Welcome to

Mind Matters

where top researchers in neuroscience, psychology, and psychiatry explain and discuss the findings and theories driving their fields. Readers can join them. We hope you will.


This week:

How Observation Beats the School of Hard Knocks




by David Dobbs
Editor, Mind Matters

;Experience keeps a dear school, but fools will learn in no other." So Benjamin Franklin summed up the high costs of first-hand learning; the wise among us, he suggests, will buy knowledge less dearly by picking it up second-hand. In an instructive essay below, Columbia University cognitive researcher Kevin Ochsner describes a recent study that illuminates what brain activity distinguishes such vicarious learning from the school of hard knocks. Read it, wonder -- and be glad you don't have to learn everything the hard way.


I Watch, Therefore I Know

Kevin Ochsner
Columbia University
New York, NY

Few questions are more fundamental than that of how we learn. Indeed, this question has been central to psychological inquiry from the time of the first experimental psychology labs in the late 1800s. Ever since, a primary goal of psychology research has been to describe how we acquire and retain the information necessary for survival. Most of this work, however, has concerned direct, first-person learning. There is another mode of learning, however, famously alluded to by Yogi Berra when he said, "You can observe a lot just by watching." As these sage and inimitable words suggest, we learn not just through direct experience but also by observing others' experiences. Although (vicarious) learning through observation is common, and in many situations may be more adaptive and efficient than learning through direct experience, few researchers had tried to unpack the bases of this ability in the brain until recently. The study under review here, "Learning Fears by Observing Others: The Neural Systems of Social Fear Transmission," by Andreas Olsson and Elizabeth Phelps, takes the exploration of vicarious learning to a new level. Imaging observational learning In this brain imaging study, Olsson and Phelps recorded neural activity in participants who were watching a video clip of someone learning through classical fear conditioning, which involves learning to associate a neutral stimulus (like a colored shape) with an intrinsically aversive stimulus (in this case a shock). After a few pairings of shape and shock, the fear learning that occurs can be measured most easily as a change in electrical resistance on the surface of the hand as one sweats in anticipation of the shock after seeing the shape -- known as a change in skin conductance response. The classical fear conditioning paradigm has been used by Joseph LeDoux, Elizabeth Phelps and others to identify a subcortical brain structure known as the amygdala that is essential for this simple, first-person, form of associative learning in both animals and humans. (For a review, see Ledoux, 2000). The amygdala's activation is crucial to this sort of learning. In functional brain scans, it will "light up" (that is, show more blood flow, which indirectly indicates activation) when a person or rat encounters a stimulus (such as a colored shape or a sound) that has become associated with a shock or other aversive stimulus. In Olsson and Phelps's study, however, the participants did not experience the shape-shock pairing directly; instead, they saw an actor in a video experiencing it. Yet the researchers found that the participants' amygdalas were active while they were watching the video, as if they were learning to associate the shock and shape themselves. And after the learning had occurred, the skin conductance showed fear responses to the shapes, also as if they had experienced the learning directly. This suggests that we learn from watching others by engaging the same learning mechanisms we use when learning first-hand. How alike are first- and third-person learning? This finding joins a growing number of similar findings in the newly emergent fields of social cognitive and affective neuroscience. Several studies have shown that watching another person perform an action or experience pain or emotion activates many of the same brain systems that are engaged when one experiences these things directly (See Decety & Grezes, 2006; Singer et al., 2004; Wicker et al., 2003). Such findings have given rise to the notion that we understand others by directly experiencing what it would be like to be them. This work has not yet shown, however, whether or how activity in these systems actually predicts behavior. One problem is that brain imaging is still a coarse tool, and it cannot resolve the content of thoughts at a fine grain. For example, although activity in the color-sensitive area of visual cortex might indicate that someone is thinking about something colorful (for example, a flower), it can't tell what specific color they have in mind (such as red). Similarly, when we see activity in the amygdala or some other brain system associated with emotion, we can't know precisely what experience someone is having. For vicarious learning, this means that the observer and the observed person may display similar patterns of brain activity in a functional imaging study even though they may not be having exactly the same feeling. Watching someone be beaten, for instance, you might feel anger while the victim is feeling mainly fear. Olsson and Phelps at least partly overcome this limitation by showing that activity in the amygdala as well as two other brain systems -- the insula and medial prefrontal cortex (mPFC) -- predicted how well participants learned. The amygdala finding is fundamental, for it is well established that activity in the amygdala typically predicts classical fear conditioning through first-person experience. If you see the amygdala is active in fear conditioning, you can be pretty confident the lesson -- the association of shock with color, for example -- is being learned. Atop this, however, Olsson and Phelps found consistent, predictive activity in the medial frontal cortex, an area strongly associated with the ability to understand other people's behavior in terms of what they are feeling or thinking. This provides one of the first findings that regions involved in mental state attribution play a role in social learning. And it clearly suggests that observational learning may depend on more than simple associative learning mechanisms. Both the direct and observational learning create an association of color with shock. But the observational learning may involve a complex social cognitive process -- likely an analysis of the observed person's behavior -- along with the more basic associative learning power of the amygdala. But how are these two processes related? Do the more cognitive processes of the prefrontal cortex mediate amygdala-based fear learning? And if so, will those neural dynamics generalize to other forms of vicarious or observational learning? Future work will be necessary to address these issues. One step beyond Tthis study suggests intriguing potential applications and implications. For clinical contexts, the basic experimental model used here could be adapted to explore impairments in observational learning -- to ask, for example, whether and how individuals with deficits in the ability to attribute mental states, such as autism and Asperger's syndrome, show deficits in social learning. Is the conversation between medial prefrontal cortex, insula, and amygdala different in these disorders? This model could also be used to explore whether humans and non-human species learn vicariously in similar ways. The mechanisms of direct basic fear learning seem to be conserved across species. How many other species have also developed (and to what extent) the ability to learn through observation as well as experience? Finally, there is the question of how we learn which things to learn through observation -- to sense when it is better, for instance, to let someone else test the rope swing. How we learn to choose between direct and observational learning is another excellent question for future research. This rich recent work by Olsson and Phelps provides an excellent start towards getting there. Kevin Ochsner is an assistant professor of psychology at Columbia University, where he investigates the psychological and neural processes involved in emotion, pain, self-regulation, self perception, and person perception.