Wednesday, April 13, 2016

Young Infants Learn By Doing

The brain is an extraordinarily complex organ that we are struggling to understand scientifically. To help us understand the way the brain works, we often use the most complex machines of an era to give us insight into what the brain is doing.  And starting in the 1950’s, the most complex machine we had was the computer.  

One influence of this computer metaphor for the mind is that it has emphasized the information that is processed in the brain over the relationship between mind and body.  After all, computers have poor bodies.  They have a keyboard and mouse as a sensory system and a display and internet connection as ways of acting on the world around it.  As a result, most research in psychology focuses on how the brain processes particular kinds of information without much focus on how the body influences that thought.

There are some streams of research that buck this trend.  In research on adults, there is work on situated cognition that explores how the environment affects the way people think.  There are also studies in embodied cognition that examine how body and mind interact.

Over the years, there has also been work on how the body influences learning in infants and children.  In my last blog entry, for example,  I talked about how experience moving influences children’s curiosity about heights.

A fascinating paper in the February, 2014 issue of Child Development by Sarah Gerson and Amanda Woodward examines how 3-month-olds learn about other people’s goals from their own actions.

If you have ever spent any time with a 3-month-old, then you know that they are not particularly coordinated.  Jessica Sommerville, Amanda Woodward, and Amy Needham developed an ingenious method for looking at actions in 3-month-olds (reported in a paper in the journal Cognition in 2005).  They attached mittens with Velcro on them to infants’ hands.  Then, the infants could reach out and move toys that also had Velcro on them.

In Gerson and Woodward’s study, all of the infants had a chance to sit with the toys for a few minutes without the mittens.  Then, some infants had the mittens put on, and they reached out and played with the toys for a few minutes.  For these infants, playing with the toy generally meant moving it back-and forth across the table.  A second group of infants was matched with the infants from the first group.  For each infant in the second group, the experimenter mimicked the actions of the infant using Velcro gloves, so that this infant got to watch something similar to what the first infant did.  That provides a control condition in which the infants are exposed to the actions without actually performing the actions themselves.

After this initial exposure, infants were tested.  It is hard to test 3-month-olds, because they cannot speak or understand much language.  Instead, researchers typically use a looking procedure.  

In this study, the two toys that were used in the first part of the study were placed on a stage.  A gloved hand reached out and picked up one of the toys.  This event was presented several times until the infants stopped looking at it.  This decrease in looking time is called habituation and is used as a measure that the children can predict what is going to happen.

Then, the positions of the two toys were switched.  So, the one on the left is now on the right and vice versa.  Now, two types of test trials were given.  In one type of test trial, the hand reached out and grabbed the toy that was now on the left.  In this case, the hand performed the same action, but the outcome was different.  In the other type of test trial, the hand picked the same object that it selected before.  In this case, the movement was different (because the hand had to reach all the way to the right), but the outcome was the same (because the same object that was touched earlier was selected again).  

The experimenters measured how long infants looked at each type of test.  The longer infants look, the more interested they are in that outcome, generally because that outcome is different from what they expected.

The infants who watched someone else acting on the toys did not look reliably longer at either type of event. Those infants who acted on the toys themselves looked for a longer time at the event with a different outcome than at the event with the same outcome.  That is, by playing with the toys, infants seemed to learn that other people often direct their actions at particular outcomes rather than just making particular movements.

The infants learned something very specific from this experience, though.  A third group of infants played with the toys using the mittens, but then the test trials involved a different set of toys than the ones they played with.  This group also did not look reliably longer at either type of test event.  That is, the infants learned only about outcomes involving the specific toys they played with.

What does all of this mean?

One way that infants learn about what other people want to do is by learning about how they perform actions in the world.  As they learn that their actions can be directed toward particular outcomes, they also learn that other people may act to bring about particular outcomes.    

However, it also appears that this learning is very conservative.  That is, infants start by learning about their interactions with particular objects.  It is valuable to keep from generalizing too far from early experience, because there are likely to be many exceptions to the early rules that infants learn.

More broadly, this research demonstrates the importance of acting on the world when learning.  Too often, even adults try to learn passively by listening to others and reading superficially.  It is important to engage with the world when learning rather than just treating the process of learning as if we were computers that were being fed information.