The discussion in the previous week focused a lot on affective states of affair. So, this week we focused on the work of Lisa Feldman Barrett and her colleagues. Feldman Barrett has recently published an incredibly interesting book on human emotion, How Emotions Are Made (2017). Our group discussion ranged between the target article and her wider project. I begin with a very brief description of this wider project in order to set the stage for the discussion of the paper.
In her wider project, Feldman Barrett argues strongly that emotion is socially constructed. She claims that there is a difference between interoceptive states (internal states relating to sensations of aversion or attraction), and full blown emotional experiences. Interoceptive states are described as being one of the key ingredients of emotion. But she goes on to claim that full blown emotion is a culturally acquired and constructed trait. Indeed, until children have acquired language Feldman Barrett claims that they do not technically have emotions and only have a simpler feeling of affect. It is the topic of language and how it shapes emotion that is the subject of this particular paper.
Emotion perception is not automatic, and is shaped by language
In this paper Feldman Barrett and colleagues assert that emotion perception is not an automatic process that take places independently of cognitive and conceptual resources. Their key point is that language is a major context in which this emotion perception takes place. They demonstrate that many of the core pieces of evidence in favour of cultural universals and automatic processing in regards to emotion perception are actually the artefacts of experimental design or evaporate when the context is altered. To see this properly it is best to start with ‘the classical view’ and the experiments on judgements of facial expressions of emotion before then moving to Feldman Barrett and colleagues’ critique.
The classical view of emotion is that there is a distinct set of emotional types (e.g. anger, fear, digust, surprise, sadness, and happiness), each of which is accompanied by distinct patterns of physical states and patterns – especially with regards to the face – that a scientist should be able to identify in a laboratory (Feldman Barrett 2017, pp. 3-4). The classical view goes on to assert that emotions are innate and no learning is required, and that we can automatically detect the basic set of emotions in others.
One set of evidence in favour of the classical view is a range of experiments in which participants were asked to examine facial expressions and make judgements about what emotion is being expressed. These experiments revealed a high level of congruence and accuracy in the judgement of emotions in others.
However, Feldman Barrett and colleagues point out that there are a number of issues here. Firstly, this experimental design is not very ecologically salient because it relies on a highly stereotyped archetypes of emotional facial expression. They are limited in terms of variability and as such do not match well onto emotions in the wild where facial expressions are highly variable and much more ambiguous. Indeed, Feldman Barrett and colleagues go on to suggest that neuroscientific and psychological evidence indicates that language acts to reduce the ambiguity of this information.
This relates to a second issue: language priming. The experimental design itself is flawed because the participants are primed into only focusing on a particular set of emotions. The facial expression to be judged is accompanied by the target set of emotions to be judged (as can be seen in the example above). When one removes this linguistic priming, the level of accuracy and congruence between participants drops markedly.
Furthermore, there are numerous ecologically salient (and farcical) situations in which the observation of a facial expression of an emotion by itself is insufficient for an adequate judgement. Feldman Barrett and colleagues use the example of expressions of pleasure that could easily be mistaken for anger without context (as can be seen in the picture below).
One can see how these kinds of mistakes are very easy to make without context. And that a solitary focus on the face is inadequate. In a less ecologically salient example. The attached main photo at the top of the page shows giant slugs photo-shopped into the hands of guitarists who are in the midst of “rocking out” or “losing themselves in the music”. The trick here shows how changing the context can alter a facial expression of emotion from one of joy or pleasure to disgust.
Lastly, in other lab based research people consistently (60-75%) mislabel fear faces as angry when put in the context of anger themselves (Carroll & Russell 1996). Feldman Barrett and colleagues point out the bad outcomes of this in scary environments such as riots and war-zones.
In summary, the evidence is against the notion that emotion perception is automatic and unmediated by a range of cognitive factors. Without context it can be very hard to identify an emotion from just a facial expression. And a key contextual ingredient in how we understand the emotions of others is language. A final note to close on: the evidence here collected by Feldman Barrett indicates that language has an extremely large impact on human emotions. This fits with the broader discussion of linguistic relativism (sometimes called the Sapir-Whorf Hypothesis) in regards to a number of other cognitive domains that have been previously discussed here (see week 3). Here is a good overview discussion in a recent talk by Lera Boroditsky which nicely summarises the wide range of evidence in favour of the notion that language dramatically shapes how we think and feel.
Carroll, J. M. & Russell, J. A. (1996) Do facial expressions signal specific emotions? Judging emotion from the face in context. Journal of Personality and Social Psychology 70 (2), 205-218.
Feldman Barrett, L. (2017) How Emotions Are Made: The Secret Life of the Brain. New York: Houghton Mifflin Harcourt.
Feldman Barrett, L., Lindquist, K. A., & Gendron, M. (2007) Language as context for the perception of emotion. Trends in Cognitive Science 11 (8), 327-332.