Figure 1 Nonsocial and social fear learning in humans. An individual learns to fear a CS through its pairing with (a) an electric shock to the wrist (fear conditioning), (b) a learning model’s expression of distress (observational fear learning), and (c) verbal information about its aversive qualities (instructed fear).
Another behavioral output, avoidance behavior, is mediated by input to the basal ganglia from the basal nucleus13.
Under most circumstances, the role of the amygdala in fear con- ditioning is best understood together with other functional regions within a greater circuitry of fear learning. This circuitry involves sensory input and motor output systems, as well as regions that contribute to explicit and conscious aspects of learning and expression of fear. For example, the hippocampus, another medial temporal lobe structure adjacent to the amygdala, is critical for coding contextual information about the fear learning situation, such as relationships between different features and the timing of events. In other words, whereas the amygdala is responsible for forming associations between somatosensory states and representations of individual stimuli (cue learning), the hippocampus is important for encoding relations between the various cues that comprise the learning context (con- textual learning). Patients with bilateral and unilateral amygdala lesions can verbally report the CS-US contingency, although they lack the normally associated autonomic response14, leading to the suggestion that the amygdala is critically involved only in implicit, nonverbal processes underlying acquisition and expression of conditioned fear. In contrast, the hippocampus is essential for consolidation and retention of explicit or declarative memory of the CS-US contingency15 and the environmental contexts that regulate conditioned fear responses16. In addition, across species, the prefrontal cortex (PFC) has a unique role in top-down regulation of affective responses through its regulation of activation in subcortical regions, such as the amygdala17,18. More specifically, the ventral (infralimbic) region of the medial prefrontal cortex (MPFC) is critical to retention of extinction of conditioned fear responses in rats19, and the human homolog of this region is involved in extinction in humans20.
The demonstration that the amygdala can operate independently from other neural systems critical to explicit expression of learned fear provides a possible explanation for the observation that a conditioned fear response can be elicited without explicit awareness of the CS21,22. A subliminal presentation of CS results in activation of the right amygdala23. Conditioned responses to subliminally presented CSs are only reported when the CSs are drawn from naturally fear-relevant stimulus categories, such as snakes, spiders and angry faces. Fear responses conditioned to fear-relevant stimuli are more resistant to modification by extinction and verbal instructions than are responses to fear-irrelevant natural categories, such as butterflies, happy faces or fear-relevant artifacts, such as broken electrical outlets and guns22,24. These observations, combined with the superior fear conditioning observed in nonhuman animals to certain types of ecologically relevant stimuli, has led researchers to posit that these particular stimuli may be prepared by evolution to engage in aversive associations. Socially and culturally defined categories can also act as prepared stimuli in a fear conditioning protocol25.
Just as the role of the amygdala in fear learning cannot be fully understood without recognizing the role of other regions in the same fear learning circuit, this kind of learning cannot be completely understood without considering the intricacy of the natural environ- ment in which it occurs. For example, fear conditioning procedures have traditionally examined learning involving direct, individual experience of an aversive stimulus, the US. However, the natural milieu of many species offers both safer and more economical alternative means to attain corresponding information about potentially noxious stimuli. The social environment provides a suitable medium to transfer emotionally significant information between individuals. Verbally communicating with a fellow human or observing a conspecific’s expressions of fear are two such means that can produce learning that shares both behavioral and neural qualities with fear acquired through fear conditioning (Figs. 3 and 4).
Observational fear learning across species Social transmission and detection of fear signals is well documented in a range of species26. The ability to detect and respond appropriately to signs of fear and pain in a conspecific probably has conferred a significant selective advantage during evolution. However, these signs not only alert the receiver about potential imminent danger, they also assign a threat value to the context or cue associated with the threat. For example, a conspecific’s fear expression may serve as an US, eliciting an immediate aversive response in the observer that becomes associated with the paired stimuli. Observational learning may also be subserved by social inference, in which the conspecific’s fear expression is a CS that was previously associated with a directly experienced aversive event (US) and may act as a secondary reinforcer in future learning.
The study of fear learning through social observation is informed by different lines of research, from emotional contagion and imitation to more complex operant tasks. Here we focus on social learning, as defined by processes contributing to formation of associations between different stimuli and expressed later in the absence of the conspecific serving as the learning model. We do not discuss simpler forms of socially facilitated and contagious fear responses, such as those seen in flocks behaving in unison, schools and herds of animals27,28 or imitation29–31. To provide an appropriate parallel to existing research on social fear learning in humans, we focus our discussion of the animal literature on social learning in the visual domain. However, similar associative mechanisms are likely to be involved in social learning relying on other modalities, such as auditory and olfactory information32.
SEPTEMBER 2007 NATURE NEUROSCIENCE