48: Emotions and Feelings

• The Modern Search for the Emotional Brain Began in the Late 19th Century

• The Amygdala Emerged as a Critical Regulatory Site in Circuits of Emotions

  • Studies of Avoidance Conditioning First Implicated the Amygdala in Fear Responses

  • Pavlovian Conditioning Is Used Extensively to Study the Contribution of the Amygdala to Learned Fear

  • The Amygdala Has Been Implicated in Unconditioned (Innate) Fear in Animals

  • The Amygdala Is Also Important for Fear in Humans

  • The Amygdala Is Involved in Positive Emotions in Animals and Humans

• Other Brain Areas Contribute to Emotional Processing

• The Neural Correlates of Feeling Are Beginning to Be Understood

• An Overall View

Elation, compassion, sadness, fear, and anger are examples of emotions. These states have an enormous impact on our behavior. But what exactly is an emotion? Unfortunately, the term emotion is commonly and confusingly used in two ways. Sometimes it refers to physiological responses to certain kinds of stimuli; when in danger, your muscles tense and your heart pounds, and you may also feel afraid. But it also refers to conscious experiences, called feelings, that often (but not always) accompany these bodily responses. We need to consistently distinguish between these two states.

In this chapter we use the term emotion to refer to the first of the two states: The set of physiological responses that occur more or less unconsciously when the brain detects certain challenging situations. These automatic physiological responses occur within both the brain and the body proper. In the brain they involve changes in arousal levels and in cognitive functions such as attention, memory processing, and decision strategy. In the body proper they involve endocrine, autonomic, and musculoskeletal responses (see Chapter 47). We use the term feeling to refer to the conscious experience of these somatic and cognitive changes. In a certain sense feelings are accounts our brain creates to represent the physiological phenomena generated by the emotional state.

In sum, emotions are automatic, largely unconscious behavioral and cognitive responses triggered when the brain detects a positively or negatively charged significant stimulus. Feelings are the conscious perceptions of emotional responses.

Emotional reactions have been conserved throughout the evolution of species. Behavioral responses that we typically call emotional responses are found in very simple organisms that may not have consciousness and thus not have feelings. A bacterial cell can detect harmful and useful chemicals and respond to these in adaptive ways. Indeed, all organisms must have such capacities to survive and thrive.

Some stimuli—objects, animals, or situations—trigger emotions automatically, even in the absence of experience. These stimuli are said to have emotional competence. In addition, some otherwise insignificant objects and events that occur in conjunction with emotionally competent stimuli can acquire emotional significance through associative learning. Thus, whereas emotionally competent stimuli are naturally significant (eg, painful or delicious), other objects and events acquire emotional competence by their association with emotionally competent stimuli.

When the brain detects emotionally competent stimuli, it sends commands to networks that control the endocrine glands, the autonomic motor system, and the musculoskeletal system (Figure 48–1). The endocrine system is responsible for the secretion and regulation of hormones into the bloodstream that affect bodily tissues and the brain. The autonomic system mediates changes in the physiological control systems of the body, including the cardiovascular system and the visceral organs and tissues in the body cavity (see Chapter 47). The skeletal motor system mediates overt behaviors such as freezing, fight-or-flight, and particular facial expressions. Together these three systems control the physiological expression of emotional states.

The autonomic and endocrine changes involved in emotional states are part of the body's homeostatic regulatory mechanisms, which are engaged whenever the body is confronted by an intrinsically charged stimulus. In fact, the body's response to strong emotion is not that different from its response to changes in other drive states or alterations in other bodily regulatory processes such as hunger, thirst, sex, and sleep, or its response to pain or changes in body metabolism that occur during vigorous exercise. These regulatory mechanisms are mediated mostly by subcortical structures—the amygdala, striatum, hypothalamus, and brain stem (see Chapter 47).

Most emotional states are observable either directly (for example, in facial expressions or other overt behaviors) or indirectly with psychophysiological or neurophysiological tests or endocrine assays. Thus many emotional responses are measurable and their neurobiological underpinnings can be investigated objectively in both human beings and experimental animals. On the other hand, measuring subjective feelings is more of a challenge and is only practical in humans.

We begin the chapter with a discussion of the historical antecedents of modern neural science research on emotion. We then describe the neural circuits and cellular mechanisms that underlie the most thoroughly studied emotion, fear. Finally, we consider how the brain processes complex social emotions and conscious feelings.

The modern attempt to understand emotions began in 1890 when William James, the founder of American psychology, asked: What is the nature of fear? Do we run from the bear because we are afraid, or are we afraid because we run? James proposed that the conscious feeling of fear is a consequence of emotions, of the bodily changes that occur during the act of running away—we are afraid because we run.

According to James, each feeling (fear, joy, anger) results from its own unique pattern of emotional expression, or bodily signature, controlled by descending connections from the cerebral cortex. A feeling comes about when the bodily expression of that emotional response enters consciousness. James's peripheral feedback theory drew on the knowledge of the brain at the time, namely, that the cortex had areas devoted to movement and sensation (Figure 48–2). Little was known about specific areas of the brain responsible for emotion and feeling.

At the turn of the 20th century researchers found that animals were still capable of emotional responses after the total removal of the cerebral hemispheres, suggesting that some aspects of emotion are mediated by subcortical regions. The fact that electrical stimulation of the hypothalamus could elicit autonomic responses similar to those that occur as emotional responses in the intact animal suggested to Walter B. Cannon that the hypothalamus might be a key region in the control of fight-or-flight responses and other emotions.

In the 1920s Cannon showed that transection of the brain above the level of the hypothalamus (by means of a cut that separates the cortex and thalamus from the hypothalamus and lower brain areas) left an animal that was still capable of showing rage. But a transection below the hypothalamus, leaving only the brain stem and spinal cord, eliminated the coordinated reactions of natural rage. This clearly implicated the hypothalamus in emotional reactions. Cannon called these hypothalamically mediated reactions "sham rage" as they lacked input from cortical areas, which he assumed were critical for the emotional experience (Figure 48–3).

Cannon and his student Phillip Bard proposed an influential theory of emotion centered on the hypothalamus and thalamus. According to their theory, sensory information processed in the thalamus is sent both to the hypothalamus and the cerebral cortex. The projections to the hypothalamus produce emotional responses (through connections to the brain stem and spinal cord) while the projections to the cerebral cortex produce conscious feelings (Figure 48–2). This theory implied that the hypothalamus is responsible for the brain's evaluation of the emotional significance of external stimuli and that emotional reactions depend on this appraisal.

In 1937 James Papez extended the Cannon-Bard theory. Like Cannon and Bard, Papez proposed that sensory information from the thalamus is sent to the hypothalamus. From there, descending connections to the brain stem and spinal cord give rise to emotional responses and ascending connections to the cerebral cortex give rise to feelings. But Papez expanded the neural circuitry of feelings considerably beyond the Cannon-Bard theory by interposing a new set of structures between the hypothalamus and the cerebral cortex. He argued that signals from the hypothalamus go first to the anterior thalamus and then to the cingulate cortex, where signals from the hypothalamus and sensory cortex converge. This convergence accounts for the conscious experience of feeling. The sensory cortex then projects to both the cingulate cortex and the hippocampus, which in turn makes connections with the mammillary bodies of the hypothalamus, thus completing the loop (Figure 48–2).

In the late 1930s Henrich Klüver and Paul Bucy removed the temporal lobes of monkeys bilaterally and found a variety of psychological disturbances, including alterations in feeding habits (the monkeys put inedible objects in their mouth) and sexual behavior (they attempted to have sex with inappropriate partners, like members of other species). In addition, the monkeys had a striking lack of concern for previously feared objects (eg, humans and snakes). This remarkable set of findings came to be known as the Klüver-Bucy syndrome.

Building on the Cannon-Bard and Papez models, and the findings of Klüver and Bucy, Paul MacLean suggested in 1950 that emotion is the product of the "visceral brain." The visceral brain included the various cortical areas that had long been referred to as the limbic lobe, so named by Paul Broca because these areas form a rim (Latin limbus) in the medial wall of the hemispheres. Later the visceral brain was renamed the limbic system. The limbic system includes the various cortical areas that make up Broca's limbic lobe (especially medial areas of the temporal and frontal lobes) and the subcortical regions connected with these cortical areas, such as the amgydala and hypothalamus (Figure 48–4).

MacLean intended his theory to be an elaboration of Papez's ideas. Indeed, many areas of MacLean's limbic system are parts of the Papez circuit. However, MacLean did not share Papez's idea that the cingulate cortex was the seat of feelings. Instead he thought of the hippocampus as the part of the brain where the external world (represented in sensory regions of the lateral cortex) converged with the internal world (represented in the medial cortex and hypothalamus), allowing internal signals to give emotional weight to external stimuli and thereby giving rise to conscious feelings. For MacLean the hippocampus was involved both in the expression of emotional responses in the body and in the conscious experience of feelings.

Subsequent findings raised problems for MacLean's limbic system theory. In 1957 it was found that damage to the hippocampus, the keystone of the limbic system, produced deficits in converting short- to long-term memory, a distinctly cognitive function. In addition, animals with damage to the hippocampus are able to express emotions, and humans with hippocampal lesions express and feel emotions normally. In general, damage to areas of the limbic system did not have the expected effects on emotional behavior.

Several of MacLean's other ideas on emotion are nevertheless still relevant. MacLean thought that emotional responses are essential for survival and therefore involve relatively primitive circuits that have been conserved in evolution, and this notion is key to an evolutionary perspective of emotion. Further, his idea that emotional states and cognitive processes involve somewhat distinct circuits and can function relatively independent of one another, as implied by Cannon and Bard and all subsequent theories of the emotional brain, also has some merit.

Although damage to most limbic areas does not have the effects on emotional behavior predicted by the limbic system theory, one limbic area was consistently shown to be involved in emotion. This area is the amygdala.

Studies of Avoidance Conditioning First Implicated the Amygdala in Fear Responses

In the mid-1950s Lawrence Weiskrantz sought to understand which region of the temporal lobe was responsible for the emotional changes characteristic of the Klüver-Bucy syndrome. To do this, he used avoidance conditioning, a form of instrumental conditioning.

In avoidance conditioning an animal learns to perform responses that successfully avoid an aversive shock, the unconditioned stimulus (US). Successful avoidance of shock reinforces the response, ie, it increases the probability of the response. Normal monkeys learn instrumental responses (ie, pressing a lever) to avoid the shock, but monkeys with lesions of the amygdala do not. Weiskrantz concluded that a key function of the amgydala was to connect external stimuli with their aversive (punishing) or rewarding consequences.

Fear has been a popular emotion in neuroscience research because it is so important for survival and also because excellent behavioral protocols are available for studying fear in animals. Following Weiskrantz's discovery, many researchers used avoidance conditioning to study the neural mechanisms of fear. However fear can also be studied using Pavlovian conditioning and by the early 1980s had become the preferred protocol.

Pavlovian Conditioning Is Used Extensively to Study the Contribution of the Amygdala to Learned Fear

In Pavlovian fear conditioning an association is learned between the US (eg, shock) and the conditioned stimuli (CS) that predict the US. For example, an emotionally neutral CS (a tone) is presented for several seconds and the animal is shocked during the final second of the CS. After several pairings of tone and shock, presentation of the tone alone elicits defensive freezing and associated changes in autonomic and endocrine activity. In addition, many defensive reflexes, such as eyeblink and startle, are facilitated by the tone alone.

Pavlovian fear conditioning is actually the first phase of avoidance conditioning. The pairing of US and CS initially results in the conditioning of a response, but in the second phase the animal learns to perform an instrumental response to avoid the shock. By the early 1980s neuroscientists began to realize that a more efficient way to study fear learning is to focus on the first stage of avoidance conditioning—Pavlovian fear conditioning—and not extend the experimental design to the second phase.

Research carried out in a variety of laboratories established that lesions of the amygdala prevent Pavlovian fear conditioning from occurring. Animals with amygdala damage fail to learn the association between the CS and the US and thus do not express fear when the CS is later presented alone.

The amygdala consists of approximately 12 nuclei, but the lateral and central nuclei are especially important in fear conditioning (Figure 48–5). Damage to either nucleus, but not other regions, prevents fear conditioning. The lateral nucleus is the input nucleus receiving information about the CS (eg, a tone) from the thalamus. The central nucleus is the output region; neurons here project to brain stem areas involved in the control of defensive behaviors and associated autonomic and humoral responses (see Chapter 47). The lateral and central nuclei are connected by way of several intra-amygdala circuits, including connections in the basal and intercalated nuclei.

Sensory inputs reach the lateral nucleus from the thalamus both directly and indirectly. Much as predicted by the Cannon-Bard hypothesis, sensory signals from thalamic relay nuclei are conveyed to sensory areas of cortex. As a result the amygdala and cortex are activated simultaneously. However, the amygdala is able to respond to a danger cue before the cortex can process the stimulus information. Given that cortical processing is required to consciously experience fear, the emotional state triggered by thalamic inputs is likely to be initiated before we consciously feel fear.

The lateral nucleus is thought to be a site of synaptic change during fear conditioning. The CS and US signals converge in the lateral nucleus; when the CS and US are paired, the effectiveness of the CS is enhanced (Figure 48–6). The lateral nucleus appears to be functionally divided. Neurons in the most dorsal part of the dorsal division appear to initiate learning when the CS and US are paired, whereas neurons in an adjacent ventral part of the dorsal division are thought to mediate the long-term memory of the CS–US association. Recent studies have also shown that synaptic plasticity occurs in specific central amygdala circuits. The central amygdala thus does not simply drive motor outputs but is also part of the circuitry through which fear associations are formed and stored, very likely by transmitting information about the CS and US from the lateral nucleus.

The emotional charge of a stimulus is evaluated by the amygdala to determine whether danger is present. If the amygdala detects danger, it orchestrates the expression of behavioral and physiological responses by way of connections to the hypothalamus and brain stem. For example, freezing behavior is mediated by connections from the central nucleus to the ventral periaqueductal gray region. But in addition, the amygdala has a variety of connections that allow it to also influence other cognitive functions. For example, through its widespread projections to cortical areas it can modulate attention, perception, memory, and decision-making. Its connections with the modulatory dopaminergic, noradrenergic, serotonergic, and cholinergic nuclei that project to cortical areas also influence cognitive processing (see Chapter 46).

The cellular and molecular mechanisms within the amygdala that underlie learned fear, especially in the lateral nucleus, have been elucidated in great detail (see Chapter 66). The findings support the view that the lateral nucleus is a site of memory storage in fear conditioning.

The Amygdala Has Been Implicated in Unconditioned (Innate) Fear in Animals

Many animals rely on innate (unconditioned) olfactory signals in the detection of threats, mates, food, and so forth. For example, rodents exhibit freezing and other defensive behaviors when fox urine is detected.

Recent studies have made considerable progress in uncovering the circuits underlying innate fear (see Chapter 47). In mammals unconditioned threats involving predator or conspecific odors are transmitted from the vermonasal component of the olfactory system (see Chapter 32) to medial amygdala. Outputs of the medial amygdala reach the ventromedial hypothalamus, which connects with the premammillary hypothalamic nucleus. In contrast to learned fear, which depends on the ventral periaqueductal gray region, unconditioned fear responses depend on connections from the hypothalamus to the dorsal peri-aqueductal gray region.

The Amygdala Is Also Important for Fear in Humans

The basic findings in animals regarding the role of the amygdala in emotion have been confirmed in studies of humans. Thus patients with damage to the amygdala fail to undergo fear conditioning when presented with a neutral CS paired with a US (electric shock or loud noise). Patients with damage to the amygdala also fail to recognize facial expressions of fear and do not generate autonomic fear responses to these.

In normal human subjects activity in the amygdala increases during CS–US pairing (Figure 48–7). This activity is especially strong when the stimuli are presented subliminally. In normal subjects fearful facial expressions also activate the amygdala, even when presented subliminally. These findings emphasize the importance of the amygdala as a subconscious evaluator of the meaning of a stimulus.

Certain forms of fear processing are unique to humans. For example, simply telling a human subject that the CS may be followed by a shock is enough to allow the CS to elicit fear responses. The CS elicits characteristic autonomic responses even though it was never associated with the delivery of the shock. Humans can also be conditioned by allowing them to observe someone else being conditioned—the observer learns to fear the CS even though the CS or US were never directly presented to the observing subject.

The emotional learning and memory capacities of the human amygdala fall into the category of implicit learning and memory (unconscious recall of perceptual and motor skills (see Chapter 65)). In situations of danger, however, the hippocampus and other components of the medial temporal lobe system concerned with explicit learning and memory (the conscious recall of people, places, and things) will encode the learning such that learned indicators of danger can also be recalled consciously.

Studies of patients with bilateral damage to the amygdala or hippocampus illustrate the separate contributions of the amygdala and the hippocampus to implicit and explicit memory, respectively. Patients with damage to the amygdala show no physiological responses to a CS (indicating no implicit learning) but have good memory of the conditioning experience (indicating explicit learning), whereas patients with hippocampal damage respond normally to the CS but have no conscious memory of the conditioning experience.

Amygdala function is altered in a number of psychiatric disorders in humans, especially disorders of fear and anxiety (see Chapter 63). In addition, the amygdala plays an important role in processing cues related to addictive drugs (see Chapter 49).

The Amygdala Is Involved in Positive Emotions in Animals and Humans

Although most work on the neural basis of emotion during the last half century has focused on aversive responses, especially fear, other studies have shown that the amygdala is also involved in positive emotions, in particular the processing of rewards. In monkeys and rats the amygdala is required for associating neutral stimuli with rewards.

Studies in nonhuman primates and rodents have followed up on Weiskrantz's suggestion that the amygdala connects stimulus rewards as well as punishers. For example, in a recent study monkeys were trained to associate abstract visual images with a reward or punisher. The meaning was then reversed (eg, by pairing a punisher with a stimulus that had previously been associated with a reward). In this way it was possible to separate the contributions of the amygdala to visual and value processing. Changes in the value of the images modulated neural activity in the amygdala, and the modulation occurred rapidly enough to account for behavioral learning.

A growing number of functional imaging studies of humans have also shown that the amygdala is involved in emotions. For example, the human amgydala is activated when subjects observe pictures of stimuli associated with food, sex, and money, or when people make decisions based on the reward value of stimuli.

In addition to the amygdala, other brain areas make important contributions to emotional processing As seen in the case of conditioned and unconditioned fear, the amygdala contributes to emotional processing as part of a larger circuit that includes regions of the hypothalamus and brain stem, eg, the periaqueductal gray region in the brain stem.

Cortical areas are also important. A number of studies in humans have implicated the ventral region of the anterior cingulate cortex, the insular cortex, and the ventromedial prefrontal cortex in various aspects of emotional processing. These cortical circuits are especially important in complex emotional states.

Complex feelings are associated with social interaction, and range from empathy and pride to embarrassment and guilt. As in the primary emotions such as fear, pleasure, or sadness, social emotions consist of specific bodily changes and behaviors and are experienced consciously as distinct feelings. These feelings make important contributions to normal social interactions.

Studies of patients with neurological disease and focal brain lesions have advanced the understanding of the neural basis of social emotions. For example, damage to some sectors of the prefrontal cortex markedly impairs social emotions and related feelings. In addition, these patients show marked changes in social behavior that resemble the behavior of patients with developmental sociopathic personalities. Patients with damage to some sectors of the prefrontal cortex are unable to hold jobs, cannot maintain stable social relationships, are prone to violate social conventions, and cannot maintain financial independence. It is common for family ties and friendships to break after the onset of this condition. Recent studies reveal that, under controlled experimental conditions, the moral judgments of these patients can be flawed.

Unlike patients with parietal or parieto-frontal damage, patients with frontal damage do not have motor defects such as limb paralysis or speech defects and thus may appear at first to be neurologically normal. Their perceptual abilities, attention, learning, recall, language, and motor abilities do not show signs of disturbance. Some patients have IQ scores in the superior range. For these reasons they return to their work and social activities after their initial recovery from brain damage. Only when they start to interact with others are their defects noticed.

In the prefrontal cortex the ventromedial sector is particularly important. In most patients with impaired social emotions this sector is damaged bilaterally (Figure 48–8), although damage restricted to the right side is sufficient to cause antisocial symptoms. The critical region encompasses Brodmann's areas 12, 11, 10, 25, and 32, which receive extensive projections from the dorsolateral and dorsomedial sectors of the prefrontal cortex. Some of these areas project extensively to subcortical areas related to emotions: the amygdala, hypothalamus, and periaqueductal gray region in the brain stem tegmentum.

Patients with these frontal lesions do not exhibit changes in heart rate or degree of palm sweating when shown stimuli that normally cause emotions, although they can describe flawlessly the pictures. Normal subjects are equally proficient at describing the pictures but also have psychophysiological responses to the pictures. Unlike normal subjects, patients with frontal lesions do not show detectable skin conductance changes, a sign of sympathetic activation, during the period that precedes making risky and disadvantageous decisions, suggesting that their emotional memory is not engaged during that critical period. Also unlike normal subjects, these patients fail in tasks in which they have to make a decision under conditions of uncertainty, and in which reward and punishment are important factors.

Interestingly, when asked about punishment, reward, or responsibility, adult patients with prefrontal damage respond as if they still have the basic knowledge of the rules, but their actions indicate that they fail to use them in real situations. This dissociation suggests that their behavioral defects are not caused by a loss of factual knowledge but rather by a difficulty in accessing that knowledge, perhaps because of defective emotional processing.

Functional imaging of normal human subjects shows that the ventromedial frontal cortex is activated during the period before a decision. That same region is activated by tasks involving punishment and reward, supporting the notion that the emotional significance of punishment and reward is relevant for decision making. Punishment and reward are frequently featured in experiments involving economic and moral decisions.

The prefrontal cortex, especially areas in the ventromedial sector, operates in parallel with the amygdala. During an emotional response ventromedial areas govern the attention accorded to certain stimuli, influence the content retrieved from memory, and help shape mental plans conceived as a response to the triggering stimulus. By influencing attention, both the amygdala and the ventromedial prefrontal cortex are also likely to alter cognitive processes, for example by speeding up or slowing down the flow of sensory representations (see Chapter 17). All of these changes are eventually incorporated in the circuitry of working memory in the dorsolateral prefrontal cortex, which contributes to the processing of conscious feelings.

We have defined feeling as the conscious experience of an emotion. Thus attempts to study the neural correlates of feelings in experimental animals are difficult because feelings are inherently subjective. Evidence for the neural correlates of feeling comes primarily from functional imaging studies of humans and from neuropsychological testing of patients with specific brain lesions.

One functional imaging study used positron emission tomography (PET) to test the idea that feelings are correlated with activity in cortical and subcortical somatosensory regions that specifically receive inputs related to the internal environment—the viscera, the endocrine glands, and the musculoskeletal system. Normal subjects were asked to recall personal episodes involving four different emotions—sadness, happiness, anger, and fear—and to attempt to reexperience as closely as possible the emotion that accompanied those events. From the moment a subject was told which episode or emotion to reexperience until the end of the scanning, the activity level in a number of cortical and subcortical regions was analyzed continuously, along with several psychophysiological parameters, such as skin conductance.

Activity changed in the insular cortex, secondary somatosensory cortex (S-II), cingulate cortex, hypothalamus, and upper brain stem, and the pattern of change differed with each emotion (Figure 48–9). Patterns of activation did not overlap. These results support the idea that at least a part of the neural substrate for feelings corresponds to the changes in the pattern of activity caused by the emotion being elicited.

When, for example, subjects experienced sadness, the subgenual sector of the cingulate cortex was activated. This region is of special interest because it is also differentially activated in patients with bipolar depression. Moreover, this region appears thinned in structural MRI scans of patients with chronic depression (Figure 48–10). The amygdala is not activated during conscious feeling, evidence that it deals mainly with unconscious emotional states.

