Human amygdala activation tracked by functional imaging studies

The general problems of human lesion data are small sample sizes, variability in the lesion size and precise location, as well as the difficulty that compensatory functions may be accomplished by other brain areas. Nevertheless, lesion studies are one unique method of investigating brain functions but they should be complemented by imaging studies with larger sample sizes of homogenuous groups of healthy subjects.

Generally three different kinds of paradigms have been applied in human imaging studies to investigate amygdala function: stimulation with affective material, classical fear conditioning and extinction, and affective startle modulation. Given the relatively small volume of the amygdala and its localization within deeper structures

of the brain, fMRI methodology with its high spatial resolution is preferrably used compared to PET or electro-/magnetoencephalographic techniques. However, one has to be aware of the technical limitations of the fMRI-technique when trying to measure amygdala activity. Attention has to be drawn to the unavoidable presence of susceptibility-induced magnetic field inhomogenities in the proximity of the amygdala.

The amygdala is neighboring the air-filled bony cavities at the skull base. These have different magnetic susceptibilities than brain tissue. Small stimulus-correlated head motions for example, that are particularly seen in psychiatric patients, are likely to result in artifactual “amygdala activations”. A reasonably reliable mapping of amygdala activity is only possible when coronal acquisitions and voxel sizes of 4-8 µl or less are employed (for a more detailed discussion of the methodological problems see Merboldt et al., 2001).

Vuilleumier et al. (2001) used event-related fMRI to determine whether the amygdala is selectively activated by fearful as opposed to neutral facial expressions and in particular whether attention modulates the amygdala response. A matching task required the subjects to decide if one of two simultaneously presented stimulus pairs were the same. Stimulus pairs were either neutral pictures of houses, or pairs of fearful or neutral facial expressions. In addition, the location of the target pair was varied within in a four-field spatial arrangement of four empty frames in which the pictures appeared. The target pair could be arranged horizontally or vertically.

Independently of the location of the target pair, either the two horizontal or the two vertical frames where highlighted in advance to the picture presentation. Subjects were instructed to attend only to the stimulus pair in the highlighted frames. This design was applied to control for attention. The question was whether processing of fearful faces is dependent on focussed attention to the stimulus or if it can also occur when attention is directed elsewhere. The authors found a preferential left amygdala activation when fearful faces were shown as opposed to neutral ones. In addition, this effect was independent of whether attention was directed to fearful targets or to distractor pictures. These results show the selective activation pattern of the amygdala in the processing of fearful stimuli that is independent of focussed attention, thereby stressing the evolutionary significance of the ‘fear system’. It is important for an organism to be equipped with an automatic danger detection system that notices any potentially harmful stimuli in its environment even when attention is captured by other less salient cues.

In a life-long process of continuous adaptation, learning processes are vital.

One simple form of learning is conditioning. Involvement of the amygdala in conditioned fear acquisition and extinction in healthy humans was demonstrated by LaBar et al. (1998) in an echoplanar fMRI study. The authors further found a habituation of amygdala responses across trials in both experimental phases. Activity during acquisition and extinction indicates the involvement in learning and storing associated stimulus-punishment-contingencies. Habituation of neural responses within and across trials may preserves the neuron’s capacity to boost its firing rate again when a novel threatening stimulus appears. Neurons that rapidly habituate are concerned with the detection of novel or changing patterns of stimulation. Threat and novelty are linked concepts in the way that novelty (as well as change) are important triggers of fear (Gray, 1987).

Although it is without doubt that the amygdala responds particularly to threatening stimuli, it has already been noted above that some authors found evidence for a more general function of processing different kinds of emotionally salient stimuli. This means that not valence per se might be the essential modulator but the more general concept of salience. The more salient a stimulus is, the more amygdala reactivity is expected. The responsiveness to salient cues of different valence was shown in a fMRI study by Breiter et al. (1996). They also confirmed earlier findings by Bordi et al. (1992, 1993; see below) in demonstrating a rapid habituation to these stimuli. Fearful, happy and neutral faces were briefly (~200ms) presented in a counterbalanced order in a passive viewing task. Bilateral anterior amygdala activation (left > right) was measured in the fearful versus neutral condition. For happy faces a left-sided anterior amygdala activation was found.

Neutral facial expressions did not lead to significant activity increases in anterior amygdala relative to a simple fixation point on a plain background. Further analyses over time revealed within and across runs decreases of amygdala activity for fearful and happy faces indicating rapid habituation.

This habituation effect was observed earlier by Bordi et al. (1992) in the lateral amygdaloid nucleus of rats in response to repeated presentations of loud aversive bursts of white noise. A large amount of cells fired only in response to the first two to five noise bursts and remained unresponsive afterwards. Even when long interstimulus intervals of several minutes separated the repeated noise bursts, habituation was observed (Bordi, 1993).

In line with Breiter (1996), Garavan et al. (2001) found selective activity in the amygdala for both positively and negatively valenced stimuli as opposed to neutral, supporting the view of processing the emotional significance of stimuli in general.

However is was interesting to see that the arousal level modulated the amygdala response for negative, but not for positive stimuli.

In conclusion, the above imaging studies demonstrate selective activation of the amygdala when a subject is confronted with salient emotional stimuli. This is particularly true for fear relevant material, but an increasing number of studies found similar activation patterns also for appetitive cues. Therefore the amygdala can be regarded as a ‘salience processor’.