The Pavlovian action-selection system consists of response sets evolved to prepare for critical species-specific behaviors such as feeding, fighting, or mating. Learning in the Pavlovian action-selection system entails recognition that one thing likely causes another, leading to taking the response set appropriate for that expected outcome. Linguistically, we categorize these response sets and identify them as emotions. In humans, many of these response sets are important to social interactions.
Although not usually thought of in terms of action-selection, emotions are an action-selection mechanism.1 Each emotion is a response set that has been learned over genetic timescales to be useful. For example, in dangerous situations, depending on emotional characterizations of the situation, one becomes afraid and one’s body prepares to flee, to stand and fight, or to submit, sometimes referred to as the fight-or-flight response (or more accurately, the fight–flight–freeze response).
There is a debate going on in the scientific community as to whether the categorization process of what we consciously label as emotions is consistent between individuals.2 These questions turn more on the issue of our conscious labels of emotions rather than the physical effect of those emotions—no one denies that anger prepares you for fighting, but other emotions (such as fear) also prepare you for fighting. Even though a given situation might make one person angry and another afraid, animals (including humans) have self-consistent response sets to consistent situations. We will return to the relationship between these response sets and emotion later in this chapter.
The advantage of having a system with these response sets is that there is a complex of body-related internal actions that need to work correctly the first time. You don’t get a chance to learn to run from the lion. You can learn that a rustle in the grass likely means that a lion is stalking you. And so you can learn to run from the rustle in the grass. In fact, you can learn that a rustle in the grass means a lion is stalking you even without being stalked. You could learn it from observing a lion stalking another animal, or you could learn it from hearing a vivid story by the fire, or some other way. In any case, what you’ve actually learned is that lions stalking prey produce a rustle in the grass. So when you hear the rustling, you realize there’s a lion, and you have the same emotional response as if a lion suddenly appeared in front of you.
What is needed is a system that can recognize the danger and implement all of the body changes quickly, all at once. That system is the Pavlovian action-selection system, in which a genetically learned action occurs in response to a given situation based on an emotional characterization of it (I like it, I will approach it. I am afraid of it, I will run away from it. I am angry at it, I will attack it. I have nowhere to run, I will freeze in fear and hope that it won’t kill me).3
In classical psychology experiments, Pavlovian learning is described as what is learned when the animal does not have to take an action in order to get the outcome—famously, a bell is rung and the food is delivered, or a light is turned on and then the subject is given a small shock.4 There is nothing the animal has to do or can do to change these events. In practice, the animal does things to prepare for the expected outcome5—the dog salivates, the rat freezes.
At one time it was assumed that the animal was learning that cues co-occur; however, we now know that Pavlovian conditioning entails recognition of informative contingencies—animals learn that when a cue is present, certain things are more likely to happen. An ever-present cue is not informative.6 Animals can even learn that certain cues predict that an event will not occur. As the lion example at the start of this chapter shows, we are learning that a cue predicts an outcome. (Pavlov’s bell worked this way as well—the bell was rung and then the dog was fed.) Stated in this way, the learning part of Pavlovian action-selection is that a stimulus (rustle in the grass, ringing a bell) predicts an outcome (a lion is stalking you, food is coming). From that prediction, you can take the appropriate action (run!, salivate).
When studied in classical psychology experiments, the action-selection part of the Pavlovian action-selection system is often seen as species-specific behavioral responses and approach or avoidance responses,7 but the largest changes are in fact somatic—changes in the body itself.8 In response to fear, for example, the heart rate increases (as does the force of each heartbeat) to move more blood to the muscles, the airways in the lungs dilate to allow more oxygen, and a host of other responses occur that prepare an animal to either fight or run. Attention increases, increasing the ability to sense stimuli.
One of the best ways to measure an animal’s fear is to measure the startle responses in a given situation. A surprising sound literally makes the animal jump; under conditions of increased anticipation of danger (fear), the animal will jump higher. By measuring the force of the muscles when the animal jumps, one can quantitatively measure its preparedness for action, and its fear.9
One does, of course, see the same effects in humans. One time, when I was first setting up my lab, I was worried about whether I could safely plug in a very expensive computer part without turning the computer off first. (The computer was in the process of running a very long computation and I didn’t want to disturb it, but I needed to access that part.) The evening before, a new Zip drive (a kind of specialized portable drive before USB flash drives were available) had caught fire and burned out at my house (with beautiful blue sparks), so I was reasonably nervous about plugging in this very expensive part. I expressed my concern to the lab, and decided that I should be the one to plug it in so that if something went wrong, it would be my fault. While I was nervously waiting to plug the part in, unbeknownst to me, one of my grad students had snuck up behind me. The instant I plugged the part in, my student whispered “bzzzzt!” in my ear. I must have jumped three feet in the air.
