or defect in an interaction – there are a number of cues that might reasonably be useful to attend to when making a decision. Is this a person I’ve known and will continue to interact with? What is the magnitude of the possible loss if I cooperate? Is there evidence that this person is disposed to like (dislike) me? These should, as normative and adaptive matters, be taken into account when making decisions about social interaction. If there is a long shadow of the future, for instance, then the expected value from subsequent interactions is greater.
In the reported work, subjects play ten rounds of a Prisoner’s Dilemma game, two blocks of five rounds holding either the hot or cold object, with the blocks counterbalanced. If I have understood the analysis correctly, the authors report that the payoff people earn is higher when the subjects in a PD pair are holding the hot object compared to the cold object. (My interest here is in the theory, not the stats, but it seems to me that the correct analysis is the frequency of Cooperate choices (a binary variable) by individuals rather than the payoff data to pairs, but it’s possible I misunderstood the analysis.)
In any case, to return to the theory, the temperature of what one is holding is a non sequitur. And I mean that in the literal, logical sense. The temperature of the room or the temperature of one’s hands is not any sort data on which such decision-making ought to be based, like choosing a mate because of their sign of the zodiac, or some other arbitrary, irrelevant property. Therefore, to the extent that temperature influences decisions, these decisions are being affected by irrelevant information, and are in error just to the extent that these effects are felt.
The ideas that motivate research of this type in the embodied cognition literature, then, rely on what strike me as a peculiar sort of byproduct. In the present case, the fact that two properties co-occur leading one to use the presence of one property when making decisions about the latter. There seems no obvious reason that noticing such co-occurrences ought to have such an effect. Further, the association seems to be particularly imperfect. People also get warm when they’re angry, possibly explaining why we have expressions such as hot under the collar. In Steven Brust’s novel, Five Hundred Years After, one character is explaining why he took care to be armed, saying:“Your Venerance, my master the Baroness gave me to understand the affair might become tolerably warm,” again illustrating how one might, from such constructions, have derived the reverse prediction.
More generally, it strikes me as just an odd sort of byproduct. Would it really have been that hard to engineer the trust system in such a way that it simply ignores the extraneous and irrelevant information about the temperature of one’s hands? Now, before the accusations of a commitment to “optimality” fly – lookin’ at you, Gary Marcus – of course there are any number of reasons that the mind might not be optimally designed. True enough. But as side effects go, this one seems particularly odd or, to put it another way, particularly easy to engineer out.
Further, it seems to me that there are just many facts about the world that don’t fit the notion. If being a little warmer makes you more trusting, should we find that our friends in the hot Philippines by and large are more trusting than our colder friends in Norway? It seems we should. But they’re not. By a far cry (see Fig. 1).
I might note that an additional source of motivation for the Storey and Workman paper is that it builds on prior results in this area, especially that of John Bargh and colleagues. I haven’t been following this work closely, but my sense is that there are at least some worries that the initial findings can be replicated. Again, I’m not an expert in this area, but it was a bit surprising to me that some concerns that have arisen in this area were not discussed by the authors.
The larger point is that most of the work of this type in embodied cognition, it seems to me, posits that the mind has the peculiar property of using a ton of irrelevant information in making decisions. Not only that, but there are any number of bits of irrelevant information the mind might use in making decisions. I myself have never really understood how one chooses which bits of irrelevant information are good candidates. Now, again, don’t get me wrong, I suspect that there are any number of cases in which the mind does indeed use irrelevant information – perhaps findings under the umbrella of the halo effect might be good exemplars. But information such as “heat” seems to me like an odd candidate for having an effect on social decision making.