Part Of: Affective Neuroscience sequence
Content Summary: 1800 words, 9 min read
Content Note: This post discusses Barrett (2017). How Emotions are Made. 

Essentialism vs Constructionism

This book extends an old debate between essentialism and constructionism. 

Essentialism interprets cognition as produced by innate mental modules which produce distinct behavioral programs. It seeks to localize function onto structures in the nervous system. Lesion evidence is considered suggestive. Essentialism relies on the abstraction hierarchy, with less abstract, more instinctive decisions operating underneath the cortex, with more abstract and reflective decisions manifesting under conscious control. It is often couched in terms of dual process theory.

Constructionism interprets cognition as whole-brain prediction cascades, with predictions flowing from control and body-oriented networks to primary sensory cortices. Perception is inference. These cascades flow through intrinsic connectivity networks (ICN), with each function subserved with ever-changing microcircuits that by their nature defy localization. This theory meshes nicely with Friston’s active inference theory, with descending predictions and ascending prediction errors. 

Classical Model of Emotion

The classical model of emotion is essentialist in character. The classical view posits at least six primary emotions: anger, fear, disgust, surprise, sadness and happiness. Each basic emotion is supposed to have a unique neural substrate, physiological signature, and motor program. The basic emotion method, championed by Paul Ekman and others, asks people to match faces to emotions. People from every culture tested tend to draw the same map between these faces and the corresponding emotions, achieving about 72% accuracy. Such cross-cultural universality is taken as evidence of innateness.

Concerningly, performance plummets when you modify the experimental setup. If you remove the emotion words and ask subjects to describe the imagined feeling in their own words, performance drops from 72% to 58%. If you remove the language component entirely, and ask whether people in two different images are feeling the same emotion, accuracy drops to 42%. Finally, subjects were asked to repeat an emotion word like “anger” over and over”. Eventually, the word becomes just a sound to the subject (“ang-gurr”) that’s mentally disconnected from its meaning. People in such a state of semantic satiation are even more impaired at emotion masking tasks, achieving 36% accuracy. 

Other evidence converges.

1. When you measure emotional expression with electromyography, there isn’t much muscular consistency within emotions. Extreme variation is the norm.
2. The search for physiological fingerprints has also failed. Anger and fear, for example, yield very similar effects on heart rate.
3. The search for neural fingerprints for various emotional circuits have also

Taken together, these data suggest that emotion may not be innate. Emotions may be conceptual in character, with a curious dependency on language. 

Constructivist Model of Emotion

Having disputing the classical theory of emotion, Barrett turns to constructing her alternate account.  

Her constructivist model of emotion is grounded in body budgets: interoceptive and visceromotor mechanisms that allostatically regulate the body. Body budgets are summarized in the two dimensions of affect: valence and arousal. While emotions are cultural constructions, affect is a human universal. The constructivist model dovetails with the circumplex theory of affect.

Next, Barrett addresses the formation of emotion concepts. There are several accounts of concept on offer: 

1. Classical concepts which specify a list of necessary and sufficient properties, like a dictionary.
2. Prototype concepts which are grounded in perceptual similarity (c.f., “I know it when I see it”)
3. Affordance concepts which categorize not by perceptual similarity, but instead via similarity of their use. 

The concept of “chair” is not perceptually coherent; chairs are instead united by how our bodies use them. Affordance concepts are specific to humans, because they require language:

> Waxman demonstrated this power of words in infants as young as three months. The infants first viewed pictures of different dinosaurs. As each image was shown, infants heard an experimenter speak a made-up word, “toma.? When these infants were later shown pictures of a new dinosaur and a non-dinosaur such as a fish, those who had heard the word could distinguish more reliably which pictures depicted a “toma,” implying that they had formed a simple concept. When the same experiment was performed with audio tones instead of human speech, the effect never materialized. 

Emotions as Social Reality

Some affordance concepts can be understood as social reality. Most things in your life are socially constructed: your job, your street address, your government and laws, your social status. Money is a classic example of social reality. Given a rectangle of paper with a dead leader’s face printed on it, and you can “buy products”. Barrett suggests that emotions are affordance concepts, synchronized through language. Emotions are like money: a pillar of social reality. 

Consider a man stamping his foot. The man thinks he is removing mud from his boot; you think he is angry; your friend thinks he is dejected. Who is right? On the constructivist model, questions of accuracy are unanswerable in an objective sense. There are no observer-independent measurements that can adjudicate. The best you can do is communicate reasons and seek consensus.  This is not to say that emotions are “just in your head.” That phrase trivializes the power of social reality. Money, reputation, laws, government, friendship, and all of our most fervent beliefs are also “just” in human minds, but people live and die for them. They are real because people agree that they’re real.

Since emotions are synchronized within a culture, they tend to vary across cultures. You can see this in how individualistic countries interpret guilt and shame differently from collectivist ones.

You are probably unfamiliar with an emotion called liget. It’s a feeling of exuberant aggression experienced by a headhunting tribe from the Philippines, the Ilongot. Liget involves intense focus, passion, and energy while pursuing a hazardous challenge with a group of people who are competing against another group. The danger and energy instill a sense of togetherness and belonging. Liget is not just a mental state but a complex situation with social rules about which activities bring it on, when it is appropriate to feel, and how other people should treat you during an episode. To a member of the Ilongot tribe, liget is every bit as real an emotion as anger or surprise. Westerners surely do experience pleasant aggression, for example video game players cultivate it during first-person shooter games. But these people are not experiencing liget with all its meaning, prescribed actions, and body-budget changes.

People moving between cultures are often shocked by the difference in emotion concepts. Smiling was not associated with happiness in Roman culture, for example.

But American culture prefer high arousal, pleasant states; immigrants often complain that their cheeks ache from smiling more frequently. But the more time someone spends in America, the more their emotions become attuned to the American context. Your pleasant emotion concepts became more granular in a process known as emotion acculturation. 

Because emotion concepts affect our body budgets, nurturing your emotional intelligence is a way to aspire towards health. For Barrett, this means practicing the use of new emotional concepts (e.g., borrowing them from other concepts) and generally pursuing emotional granularity. Emotional granular individuals were 30 percent more flexible when regulating their emotions, less likely to drink excessively when stressed, and less likely to retaliate aggressively against someone who has hurt them. In contrast, lower emotional granularity is associated with all sorts of afflictions. While the causal role for emotional granularity is not yet known, it conceivably plays a role. It would explain why parents find it helpful to teach their children to “use their words”. 

Affective Programs in Subcortex

Barrett’s treatment of neuroscience is rather cortex-oriented. What about the subcortex? Barrett’s skepticism notwithstanding, I find the evidence for modules in the subcortex to be persuasive

Consider anger. Electrical stimulation of the brain in specific subcortical areas elicit three subvarieties of anger behavior: reactive anger, predatory behaviors (quiet biting attack), and inter-male aggression. These behaviors also are differentiated pharmacologically. Chlordiazepoxide reduces affective attack and increases predatory attack; amphetamine increases affective attack without no effect on predatory attack. In the evolution of Sapiens, affective anger and predatory behaviors seem to respond differentially to selection pressures. Modular circuits have also been identified for fear behaviors, anxious behaviors, seeking behaviors, and other behavioral expressions of emotion.

Such affective programs in the subcortex are organized hierarchically. For example, anger is organized as cortex > amygdala > hypothalamus > periaqueductal gray (specific nuclei within these structures). Lesions to higher areas does not diminish affective behaviors, but damage to lower areas are much more severe. Higher cognition appears to play a role in modulating these programs. Decorticate animals activate their affective programs more vigorously, and are less receptive to change. Affective programs are also evolutionarily ancient; we find them in nearly every species studied. 

Approaching The Debate

Constructionists emphasize the role of cortex in emotional experiences and emotion perception, whereas essentialists emphasize the role of subcortical structures in emotional behaviors. 

Barrett claims that emotions have no neural signature, and therefore disagreements about emotional attribution (“I did it because I was angry”) cannot be resolved by observer-independent measurement. Perhaps consciousness does not have privileged access to these states, and self-report defaults to confabulation. Perhaps emotion perception must resort to consensus seeking. But it seems possible to objectively resolve questions of emotional behavior and even changes in disposition. 

If you assume that these subcortical programs generate conscious feelings, then you must conclude that animals have conscious feelings. But many constructionists believe that only affect (arousal and valence) are accessible to consciousness. In this view, the complexity of emotional experiences must come from concepts. 

Barrett claims that only humans experience emotions. This claim rests on four propositions

1. Emotion concepts operate independently of affective programs, beyond the dimensions of salience and arousal.
2. Emotions are goal-concepts, with teleological rather than perceptual coherence.
3. Goal-concepts are only acquired from language, non-human primates cannot acquire them.
4. Emotional goal-concepts can be synchronized with language, and are embedded in the fabric of social reality.

I am skeptical that all four are true. That said, I am not in a position to properly evaluate them. 

I will say that constructionism brings much-needed attention to the importance of conscious concepts, how they inform emotional inferences about e.g., facial expressions. It sheds light on how culture produces different responses to shame and guilt. And it makes me wonder exactly how 

Until next time.