Another related effect in humans is that we tend to sweat when we are emotionally aroused.10 Because sweat is basically salt water, it conducts electricity, and this is measurable by an increase in skin conductance. Even very subtle changes are detectable as a change in the conductivity of the skin. These changes are known as the galvanic skin response (GSR) or the skin-conductance response (SCR) and are a surprisingly reliable measure of emotional arousal.
Neurophysiologically, these responses are controlled by small bundles of neuronsA in the base of the brain, called the hypothalamus,B and by structures at the top of the spinal cord, particularly the periaqueductal grayC and the amygdala.D
These brain structures interact with the rest of the body through a feedback loop between the hypothalamus in the brain and organs in the body via hormones circulating in the blood. The hypothalamus sends information to (and receives signals from) a structure called the pituitary gland, which is a collection of hormone-releasing and hormone-sensing cells that translate neural signals to and from hormonal signals.12 This pathway from brain to body is actually a complete loop that includes signals from the hormones to the rest of the brain as well. Neurons in the brain are cells and thus require nutrients, which they get from blood.E These hormones (such as adrenaline, testosterone, estrogen, and glucocorticoids) can dramatically affect neural function, and even neural structures.13 The brain is tightly coupled to the rest of the body through these neural–hormonal interactions.
Interestingly, one of these hormone-releasing glands that connect the brain to the body is the pineal gland, which was the specific brain structure that the philosopher René Descartes (1596–1650) identified as the connection between the body and the soul.14 Descartes argued that animals were machines, but humans are not because humans have a soul. Descartes argued that the connection between the body and the soul lay in the pineal gland. This hypothesis (that the soul and body are different things) is called dualism, and is generally phrased in the modern era in terms of a hypothesis that the mind and body are different things. Of course, as everyone who has studied anatomy knows (including Descartes, who was an exceptional anatomist, well familiar with both human and nonhuman anatomy15), all mammals have pineal glands.16 We will return to this discussion at the end of the book, when we directly address dualist and materialist philosophies.
An attentive reader may be wondering at this point in our discussion of “emotion” where exactly is the emotion? Anger, jealousy, lust, fear, happiness, sadness—these are feelings; they are more than a body’s response to stimuli. Emotions are more than the fight-or-flight response. The answer lies in the fact that we have also developed a mechanism for monitoring these bodily responses, and we have a mechanism for categorizing them.17
William JamesF (1842–1910) and Carl Lange (1834–1900) first proposed that emotions were a connection between the mind and the body.18 Everyone knows that when we get angry, our heart rate increases and our capillaries dilate to allow more blood to our muscles, which makes us flush and actually heats our skin. As pointed out by George Lakoff and Mark Johnson in their discussion of languages, the metaphors that we use for describing emotions reflect the actual bodily changes occurring during those emotions.19
As was mentioned when I introduced the situation-recognition system (Chapter 6), I suspect that the cortex is one big categorization machine. (A computational discussion of the mechanisms of categorization can be found in Appendix C. As discussed therein, the interaction between these categorization processes and reality produces categories that are often similar but can diverge between individuals. The fact that not everyone agrees on the specific boundaries of these categories does not imply that there is not a species-based set of responses that are categorized similarly between individuals.) If we applied this categorization machinery to our internal, visceral sensory system (How fast is your heart beating?), we would categorize these Pavlovian responses. Different parts of the cortex receive different inputs and categorize different aspects—visual cortex is processing visual input because it receives input from the retina (via the thalamus); auditory cortex is processing auditory input because it receives input from the cochlea (again via the thalamus and other structures). Is there a cortical area that receives input from the hypothalamus, that can recognize these internal states, something that receives visceral inputs?
Anatomically, that part of the brain is the ventromedial prefrontal cortex,G sometimes known as the orbitofrontalH cortex, but also including the insula.I These neurons receive input from the body to both represent and control the emotional bodily states, a process referred to as interoception.21 This localization of emotional processing was initially discovered by studying patients who had damaged prefrontal cortices.22 These ventral prefrontal cortices have the necessary anatomical connectivity, are known to be active during decision-making (as evidenced by fMRI), and play an important role in decision-making (as evidenced by lesion studies).23
The most famous of these patients was Phineas Gage, a railroad worker who in 1848 had a tamping iron (a 13-pound, 3-foot-long, 1.25-inch-diameter cast-iron rod) blasted through his skull. His job was to tamp (pack) down the blasting powder before it was set off. Unfortunately, the process of tamping it down set it off and drove the iron stick through his skull. Remarkably, Gage survived; even more remarkably, he recovered from his injuries and lived for an additional dozen years.24 Gage had surprisingly specific if complex deficits from his injuries—his physical and sensory abilities, his language and intellectual abilities, his ability to perceive the world, his memory, and his ability to take action were all intact. What changed, however, was his personality and his ability to make decisions. Reconstructing the path of the iron bar through a brain sized to match Phineas Gage’s suggests a complete destruction of the prefrontal part of the brain, including both ventral and dorsal aspects. The dorsal prefrontal cortex seems to be involved in the ability to plan, while the ventral prefrontal cortex seems to be involved in the ability to recognize emotions.25 If both of these were damaged, Gage’s ability to make decisions, particularly intelligent ones, would have been severely compromised.
Literature has often described the difference between humans and robots as emotional—humans have emotion, robots don’t. From Star Trek’s Data, who spends his life wishing he had the emotional interactions that his colleagues do, to the replicants in Ridley Scott’s Blade Runner (and in Phillip K. Dick’s original source book Do Androids Dream of Electric Sheep?, which is a much better title), who are identifiable through their lack of an emotional response, literature has identified the unemotional as “robotic.” In fact, in Blade Runner/Do Androids Dream of Electric Sheep?, the replicants are only identifiable by the fact that when they look at emotional pictures, their eyes do not dilate. In Descartes’ Error, Antonio Damasio follows the description of Phineas Gage with a description of a modern patient (referred to as “Elliot”) with known damage to the ventral prefrontal cortex (confirmed by structural MRI),J who has a specific deficit in emotion. Like the replicants in Blade Runner, this patient was blind to emotion: unable to process feeling, to recognize it, or to show it. In situations where emotions are all but inevitable, Elliot showed no reaction. Damasio quantified this by measuring skin conductance, and found that Elliot’s skin conductance was completely flat.
Historically, philosophers have suggested that there are two decision-making systems, an emotional (hot) decision-making system and a cognitive (cold) system. This hypothesis can be traced back to Plato, who described the human as a charioteer driving two horses, one well behaved and always doing the right thing, and the other wild and hard to control.27;K St. Augustine describes two souls, one living in the City of Man, connected to the animals, filled with emotion, lust, and anger, and the other in the City of God, separated from the animals, filled with reason and deliberation.29 (The inherent dichotomy of this is embodied in the wonderful line from Augustine’s Confessions: “Make me chaste, O Lord, but not today.”30)
Freud’s theory of the unconscious follows this same dichotomy. Freud proposed that there were three components, the id, the ego, and the super-ego.31 This tracks Plato’s three beings in the chariot—the wild horse that needs to be held in check (the id), the angelic horse trying to lead one to reason (the super-ego), and the charioteer trapped in the middle (the ego). Much of modern psychiatry still remains within this dual-systemL hypothesis, based in large part on Freudian concepts of the unconscious. It is, of course, as we have seen, fundamentally wrong. You are not separate from your emotional self, nor are you a charioteer driving two other decision-makers; you are the sum of all of your decision-making systems.
Modern dual-process theories argue for a separation between an unconscious, impulsive (Dionysian) system and a conscious, executive-function (Apollonian) system.32 The typical analogy presented is that of a horse and rider, although some authors refer to this as an elephant and its rider. In this analogy, you are the rider trying to steer the horse (or the elephant), but the horse (elephant) has a mind of its own and sometimes goes its own way.
Translating these dual-process theories into our four action-selection systems suggests that the conscious being you are is the Deliberative system, while the Reflexive, Pavlovian, and Procedural systems are nonconscious. Even if we assume that there are multiple nonconscious systems, this multiple-process theory suggests that the conscious self struggles to control the other systems. This differentiation between the noncognitive (unconscious) other and the cognitive (conscious) self requires that we remove emotion from ourselves and place it in the “other” box. I don’t know about you, but I have always felt (an emotional word!) that my emotions are a critical part of who I am.
Separating the cognitive “self” and the noncognitive “other” also requires that we separate out the Procedural (habit, cached-action) system as part of the “other.” The Procedural system is not emotional in any sense of the word—it is a simple storage of stimulus–response pairs and response–response action chains. Instead of the two processes inherent in dual-process theory (emotional, unconscious, the horse or elephant vs. cognitive self, the rider), we would have to separate three systems (emotional, procedural, self). But, again, the procedural responses are part of us. We will see that these stimulus–response pairs and these action chains are the key to athletes hitting baseballs, making plays, throwing and catching footballs, and making other trained responses. When asked, most athletes talk about coming into themselves, not going out of themselves, when they are “in the zone.”
As pointed out by Antonio Damasio in several books, people like “Elliot” with impaired emotional systems do not make normal decisions.33 In fact, there is now some evidence that people who are psychopathic or sociopathic are calculating and unemotional, making decisions independent of any emotional sensitivity.34
When talking about emotions and the role of emotion in decision-making, it is important to note that our memories of the past are reconstructed, rebuilt each time anew and are not necessarily correct. (We will address this issue in depth in our discussions of episodic future thinking, episodic memory, and deliberation in Chapters 9 and 16.) This means that if we took an action based on one of our other systems—for example, because of our Procedural learning system—we may imagine that we took that action because our Deliberative system wanted to, or we may imagine that we took that action because of an emotional response. Although one may remember taking an action “because it felt right,”35 in fact, the emotional parts of our brain that actually encode “feeling right” might not have been active at all at the time; instead, you may have back-constructed the emotion.36 This has been explicitly seen in studies of craving and relapse to drug addiction: most addicts will report that they remember an unstoppable craving in the days before relapse, but prospective studies that look at addicts during relapse find that only a small proportion actually show craving.37;M
That being said, emotions do play a strong role in our decision-making. Often, situations in the world or situations with our fellow humans produce emotions, which guide our decisions. Emotions can be described as our interpretation of reactive states that have been learned over evolutionary timescales, producing changes in our bodily responses and biasing our actions.
When walking along a dark road at night, sounds begin to make us nervous, and we begin to think we may be seeing things. The developing anxiety enhances our attention to the point where we may even see things that aren’t there. I have not gone into depth about perception in this book, in large part because so much is known about it that I would need an additional book to explain it all.38 Two important concepts, however, are useful here. First, perception is fundamentally an interpretive action.39 Although our retina responds to light, we do not see “light,” we see objects—trees, houses, faces. Although the neurons in our cochlea in our ears respond to sound, we do not hear “sound,” we hear voices and music. Illusions occur because the brain’s interpretation mechanisms are being tricked and process information incorrectly. Second, perception is fundamentally about detecting signals.40 This means that one has to make an estimate of whether the perceived set of light signals is a face or something else. When I look at a tree in broad daylight, it’s pretty easy to realize that’s a tree. But late at night, in the dark, is that a tree waving in the wind or something more sinister?
One time, shortly after we graduated from college, my wife and I were visiting Ireland. We were on the Aran Islands, a very rural part of the westernmost part of Ireland. We were walking back from the pub to the bed-and-breakfast where we were staying. It was pitch black except for the barest of starlight peeking through the quickly moving clouds. Well versed in Irish poetry and Irish myth, it was pretty easy to get spooked by the darkness. By the edge of the road ahead of us, we saw this orange light floating in the air, like a will-o’-the-wisp. There was this smell in the air, like a dog or a wolf. By the time we got to the light, our adrenaline was at full strength and I was ready to jump. As we passed, a young man, sitting on his front stoop, smoking a cigarette, his hand on the head of this big, furry dog, nodded to us, said “Evenin’,” and asked us where we were heading. We told him and he said it was about a mile down the road. And we went on, laughing at our emotional interpretations of the floating orange light.
In signal detection, there are two types of errors one can make, false negatives and false positives. A false negative entails not recognizing something that is there, while a false positive entails recognizing something as being there when it is not. Most signal-detecting mechanisms have parameters (gains and thresholds) that can be adjusted to trade off the likelihood for making false-positive errors from false-negative errors.41 The brain is no exception.
