What Are All These Emotions? Lisa Feldman Barrett’s Constructionism about Emotions (part 1)

 I.

 

When I started this blog, I said I wanted to begin with a focus on personality disorders and emotions. I have somewhat made good on the first promise, but not the second. In this post, I want to begin to explore some of my own perplexities about emotions. I am going to start with Lisa Feldman Barrett’s book How Emotions Are Made: The Secret Life of the Brain (2017).

 

Barrett is a psychologist and neuroscientist whose work I first encountered through mental health Twitter, back when academic Twitter was still a thing. She is very good at popularizing her views, and I should note right away that she is very clear, if you watch her videos or listen to her podcasts, that How Emotions Are Made is a popularization—and thus at times a drastic simplification—of the views that she and her lab have worked out in hundreds of academic papers. My impression of Barrett is that she revels in being a bit of a bête noire, that she likes to play the role of depriving us of the self-evident truths imparted in pre-school—or MSW programs. She’s not here to tell us that Santa doesn’t exist; she’s here to tell us that the brain and the emotions aren’t what we were told they were.

 

In what follows, I’m going to give a sympathetic reading to the first fifty pages of How Emotions Are Made, with an eye in particular to the implications for psychotherapy. At a later date, I will give air time to some rival views, especially those of the late affective neuroscientist Jaak Panksepp.

 

 

II.

 

Barrett begins with a revolutionary narrative. Up until now, a certain theory of emotions has held sway. She calls this the “classical view.” I am not entirely convinced that this view is as old or as entrenched as Barrett says it is, but it has certainly been prominent since at least the 1960s. The classical view of emotion goes something like this:

 

There are distinct basic emotions. Each basic emotion has a unique fingerprint in the body and brain that allows it to be differentiated from the other basic emotions. Moreover, each basic emotion has a unique matching facial expression and shows up distinctively in voice and posture. These basic emotions and their corresponding expressions are universal. The basic emotions arose through evolution. They function as automatic responses to outside events, like reflexes.

 

Barrett thinks that everything about this view is wrong. Against it, she opposes what she calls the “theory of constructed emotion” (what she more mildly calls the “conceptual act theory of emotion” in her academic writings—see p. 370n9), which I will discuss more at a later date. In a nutshell, Barrett thinks that emotions are constructed—socially, but also psychologically. They are constructed through concepts, which our brain uses to make predictions about how the world and our own bodily states will be. They are not universal except insofar as the emotion concepts we use are universal; however, the appearance of universality arises from the ease with which many of us learn and teach each other emotion concepts. Having an emotion like fear or anger is a bit like perceiving a ball or strike in a baseball game (this is my analogy). It’s something we have to learn how to do, but once we learn how to do it, it comes across as more or less automatic.

 

For now, I want to focus on Barrett’s empirical criticisms of the classical view. Part of what is so compelling about this book is Barrett’s extremely readable criticism of the assumptions baked into classical experiments.  

 

 

III.

 

According to Barrett, the classical view derives from Darwin’s The Expression of the Emotions in Man and Animals, which claims that human beings and other animals express emotions in consistent and universal ways. Darwin’s position received important uptake in the 1960s among a group of researchers the most famous of which is the psychologist Paul Ekman. (Going forward, I am going to use his name as a stand-in for his team.) In the 1960s, Ekman and colleagues trained actors to produce a series of photographs exemplifying what they held to be the six basic emotions: anger, fear, disgust, surprise, sadness, and happiness (pp. 4-5). You can find these photographs easily through a Google Image search, for example, here.

 

What Barrett calls the “basic emotion method” is then used to design experiments that go thus: Experimenters give people a sequence of faces. Next to each face, the six basic emotions will be listed; the subjects have to match the emotion to the face, much like a multiple-choice test. There are variations on this design, such as experiments where two faces will be displayed along with a story, such as “Her mother just died, and she feels very sad.” Subjects then have to choose the correct facial expression (pp. 5-6). The studies have been done many times, and they appear to show high levels of correct answers and high levels of agreement across cultures. In particular, a famous study with a remote tribe in Papua New Guinea (Ekman et al, “Pan-Cultural Elements in Facial Displays of Emotion,” Science (1969)) appeared to lend credence to the idea that there are discrete basic emotions, that these emotions are universal, and that they have their “fingerprint” in human facial expressions.

