Science Magazine Publishes CEU Research

January 7, 2011

Led by Agnes Melinda Kovacs, Department of Cognitive Science, CEU/Hungarian Academy of Sciences/ Cognitive Neuroscience Sector, International School for Advanced Studies (SISSA), a group of researchers learned through their recent breakthrough research that babies seem to understand the mental states of others much earlier than previously thought. The research, published in Science Magazine on 24 December 2010, reached the conclusion that 7-month-old babies already automatically take into consideration the knowledge of others. Erno Teglas, Department of Cognitive Science, CEU/Hungarian Academy of Sciences/SISSA, and Ansgar D. Endress, Massachusetts Institute of Technology (MIT)/SISSA were the other researchers in the project.

Humans are guided by internal states such as goals, desires and beliefs. The ability to attribute mental states to others is usually termed “theory of mind.” These abilities play a crucial role in ensuring the efficiency of social interactions, as they do not only allow us to predict the behavior of others, but also to adjust our own behavior before the predicted behavior has actually taken place. We encounter such situations on everyday basis. For instance, when intending to cross the street we notice that the person next to us looks in the wrong direction and does not see the approaching vehicle, we will try to make sure that this person will not become the victim of his or her own false beliefs.

While such situations would imply that humans spontaneously compute the beliefs of other people, research in the last 25 years has suggested that, at least under laboratory conditions, children do not seem to be able to take into consideration the false beliefs of others before the age of four. Moreover, it has been suggested that even in adults, reasoning about others’ beliefs might be laborious and effortful, just as it is laborious and effortful to mentally multiply 17 by 13.

In the study the researchers used a new paradigm to find out whether adults and infants as young as seven months automatically compute the beliefs of others, even in situations where the presence of another agent is irrelevant to their task. With adults, they used a simple visual detection paradigm, where participants had to detect the presence of a ball. It is known from every-day life experiences, and it is also proven by scientific studies that our expectations and knowledge modulate our behavior. For instance, when a person arrives at a crowded airport and spots his/her best friend, s/he will be much quicker to notice the friend if s/he knew in advance that the friend was waiting for him/her, as opposed to not knowing that the friend was waiting.

The researchers used the speeding up of behavior in response to expected events in a simple task, where adult participants watched short video animations with a screen that might hide a ball; once the screen fell, participants simply had to indicate whether they saw the ball. Just like the person at the airport would detect his/her friend more rapidly when s/he is expecting the friend, researchers found that participants of the test were faster to detect the ball when they had seen the ball rolling behind the screen, and expected to find it there, compared to a situation where participants had seen the ball rolling out of the scene, and, thus, did not expect to find it behind the screen.

The researchers were interested in some crucial conditions and asked whether not only the participants’ own beliefs would speed up their reactions, but also the “beliefs” of an animated agent they see in the films. In particular, researchers measured whether participants would be quicker to detect the ball if only the animated agent believed it to be behind the screen, participants have seen the ball rolling away. Specifically, if the agent walks out of the scene before the ball rolls away, s/he should “think” that the ball is still behind the screen. Participants were not only quicker in detecting the presence of the ball when they themselves believed the ball to be behind the screen, but also when the agent believed this to be so, even though the agent’s beliefs clashed with their own. This suggests that just watching animations involving an agent in which the scene changed in the absence of the agent, led participants to automatically compute the agent’s (false) beliefs, even though the presence of the agent was in fact irrelevant to the task they had to perform.

Researchers asked similar questions from 7-month-old infants. While adults are faster to react to stimuli that match their expectations, or the expectations they attribute to another agent, infants look longer at stimuli that do not match their expectations (or possibly the expectations they attribute to another agent). Results showed that infants looked longer at finding no ball behind the screen when the agent expected a ball behind the screen compared to when s/he expected no ball. In other words, just as adults, 7-month-old infants cannot help computing the beliefs of other agents.

The data of researchers suggest that human infants and adults automatically track the beliefs of other agents. Computing the beliefs of others does not seem to be particularly effortful; humans seem to do it as soon as they see other people, or even an animated agent. Seemingly, in implicit tasks, the beliefs of others influence the behavior of infants and adults similarly to their own beliefs.  While this finding may seem problematic for an individual, since it may make individual behavior susceptible to the beliefs of others that do not reliably reflect the current state of affairs, these abilities, in fact, enable the complex social interactions characteristic to human societies.

If you wish to obtain a copy of the article, please write to Melinda Harlov at the External Relations Office: