12 month olds, but not 6 month olds predict other people's actions

No, this is not like voodoo prediction where they will know what will happen 12 years hence.

All of us, however, are capable in degrees of predicting what is going to happen over short time scales. This predicition falls into two general categories. First, we can predict the behavior of inaminate objects such as knowing how a ball will flight when we hit it just-so with a bat. That implies that we understand how physics work on some inituitive level.

Second, we can understand how animate objects such as people behave. For example, if I see someone removing objects from a container to a place on the table, I can reasonable predict that where they will move new objects based on a similar course that the old objects took. It is a prediction based on consistencies in what I observed before about human motion, and it is based on the premise that people operate in predictable ways. However, to do that I need to first recognize that the other person is a person, that they possess agency in psychology speak.

Much has been written about the neurological machinery that allows people to recognize agency. (I posted on this issue with respect to morality a couple weeks ago.) The ability to recognize agency has often been attributed to the Mirror Neuron System (MNS) -- a set of neurons in the brain that activate when a person views another object or person to which they attribute agency.

In understanding this development of agency and the MNS, it is reasonable to ask whether they develop at the same time. We know that the MNS is active by around a year of age. The question is whether skills such as the prediction of human motion are also active around that age.

Terje Falck-Ytter et al, publishing in Nature Neuroscience, have addressed this issue:

Recently, a strong claim has been presented about the role of the MNS in human ontogeny. According to the MNS hypothesis of social cognition, the MNS constitutes a basis for important social competences such as imitation, 'theory of mind' and communication by means of gesture and language. If this hypothesis is correct, the MNS should be functional simultaneous with or before the infant's development of such competencies, which emerge around 8-12 months of life. Furthermore, according to the MNS theory, proactive goal-directed eye movements during the observation of actions reflect the fact that the observer maps the observed actions onto the motor representations of those actions. This implies that the development of such eye movements is dependent on action development. Therefore, infants are not expected to predict others' action goals before they can perform the actions themselves. Infants begin to master the action shown in our study at around 7-9 months of life. Thus, if the MNS is a basis of early social cognition, proactive goal-directed eye movements should be present at 12 but not at 6 months.

Although habituation studies indicate that young infants distinguish between means and ends when observing actions, no one has tested the critical question of when infants come to predict the goals of others' actions. Using a gaze-recording technique, we tested the MNS hypothesis by comparing the eye movements of adults (n = 11), 12-month-old infants (n = 11) and 6-month-old infants (n = 11) during video presentations showing nine identical trials in which three toys were moved by an actor's hand into a bucket (Test 1). We compared gaze behavior, consisting of (i) timing (ms) of gaze arrival at the goal (the bucket) relative to the arrival of the moving target and (ii) ratio of looking time in the goal area to total looking time in combined goal and trajectory areas during object movement.

In adults, the MNS is only activated when someone is seeing an agent perform actions, not when objects move alone. However, according to the teleological stance theory14, seeing a human agent is not necessary for infants to ascribe goal-directedness to observed events. This theory states that objects that are clearly directed toward a goal and move rationally within the constraints of the situation are perceived as goal-directed by 12-month-old infants. This implies that seeing an interaction between the actor's hand and the toys is not necessary for eliciting proactive, goal-directed eye movements. By comparing the gaze performance of adults (n = 33) and 12-month-olds (n = 33) in three conditions, we evaluated this alternative hypothesis (Test 2). The first condition, 'human agent', was identical to the one in Test 1. In the 'self-propelled' condition, the motion was identical to that in the human agent condition except that no hand moved the toys. We also included a 'mechanical motion' condition. We assigned subjects randomly to the conditions, and sample sizes were equal. (Citations removed. Emphasis mine.)

Basically what the authors did was the following. You show adults, 12 month olds, and 6 month olds videos of objects either being moved by a person -- the human agent condition (HA) -- or being moved by their own volition. The own volition has two types. In one type, they move on the same courses that they would have by the person. This is called self-propelled (SP). In the other type, they follow computer constructed courses that do not necessarily resemble those that they would take if a person picked the object up. This is called mechanical motion (MM). (This last control is necessary because you would like to distinguish whether the ability of the subject to predict is related to whether the motion is biological -- looks like it was generated by a person -- or not.)

The end-points that the experimenters measured was the tendency of the subjects to look where the objects are going. This gets back to the example that I gave at the beginning of the post. If we can infer agency in a process, then we can predict the outcome of this process. One of the ways we show that we have predicted the outcome is by looking where the object is going to go rather than where it is right now.

Here are the results (click to enlarge):

On the right you can see the arrival time of the subjects gaze prior to the arrival of the object -- how much their gaze beat the object there suggesting how quickly they predicted the arrival.

You can see that for human agency type motion, both the adults and the 12 month olds predicted the arrival earlier than it got there. However, for 6 month olds, they do not predict with their gaze the arrival of the object. Interestingly, for adults and 12 month olds, the prediction did not take place when the object was not seen to be moved by a person -- the SP and MM conditions.

On the left you can see that this data is recapitulated by the ratio of the time spent gazing at the goal to the total gaze time.

These results show that adults and 12 month old possess a similar understanding of agency in human movements, but that 6 month olds do not.

What does this tell us about how the understanding of agency develops and the MNS?

The authors summarize three conclusions:

The present study supports the view that action understanding in adults results from a mirror neuron system that maps an observed action onto motor representations of that action. More importantly, we have demonstrated that when observing actions, 12-month-old infants focus on goals in the same way as adults do, whereas 6-month-olds do not (Test 1). The performance of the 6-month-olds cannot originate from a general inability to predict future events with their gaze, because 6-month-olds predict the reappearance of temporarily occluded objects. Finally, in terms of proactive goal-directed eye movements, we found no support for the teleological stance theory claiming that 12-month-old infants perceive self-propelled objects as goal-directed (Test 2).

The interesting one to me is the bit about predicting actions in 6 month year olds. We know that if you roll a ball behind a screen, babies at even 6 months of age can predict that it will come out the other side -- that it obeys physical laws. Their deficit in this regard is not general prediction, but prediction of the behavior of systems that express agency. This deficit subsides when they reach 12 months of age, in a time scale that parallels the development of the MNS.

Also, the so-called teleological stance theory -- the agency is inferred in objects showing directed movements is not accurate. The 12 month old babies can distinguish between objects that look like they have agency and objects that are actually being moved by people.

I love experiments like this 1) because they involve babies and I think babies are cool and 2) because it gets to the nitty gritty details of how human beings recognize each other. It's cool to think that 12 month olds are already on top of that.

You know how babies develop stranger anxiety around 9 months. I wonder if the development of agency and the understanding that there are other people in the world has something to do with that.

Hat-tip: Faculty of 1000.