 

These are popular views. One could even say these views are materially embedded in Western culture—just think of emojis. Think of books and TV shows for children. Barrett also brings up the failed attempt by the Transportation Security Administration (TSA) to apply Ekman’s ideas to airport security. These ideas have also been influential in psychotherapy. Marsha Linehan, for example, cites several of Ekman’s papers in the primary DBT treatment manual, Cognitive-Behavioral Treatment of Borderline Personality Disorder (1993). (The actual extent to which the DBT theory of emotions aligns with Ekman’s view is a much more complicated matter, however.)

 

 

IV.

 

Barrett identifies a number of problems with the “basic emotion method” (again, this is what she calls the research program initiated by Ekman and colleagues). First, the original facial expressions were arrived through stipulation—the actors were coached by Ekman and his team to produce specific expressions (p. 10). To take out the element of subjective judgment, one might try other methods focused on what humans actually do with their faces when experiencing emotions, such as facial electromyography (EMG) experiments—hooking electrodes up to facial muscles and then evoking emotions. Reviewing the research on facial EMG, Barrett concludes, “In study after study, the muscle movements do not reliably indicate when someone is angry, sad, or fearful; they don’t form predictable fingerprints for each emotion” (p. 7).

 

Another problem is that context matters a great deal in reading faces. For example, Barrett’s lab took photographs from the book In Character: Actors Acting and then compared three conditions: subjects who just saw the face photographs in the book, subjects who only read the stories paired with the photographs, and subjects who saw the photographs and read the stories. The three groups assigned emotions in markedly different ways. Barrett illustrates the point by placing a photograph of actor Martin Landau in between Ekman’s photographs of surprise and fear; she asks the reader to determine which of the two emotions Landau is displaying. As it turns out, we cannot really make sense of Landau’s face without more information about what the scene prompt has him looking at (pp. 9-10).

 

This point about context makes a lot of sense to me. It chimes with both my philosophical and my therapeutic commitments. Philosophically speaking, there is a rich tradition in phenomenology (I’m thinking of Heidegger and Merleau-Ponty, among others) that says something like this: there is no such thing as an isolated perception; we are always taking in and engaging with a larger whole, and it is against the backdrop of this whole—call it “background,” “situation,” “context,” “world,” “Umwelt,” or whatever you’d like—that we make sense of particular things. Therapeutically, it makes sense too: mentalizing, or our complex ability to perceive what is going on in others’ minds, including their emotions, depends on taking in all sorts of information, not just external physical cues. We need to know what has been happening in the person’s world, what they have been trying to do or getting ready to do. Barrett illustrates this nicely with a photograph of Serena Williams, the details of which I will not give away, as the effect rather depends on reading the book in order.

 

Other attempts to find a “fingerprint” for emotions also fail, Barrett argues. For example Ekman and colleagues also have studied autonomic nervous system indicators, such as heart rate, temperature, and skin conductance, in relation to evoked emotions. (Most famously in Ekman et al, “Autonomic Nervous System Activity Distinguishes among Emotions,” Science (1983).) Basically, hook people up to some instruments that will measure these things, then do something to elicit emotions and track what happens. Here is the problem: “The famous 1983 study evoked emotion in a curious way—by having test subjects make and hold a facial pose from the basic emotion method…. While face-posing, subjects could use a mirror and were coached by Ekman himself to move particular facial muscles” (p. 12). In short, there is rather a lot of circular reasoning going on. There are other problems as well, including the fact that four meta-analyses have failed to yield physiological fingerprints (pp. 12-14).