This means that perception, like any signal-detection machinery, has to trade off false negatives (missing something that is there) for false positives (identifying something as being there when it is not). As we become anxious, even frightened, by the dark, our hearts start to pound, our muscles become tense, ready to spring into action, and our attention looks for anything that might demand a response. At the sound of a branch cracking beneath our feet, we might literally jump in the air. (This is fear-potentiated startle, which we have seen before.)
Emotions are particularly important for social interactions. Love at first sight may or may not exist, but lust at first sight certainly does. A person who responds to a situation (say catching a cheating spouse) with anger and rage is going to select fundamentally different actions to take than someone who responds to the same situation with sadness and dejection. In a now-classic example, people offered unfair rewards in the ultimatum gameN show increased activity in the anterior insula (involved in negative emotions) and reject the unfair offer with anger and disgust.43
Studies of emotion have found that not only are many decisions correlated with emotional reactions to the decision itself, but also that unrelated emotional states can affect decisions.44 Inducing a feeling of anxiety before being faced with a decision makes people less willing to take risks; inducing a feeling of anger makes people less likely to deliberate over decisions and more likely to respond quickly.45 This observation (that being emotionally frightened, angry, or sexually aroused makes one less likely to deliberate cognitively) is almost certainly where Plato, St. Augustine, and Freud got their dichotomy between emotion and cognition pulling the person in two directions. However, as I hope I’ve made clear, the interaction is much more complicated than that.
Emotion is often seen as part of our “animal” heritage, while cognition is seen as a “higher,” “more human” decision-making system.46 As I hope I’ve shown throughout this book, that’s a mistaken dichotomy. Rats deliberate over decisions, and emotions are fundamental to the interactions that make us human. A question that arises, however, is whether animals have emotions and feelings. All pet lovers, particularly those with a mammalian pet (a cat, a dog), will tell you that their pets have feelings, but of course, as we saw in the very beginning of the book, we are notoriously bad at judging the agency of nonhuman creatures. My father is convinced that his GPS gets annoyed when it has to recalculate a route because he missed a turn.O On the other hand, apes who have been taught limited language (such as Kanzi the chimpanzee or Koko the gorilla) are able to express emotions in words.47 These apes use emotional words in very parallel ways to humans. In his book The Expression of the Emotions in Man and Animals, Charles Darwin noted that certain emotional expressions cross cultures and species. Anger, disgust, laughter, joy, sadness: these are all consistent signals across all humans and all cultures.
In fact, as we saw in Chapter 4, these facial expressions are similar across mammals. Kent Berridge and his colleagues studied facial expressions of distaste (in response to bitter liquids) and appeal (in response to sweet liquids) and found similar facial expressions across rats, monkeys, and humans (sticking one’s tongue out, making a “yuck” face for bitter and licking one’s lips for sweet). Berridge and colleagues were able to show that these expressions are not about taste on the tongue because they could change distaste to appeal and vice versa by manipulating the opioid system in rats.48
An interesting question, of course, is where emotion arises in the evolutionary timeline. Do bacteria have emotions? Probably not as we understand them. Do reptiles? Unlikely. Again, scientists have not been able to reliably identify them. Do mammals? Almost certainly. One intriguing proposal put forward by Michel Cabanac, Arnaud Cabanac, and André Parent is that amniotes (mammals, birds, and some reptiles) developed a categorization system that can observe the body’s Pavlovian action-selection system and that this categorization system is what we now recognize as emotion.49 This is consistent with William James’ concepts of emotion as actually residing in visceral changes in the body and Antonio Damasio’s concepts of emotion as the recognition and memory of bodily responses, which he refers to as the somatic marker hypothesis.50
Whenever emotions appeared evolutionarily, they are clearly an important part of human decision-making systems. Emotions need to be seen not only as feelings that drive other decision-making components to act, but as action-selection systems in their own right. Emotions interact with our categorization of feelings through visceral changes in the body, which we identify as those same emotions through memory and metaphor: the vigilance of fear, the tension of anger, the heat of lust, the warmth of comfort, the breaking-heart pain of sadness.
• Antonio Damasio (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. New York: Quill Press.
• Antonio Damasio (2003). Looking for Spinoza: Joy, Sorrow, and the Feeling Brain. New York: Harcourt.
• Joseph E. LeDoux (1996). The Emotional Brain. New York: Simon and Schuster.
• George Lakoff and Mark Johnson (2003). Metaphors We Live By. Chicago: University of Chicago Press.