The Moral Life of Babies

In my last post, I raised some questions about Paul Bloom's article in today's Times Magazine entitled, “The Moral Life of Babies.” In it, Bloom summarizes some of his findings about the “naive morality” that babies possess from the moment they are born. From this finding, Bloom concludes not only that there is a universal morality, but that morality has a biological basis.

In my last post, I tried to argue that it is not necessarily obvious that there is a universal morality. It is far from clear that we could discover any single moral principle that all societies share. If we can't, then there is no universally shared morality, how could morality have a biological basis?

Today, I would like to question Bloom's conclusions from a slightly different angle. Let's stipulate that all societies value compassion (just for the record, this seems unlikely to me). Bloom wants to argue, on the basis of his research, that this value is biologically based. He hypothesizes that if babies exhibit this value, then presumably it is encoded in their biology rather than coming about by socialization.

However, there is a problem. How are can we know what a baby is thinking? Not only are babies unable to speak, but newborns can barely move. However we investigate babies moral values, it will have to be indirectly. Bloom has a solution. For some time, infant psychologists have believed that “the eyes are a window to the baby’s soul. As adults do, when babies see something that they find interesting or surprising, they tend to look at it longer than they would at something they find uninteresting or expected. And when given a choice between two things to look at, babies usually opt to look at the more pleasing thing. You can use 'looking time,' then, as a rough but reliable proxy for what captures babies’ attention: what babies are surprised by or what babies like.”

According to infant cognition specialists, if a baby looks at one thing longer than another, the baby either prefers it or is surprised by it. For example, an infant will look at his or her mother's face longer than an unfamiliar face presumably because the baby prefers his or her mother's face to the stranger's.

Similarly, a baby who witnesses an impossible event (like one animated object passing through another) and a possible event (two animated objects bumping to each other), the baby will look longer at the impossible event presumably because the baby is surprised.

Here is an example of one of Bloom's experiments that shows how he used “look time” to deduce babies' moral values:

“Our experiments involved having children watch animated movies of geometrical characters with faces. In one, a red ball would try to go up a hill. On some attempts, a yellow square got behind the ball and gently nudged it upward; in others, a green triangle got in front of it and pushed it down. We were interested in babies’ expectations about the ball’s attitudes — what would the baby expect the ball to make of the character who helped it and the one who hindered it? To find out, we then showed the babies additional movies in which the ball either approached the square or the triangle. When the ball approached the triangle (the hinderer), both 9- and 12-month-olds looked longer than they did when the ball approached the square (the helper). This was consistent with the interpretation that the former action surprised them; they expected the ball to approach the helper.”

Bloom interprets his observations as follows: Because the yellow square helped the ball, the babies expected the ball to approach the square. When they saw an animation in which the ball approached the triangle, who had hindered the ball, the babies were surprised, which the babies indicated by looking at this animation longer.

Recall that longer look time can either indicate preference or surprise. How did Bloom know whether the baby was surprised that the ball approached the triangle, or whether the baby preferred to see the ball approach the triangle?

Perhaps babies looked longer at the ball approaching the triangle because they liked that animation better. Maybe babies are totally immoral. They desire the opposite of what we would call moral behavior. They wanted the ball to approach the triangle that had mistreated it, and therefore, they looked longer at this preferable outcome.

Here's another: perhaps babies share adult morality, but they believed that the ball was approaching the triangle to attack it. Again, this outcome seemed preferable and so the babies looked at it longer.

Bloom seems to interpret the babies' increased look time as expressing surprise because an Bloom (and those who share his moral outlook) would be surprised to see a person react in a friendly way to someone who had hindered them. The point is that Bloom is reading his own morality onto the baby.

It seems like no matter what the baby does, the researcher can interpret that baby either as saying “I like what I see,” or “I am surprised by what I see.” One of these two statements probably applies in almost any situation. Imagine a baby who watches a video of a person hitting another person. And then watches a video of a person hugging another person. The baby looks longer at the hugging video than the hitting video. This could either mean that the baby finds hugging surprising, or that the baby prefers hugging over hitting. What conclusion can be drawn from this experiment?

Can we even assume that the baby understands the animation of the ball, square and triangle, maybe the baby sees the triangle as trying to pull the ball up the hill, while the square is trying to drag it back down. The researcher needs to supply a great deal of information to be able to make sense of the baby's reactions. Basically, Bloom assumes that the baby interprets the events similar to the way the Bloom interprets the events.

How much of the moral lives of babies is in the heads of the baby and how much is in the mind of the theoretician? Is it possible that we so badly want babies to possess rudimentary morality that we will interpret any observations such that they prove what we want them to prove?