 

Finally, one might expect the “fingerprint” for emotions to lie in the brain, rather than in the face or the autonomic nervous system. Unfortunately, the results here have been equally disappointing. A tradition dating to the 1930s has localized the “fear circuit” in the amygdala, but the research has been subsequently debunked. For example, Barrett’s own research suggests that amygdala activation is actually linked to seeing novel faces, and that activation decreases if you see the same face again or see the face in profile. The more important point is that of the “degeneracy” of neurons: there are many different neural pathways that can lead to fear, and a single “core system” in the brain can be involved in many different experiences and mental processes (pp. 17-19). Subsequent meta-analysis of neuroimaging studies by Barrett’s team failed to yield any neural fingerprints for anger, disgust, happiness, fear, or sadness (20-21): “no brain region contained the fingerprint for any single emotion” (22).

 

 

V.

 

Having argued against the “fingerprint” component of the classical view, Barrett then takes on the universality claim. Ekman-style experiments have purported to show that the vast majority of people around the world, independent of culture, “correctly” identify the basic emotions in the face photographs in question. But there is a simple alternative explanation for this consensus: the very design of the experiment primes people to answer the questions in the way the experimenters expect. Anyone who has ever taken a multiple-choice test gets this. The answer choices shape what you see.

 

To avoid this problem, some experiments use free labeling: you see a face and then just have to say what emotion you see. Only about 58% of subjects get these questions “right,” compared to about 85% in the West and 72% in places like Malaysia (p. 45). Even lower “success” rates follow when you take out the word “emotion” from the question and just ask “What word best describes what’s going on?” In short, the priming that occurs in the forced-choice nature of the classic Ekman-type experimental design could be creating the impression of the universality of emotion facial expressions (pp. 45-46).

 

Barrett’s team conducted a study of the faces with the remote Himba tribe in norther Namibia, which has had little contact with Westerners. Her team had tribe members freely sort the faces and then label them, which resulted in markedly different categories from Ekman’s six basic emotions (pp. 46-49). To be sure, Disa Sauter and her colleagues also carried out a facial expression study with this tried and arrived at results that appeared to confirm Ekman’s views. Barrett’s team subsequently replicated the experiment using the published methodology and arrived at different results from Sauter et al; it eventually turned out that Sauter’s team had not disclosed in published form an important step of their design, one that most likely had the effect of teaching the Himba Western emotion categories prior to completing the experiment (pp. 49-50).

 

Other evidence can be marshalled here. Smiling was not associated with happiness in ancient Rome, and the Romans used a whole variety of other gestures that we do not (p. 51). Other studies from around the same time as Ekman’s original one yield results that spoke against universality, but they were published in less prestigious venues (p. 52). Cultural anthropology tells us that different cultures exhibit dramatic variation in how they understand emotions (p. 53). And so on.

 

 

VI.

 

In the next post, I will turn to Barrett’s alternative, constructionist account of emotions. To conclude this post, I think Barrett’s critique of the classical theory has powerful implications for psychotherapy. First, most contemporary psychotherapies involve a psychoeducation component about emotions. It is worth asking to what extent those psychoeducation components are premised on a faulty theory of emotion—a much larger question, of course, than I can go into here. After all, these psychoeducation pieces usually try to teach patients how to recognize and identify discrete emotions. This is not necessarily a bad thing! However, it may turn out that what is actually going on is better thought of as teaching patients how to construct, rather than observe, their emotional experiences. Second, Barrett raises important questions for the whole topic of cultural competency. I also find myself wondering about how we should think about conditions such as alexithymia or autism spectrum disorder in which individuals perceive or exhibit emotion in ways that are considered outside of the cultural norm. Third, it is worth asking what the implications of the critique of the classical theory are for the field of psychopathology—and especially for the whole messy field of emotional disorders. In this blog, I have already had occasion to discuss the contemporary debate between categorical and dimensional approaches to the classification of mental disorders, but I have focused mainly on personality disorders; however, similar questions can be raised about emotional disorders (major depression, bipolar disorder, anxiety disorders, and so on). Finally, what implications does Barrett’s critique have for the distinction common to so much of psychotherapy between “primary” and “secondary” emotions?

It may sound as though I accept Barrett’s position wholesale. I am not yet entirely convinced, though. In particular, I worry that what she calls the classical theory of emotion bundles together too many different things. Logically speaking, there is no contradiction in rejecting the fingerprint theory while accepting the existence of a few basic emotions. I do find myself worrying that Ekman is being used to set up a bit of a straw man.