Instructed knowledge shapes feedback-driven aversive learning in striatum and orbitofrontal cortex, but not the amygdala

摘要

Around the start of the twentieth century, Pavlov discovered that dogs salivate upon hearing a bell that has previously signaled that food is available. This phenomenon, in which a neutral stimulus (the bell) becomes associated with a particular outcome (such as food), is known as classical conditioning. The network of brain regions that supports this process – which includes the striatum, the amygdala and the prefrontal cortex – seems to work in a similar way across most animal species, including humans. However, humans don’t learn only through experience or trial-and-error. We do not need to burn our hands to learn not to touch a hot stove a verbal warning from others is usually sufficient. Experiments have shown that giving people verbal instructions on how to obtain rewards alters the activity of the striatum and prefrontal cortex. That is, the instructions interact with the circuit that also supports learning through experience. But is this the case for learning how to avoid punishments? That process depends largely on the amygdala, and it is possible that systems designed to detect threats may be less sensitive to verbal warnings. To address this question, Atlas et al. taught people to associate one image with a mild electric shock, and another with the absence of a shock. After a number of trials, the relationships were reversed so that the previously neutral picture now predicted a shock and vice versa. Telling the participants about the reversal in advance triggered changes in the activity of the striatum and part of the prefrontal cortex. By contrast, such warnings had no effect on the amygdala. Instead, the activity of the amygdala changed only after the volunteers had experienced for themselves the new relationship between the pictures and the shocks. A key next step is to find out whether this distinction between the two types of learning signals (those that can be updated by instructions and those that cannot) is specific to humans. While the current study relied upon language, there are other methods that could be used to explore this issue in animals. Furthermore, knowing that the human brain has a specialized threat detection system that is less sensitive to instructions could help us to understand and treat anxiety disorders. Atlas et al. hope to test this possibility directly in the future.