> Specifically, we found that after selecting from two equally attractive objects, infants subsequently devalued the initially unchosen object. We ruled out the possibility that this reflects a desire for novelty or any intrinsic preference among the options.
The experimental design that allowed them to do this is simple, but pretty clever. The original version of the experiment is based on first letting the baby choose between toys A and B, and then demonstrating that if A was chosen, the baby prefers a new toy C over B. (All of the toys were equally likely to be preferred by a control group.)
Since it's not possible to know whether the initial choice was random or based on individual preference, they did a modified version of this experiment. After the baby is shown the first two toys, they are placed under boxes and secretly swapped. Despite originally having chosen B, the baby instead gets to interact with A. In this case, the babies still fairly consistently choose C (the new toy) over B (the one they originally preferred but didn't get to interact with).
And as the press release does mention, simply giving a baby a toy randomly chosen by the experimenter doesn't cause the same effect. So the reasonable inference is that the actual experience of making a choice (or at least having the illusion of doing so) reduces one's attachment to the option that wasn't chosen.