In humans, sectors of the insular cortex that are activated during recall of feelings are also activated during the conscious sensation of pain and temperature. The insular cortex receives homeostatic information (temperature and pain signals, changes in blood pH, carbon dioxide, and oxygen) through pathways that originate in peripheral nerve fibers. These afferent fibers include, for example, the C and Aδ fibers. Those fibers form synapses with neurons in lamina I of the spinal cord's posterior horn or the pars caudalis of the trigeminal nerve nucleus in the brain stem. The pathways from lamina I and the trigeminal nucleus project to brain stem nuclei (nucleus of the solitary tract and parabrachial nucleus); from there to the thalamus and then to the insular cortex. The identification of this functional system is further support for the idea that signals in the afferent somatosensory pathways play a role in the processing of feelings. Moreover, in patients with pure autonomic failure, a disease in which visceral afferent information is severely compromised, functional imaging studies reveal a blunting of emotional processes and attenuation of activity in the somatosensory areas that contribute to feelings.

Like other feelings, social feelings engage the insular cortices and the primary and secondary somatosensory cortices (S-I and S-II), as has been shown in functional neuroimaging experiments evaluating empathy for pain and, separately, admiration and compassion.

Insights about the neural correlates of feeling also have come from examination of patients with focal lesions. Damage to the right somatosensory cortex (S-II, S-I, and insula) impairs social feelings such as empathy. Consistent with this finding, patients with lesions in the right somatosensory cortex fail to guess accurately the feelings behind the facial expressions of other individuals. This ability to read faces is not impaired in patients with comparable lesions of the left somatosensory cortex, indicating that the right cerebral hemisphere is dominant in the processing of at least some feelings. Body feelings such as pain and itch remain intact as do feelings of basic emotions such as fear, joy, and sadness.

On the other hand, damage to the human insular cortex, especially on the left, can suspend addictive behaviors, such as smoking. This suggests that the insular cortices play a role in associating external cues with internal states such as pleasure and desire. Interestingly, complete bilateral damage to the human insular cortices, as caused by herpes simplex encephalitis, does not preclude emotional feelings or body feelings, suggesting that the somatosensory cortices and subcortical nuclei in the hypothalamus and brainstem are also involved in generating feeling states.

Finally, the neural basis for the hedonic component or pleasure aspect of feeling states is being elucidated in animal studies by Kent Berridge and his colleagues. These studies consistently implicate the nucleus accumbens and other nuclei of the basal ganglia, especially in the ventral striatum and ventral pallidum. A growing number of functional imaging studies of humans—especially in the field of neuroeconomics, which combines research methods from neuroscience, experimental and behavioral economics, and cognitive and social psychology to explore decision-making in humans—point in the same direction.

In the overall physiology of life regulation emotional states sit between the simple processes of reflexes and homeostatic regulation, on the one hand, and cognitive processes on the other.

Emotions serve as cues to appropriate behavior in response to challenges and opportunities in the environment, allowing an organism to deploy specific advantageous behaviors rapidly. The neural processing responsible for emotional states and their effect on behavior is largely unconscious, much as Freud had predicted. Indeed, functional imaging studies show that the amygdala is readily activated by stimuli that are prevented from entering awareness. But even though the initial neural processing of emotionally competent stimuli may be unconscious, that process can lead to feelings, the conscious awareness of the brain's response to the stimuli.

What might the adaptive role of feelings be? Given that emotional states can arise automatically and effectively, what would be the advantage of bringing to consciousness the physiological changes that constitute emotions? The observation of brain-damaged patients in whom feelings are severely compromised gives one possible answer: The loss or impairment of the neural processes responsible for feelings diminishes the ability to anticipate and plan behavior.

Conscious feelings facilitate learning about objects and situations that cause emotional responses. Thus feelings enhance the behavioral significance of emotions and orient the imaginative process necessary for planning of future actions. In brief, unconscious emotional states are automatic signals of danger and advantage, whereas conscious feelings, by recruiting cognitive abilities, give us greater adaptability in responding to dangerous and advantageous situations. Indeed, both emotions and feelings also play a major role in social behavior, including the formation of moral judgments and the framing of economic decisions.