Some time ago, I alluded to a very real moral problem: Observed behavior, on its own, does not necessarily give you much insight into the moral value of the action. While people can generally agree in the abstract that killing is morally wrong, there appear to be some unspoken assumptions that go into such a thought. Without such additional assumptions, there would be no understanding why killing in self-defense is frequently morally excused or occasionally even praised, despite the general prohibition. In short: when “bad” things happen to “bad” people, that is often assessed as a “good” state of affairs. The reference point for such statements like “killing is wrong”, then, seems to be that killing is bad, given that it has happened to someone who was undeserving. Similarly, while most of us would balk at the idea of forcibly removing someone from their home and confining them against their will to dangerous areas in small rooms, we also would not advocate for people to stop being arrested and jailed, despite the latter being a fairly accurate description of the former.
Figuring out the various contextual factors affecting our judgments concerning who does or does not deserve blame and punishment helps keep researchers like me busy (preferably in a paying context, fun as recreational arguing can be. A big wink to the NSF). Some new research on that front comes to us from Hamlin et al (2013), who were examining preverbal children’s responses to harm-doing and help-giving. Given that these young children aren’t very keen on filling out surveys, researchers need alternative methods of determining what’s going on inside their minds. Towards that end, Hamlin et al (2013) settled on an infant-choice style of task: when infants are presented a the choice between items, which one they select is thought to correlate with the child’s liking of, or preference for, that item. Accordingly, if these items are puppets that infants perceive as acting, then their selections ought to be a decent – if less-than-precise – index of whether the infants approve or disapprove of the actions the puppet took.
In the first stage of the experiment, 9- and 14-month old children were given a choice between green beans and graham crackers (somewhat surprisingly, appreciable percentages of the children chose the green beans). Once a child had made their choice, they then observed two puppets trying each of the foods: one puppet was shown to like the food the child picked and dislike the unselected item, while the second puppet liked and disliked the opposite foods. In the next stage, the child observed one of the two puppets playing with a ball. This ball was being bounced off the wall, and eventually ended up by one of two puppet dogs by accident. The dog with the ball either took it and ran away (harming), or picked up the ball and brought it back (helping). Finally, children were provided with a choice between the two dog puppets.
Which dog puppet the infant preferred depended on the expressed food preferences of the first puppet: if the puppet expressed the same food preferences as the child, then the child preferred the helping dog (75% of the 9-month-olds and 100% of the 14-month-olds); if the puppet expressed the opposite food preference, then the child preferred the harming dog (81% of 9-month-olds and 100% of 14-month-olds). The children seemed to overwhelming prefer dogs that helped those similar to themselves or did not help those who were dissimilar. This finding potentially echos the problem I raised at the beginning of this post: whether or not an act is deemed morally wrong or not depends, in part, on the person towards whom the act is directed. It’s not that children universally preferred puppets who were harmful or helpful; the target of that harm or help matters. It would seem that, in the case of children, at least, something as trivial as food preferences is apparently capable of generating a dramatic shift in perceptions concerning what behavior is acceptable.
The effect was then mostly replicated in a second experiment. The setup remained largely the same with the addition of a neutral dog puppet that did not act in anyway. Again, 14-month-old children preferred the puppet that harmed the dissimilar other over the puppet that did nothing (94%), and preferred the puppet that did nothing over the puppet that helped (81%). These effects were reversed in the similar other condition, with 75% preferring the dog that helped the similar other over the neutral dog, and preferred the neutral over the harmful puppet 69% of the time. 9-month-olds did not quite show the same pattern in the second experiment, however. While none of the results went in the opposite direction to the predicted pattern, the ones that did exist generally failed to reach significance. This is in some accordance with the first experiment, where 9-month-olds exhibited the tendency to a lesser degree than the 14-month-olds.
So this is a pretty neat research paradigm. Admittedly, one needs to make certain assumptions about what was going on in the infant’s heads to make any sense of the results, but assumptions will always be required when dealing with individuals that can’t tell you much about what they’re thinking or feeling (and even with the ones who can). Assuming that the infant’s selections indicate something about their willingness to condemn or condone helpful or harmful behavior, we again return to the initial point: the same action can be potentially condemned or not, depending on the target of that action. While this might sound trivially true (as opposed to other psychological research, which is often perceived to be trivially false), it is important to bear in mind that our psychology need not be that way: we could have been designed to punish anyone who committed a particular act, regardless of target. For instance, the infants could have displayed a preference towards helping dogs, regardless of whether or not they were helping someone similar or dissimilar to them, or we could view murder as always wrong, even in cases of self-defense.
While such a preference might sound appealing to many people (it would be pretty nice of us to always prefer to help helpful individuals), it is important to note that such a preference might also not end up doing anything evolutionarily-useful. That state of affairs would owe itself to the fact that help directed towards one individual is, essentially, help not directed at any other individual. Provided that help directed towards one person is less likely to pay off in the long run (such as individuals who do not share your preferences) relative to help directed towards others (such as individuals who do share you preferences), we ought to expect people to direct their investments and condemnations strategically. Unfortunately, this is where empirical matters can become complicated, as strategic interests often differ on an individual-to-individual, or even day-to-day basis, regardless of there being some degree of overlap between some broad groups within a population over time.
Finally, I see plenty of room for expanding this kind of research. In the current experiments, the infants knew nothing about the preferences of the helper or harmer dogs. Accordingly, it would be interesting to see a simple variant of the present research: it would involve children observing the preferences of the helper and harmer puppets, but not the preferences of the target of that help or harm. Would children still “approve” of the actions of the puppet with similar tastes and “disapprove” of the puppet with dissimilar tastes, regardless of what action they took, relative to a neutral puppet? While it would be ideal to have conditions in which children knew about the preferences of all the puppets involved as well, the risks of getting messy data from more complicated designs might be exacerbated in young children. Thankfully, this research need not (and should not) stick to young children.
References: Hamlin, J., Mahajan, N., Liberman, Z., Wynn, K., (2013). Not like me = bad: Infants prefer those who harm dissimilar others. Psychological Science.