The Moral Instincts of Infants

The Washington Post has an excellent piece with a provocative title.  “The disturbing thing scientists learned when they bribed babies with graham crackers”, by Ana Swanson, does not quite deliver on its title.  If anything, I think, what the scientists learned is the opposite of disturbing.  It does, however, deliver a fine summary of a very interesting study.  Moreover, it is bristling with links to scholarship on the subject, including the study it focuses on. 

The study, “Costly rejection of wrongdoers by infants and children,” by Arber Tasimi and Karen Wynn (Cognition 151 (2016) 76–79), came out last month.  It begins by placing the study’s focus in context. 

From infancy to adulthood, humans exhibit an aversion to individuals who treat others poorly. Even in the first months of life, infants reject agents who behave badly (Hamlin & Wynn, 2011; Hamlin, Wynn, & Bloom, 2007, 2010), and before their first birthday, not only avoid wrongdoers themselves, but expect others to do so as well (Kuhlmeier, Wynn, & Bloom, 2003)…

Here we ask about the strength of this aversion: Is it sufficiently powerful to lead people to resist one of the most alluring aspects of everyday life: profit?.

In a range of studies, children show a tendency to dislike persons who harm other persons.  This aversion emerges very early in life, even before the emergence of linguistic ability.  It is thus unlikely to be learned behavior and it is certainly not taught behavior.  Tasimi and Wynn conducted two experiments to measure whether infants and older children could be bribed to suppress this aversion to wrong doers. 

In the first experiment, one hundred and sixty children ages 5 to 8 were shown photographs of two fictitious benefactors Craig and Max.  Children randomly assigned to the baseline condition were told nothing else about the benefactors.  Then they were invited to accept a prize (stickers) from one of the two.  The offers were unequal.  Some of the children were offered one sticker from one benefactor and two from the other.  Some faced a one to four choice, others a one to eight choice, and others a one to sixteen choice.  The fictitious identities were randomly switched to control for name (or face?) preference. 

Not surprisingly, the children almost always chose the better offer.  Who wouldn’t?  That established a baseline measure: what the children would choose when they knew nothing about Craig and his buddy. 

Children randomly assigned to the character-information condition were presented with the same two fictitious persons but were also told that one of the two benefactors was mean.  He hit someone on the playground.  The other is always nice.  He hugged someone on the playground.  Then the children were divided into groups and faced the same assortment of offers: 1:2, 1:4, 1:8, and 1:16.  The contrast with the baseline condition was striking.  Fewer than 25% of the children accepted the offer of the mean benefactor when the offer was one to two.  Fewer than fifty percent accepted the offer from the mean person when the cost was one to four or one to eight.  Only in the case of a one to sixteen contrast did a majority of children make a deal with the devil.  The results were still slightly lower than the baseline results. 

The children were willing to pay a significant cost to deal with the do-gooder rather than the wrong-doer.  At the very least, this suggests that the children liked the one and disliked the other.  It may suggest that this is a case of altruistic punishment.  The subjects were willing to pay a personal cost to inflict a cost on a transgressor and to reward a helper.  Perhaps this is the same thing.

In the second experiment, the subjects were sixty-four 12 to 13 month old infants.  In this case the competing benefactors were rabbit puppets identified only by their orange or green shirts.  The prizes were graham crackers.  The same controls were instituted, with the rabbits switching shirts.  This time the distribution was either one to two or one to eight.  As in the first experiment, a baseline condition was tested and again the infants preferred more crackers to fewer. 

The children assigned to the character-information condition watched as one rabbit assisted a lamb puppet in opening a box to get at something the lamb wanted.  The other rabbit then slammed the box closed, frustrating the lamb.  The infants preferred the gift of one cracker from the good rabbit when the cost was one to two.  They held their nose and dealt with the bad rabbit when the bribe increased to eight over one. 

As I said earlier, I think that this is the very opposite of disturbing.  It suggests a robust moral instinct that emerges before the infants can talk or engage in “reputation management.”  The older children may have been worried about what the experimenter would think about their choices, but the infants were too young for that.  They just didn’t like what they were seeing when they saw the bad rabbit reveal herself.  To quote the authors:

The current findings show that a willingness to pay personal costs to avoid transactions with wrongdoers is an early-emerging and fundamental aspect of human nature. Our study contributes to a growing literature uncovering the origins and nature of social preferences, and extends this work by highlighting the psychological significance of social assessments to young humans.

This tells against the social science model according to which we are born amoral that morality is simply a social construct.  It appears that in fact we are by nature moral animals.  That is an important finding.  That children can be bribed is not.  They are not little angels any more than adults are. 

I think that there are profound consequences for ethics in this study.  Sacrificing self-interest for the sake of justice is both beautiful and good.  We may admire individuals who cannot be purchased at any price, and such persons frequently emerge as heroes in our literature.  The character of John Proctor in The Crucible comes to mind.  He ultimately sacrifices his life rather than confess to a lie.  That is beautiful.  And yet…

Do we really want our morality to be that inflexible?  Maybe.  Maybe not.  Here I am instructed by that moral authority, my beagle.  Bella knows that when I walk to her bowl with a scrap of food, she has to sit before she gets it.  I suppose I could easily teach her to wait before I say ok to gobble it up.  It’s a good thing that dogs can internalize such rules.  It made the alliance between our canine and human ancestors possible and so richly rewarding for both.  Beagle socialization reveals the primitive form of evolved psychological mechanisms that underwrites human morality. 

It would not do, however, for the beagle to be too good.  I wouldn’t want her to starve to death with food in front of her just because I wasn’t around to bless her meal.  So her evolved social instincts compete with her evolved appetite.  When she gets hungry enough, she will throw caution to the wind. 

By way of analogy, suppose that one of my loved ones is kneeling in a line of hostages.  A gunman is going down the line asking each person “are you an American?”  He shoots anyone who says yes and leaves unharmed anyone who says no.  What would I want my loved one to do?  I say lie.  Despite what John of Patmos says about liars, and despite my view that lying is immoral, I would prefer that self-interest trump righteousness in this case. 

If self-interest seems unimportant to you, then consider this scenario.  An SS officer is at your door asking you if you have seen any Jews in the neighborhood.  The truth is that you have.  There are six of them hiding in your basement.  I would argue that telling a lie in this instance is not only morally permissible, it is morally obligatory. 

We might wish that human beings were incorruptible, but probably we should not.  That would make us like the ants, blindly following rules with the capacity to deliberate.  Our corruptible nature is one of the costs of being genuinely moral beings.