Essentialist Biases in Reasoning About Emotions

Berent, Iris; Feldman Barrett, Lisa; Platt, Melanie

doi:10.3389/fpsyg.2020.562666

ORIGINAL RESEARCH article

Front. Psychol., 23 September 2020

Sec. Cognitive Science

Volume 11 - 2020 | https://doi.org/10.3389/fpsyg.2020.562666

Essentialist Biases in Reasoning About Emotions

$\r\nIris Berent*$ Iris Berent^*

Lisa Feldman Barrett

Melanie Platt

Department of Psychology, Northeastern University, Boston, MA, United States

A large literature debates whether emotions are universal and innate. Here, we ask whether reasoning about such matters is shaped by intuitive Essentialist biases that link innateness to the material body. To gauge the perception of innateness, we asked laypeople to evaluate whether emotion categories will be recognized spontaneously by hunter–gatherers who have had no contact with Westerners. Experiment 1 shows that participants believe that emotions are innate and embodied (facially and internally) and these two properties correlate reliably. Experiment 2 demonstrates that the link is causal. When told that emotions are localized in specific brain areas (i.e., embodied), participants concluded that emotions are innate. Experiment 3 shows that this naïve view persists even when participants are explicitly informed that these emotions are acquired. Our results are the first to suggest that laypeople incorrectly believe that, if emotions are embodied, then they must be innate. We suggest that people’s failure to grasp the workings of their psyche arises from the human psyche itself.

Introduction

A large literature debates whether emotions are universal and innate and, correspondingly, whether they are uniquely expressed in the face, body, and brain (e.g., Ekman et al., 1969; Tooby and Cosmides, 2008; Sauter et al., 2015; Siegel et al., 2018 vs. Gendron et al., 2018; Barrett et al., 2019, for proponents and opponents of innate emotions, respectively). Despite hundreds of studies on the topic, the debate shows no signs of abating. The factors that contribute to the persistence of the innateness debate are unknown. Here, we explore the possibility that the stalemate emanates from biases inherent in naïve psychology–the tacit beliefs that people hold about the bodies and the minds of others (Baron-Cohen et al., 1985; Keil, 1986; Spelke, 1994). Very little research has specifically addressed the role of naïve psychology in views of emotion. To begin investigating this question, we examined laypeople’s beliefs about whether emotions are universally expressed in the material body and whether such beliefs are linked to their reasoning about innateness.

Previous research has shown that adults rate emotions highly on biological attributes such as “naturalness,” akin to bodily state and natural kind categories, as distinct from abstract nominal kinds and cognitive states (Lindquist et al., 2013). These observations suggest that laypeople’s understanding of emotions is guided by Essentialism–an intuitive psychological principle that governs reasoning about living things (Keil, 1986; Gelman, 2003). Per Essentialism, living things are believed to possess an innate immutable essence that transfers from parent to their offspring (Gelman, 2003). For example, young children know that racoons cannot turn into skunks by painting their exteriors, as such manipulations do not alter their inherent essence (Keil, 1986). Children also know that a baby rabbit raised by monkeys will maintain the properties of its biological kind (e.g., it will grow up to exhibit long ears and eat carrots, rather than have short ears and eat bananas; Gelman and Wellman, 1991). These observations suggest that, in naïve biology, innate biological traits must possess a special immutable essence.

In addition, a careful read of the literature on essentialism suggests that people further believe this innate essence must form part of the material body (for review, Berent, 2020). Children, for example, assert that a puppy inherits its brown color from its mother by the transfer of some tiny brown pieces of matter (Springer and Keil, 1991). Children further believe that this innate essence is physically localized in the material body (i.e., the body of the puppy and its mother). Furthermore, the essence is presumed to be in a specific place within the body. So when children are invited to discover which animal is hidden inside a fossil, they insist that the sample be taken from the center of the fossil, but they do not do the same when interrogating non-organic material, such as rocks or metals (Newman and Keil, 2008). Other findings suggest that material bodily traits (e.g., foot size) are more readily perceived as innate than psychological characteristics (e.g., shyness, Heyman and Gelman, 2000). These results suggest that, in naïve biology, not only are innate traits embodied, but also conversely, embodied biological traits are innate.

This Essentialist belief provides a powerful causal mechanism for reasoning about the innateness of psychological traits, such as emotions. If people assume that (a) innate traits must possess an inborn essence and (b) the essence must form part of the material body (i.e., essences are embodied), then their tendency to interpret psychological traits as innate will depend on their perceived materiality and embodiment. Traits that are perceived as immaterial and disembodied should be viewed as ones that are unlikely to be innate (i.e., a negative innateness bias), whereas those that are readily linked to the material body would be presumed innate (i.e., a positive innateness bias).

The contrast between knowledge (epistemic states, such as knowing that “objects are cohesive”) and emotions (e.g., “anger”) allows us to evaluate this prediction. Intuitively, emotions appear to be embodied in the material body (e.g., in the face), whereas knowledge seems disembodied and immaterial. If our intuitive nativist intuitions are guided by our beliefs about the materiality of psychological traits, then we expect people to exhibit opposite nativist intuitions toward knowledge and emotions (Berent et al., 2019a; Berent, 2020).

Recent studies from our lab have evaluated people’s nativist intuitions toward knowledge. We found that people presume that knowledge is not innate, and they maintain this bias despite explicit evidence to the contrary (Berent et al., 2019b; for converging evidence, see Wang and Feigenson, 2019). Moreover, the perception of knowledge as innate was linked to its perceived materiality. Specifically, epistemic states tended to be viewed as immaterial–the less material the trait, the less likely it was to be viewed as innate (Berent et al., in press). Thus, for epistemic traits (i.e., knowledge), people are negatively biased against innateness (Berent et al., 2019b, in press).

Here we test for the complementary (positive) bias for emotions. Observing that a given emotion category (e.g., fear) is expressed in the body (e.g., in the face), people would presume that this emotion is innate. Moreover, if in intuitive psychology, most emotions are embodied, then people would tend to view emotions as innate. We hypothesize that this embodiment–innateness link is causal: people are inclined to view emotions as innate precisely because they believe that emotions are embodied.

Our three experiments test this hypothesis. Experiment 1 asked whether the (perceived) innateness of an emotion category (e.g., fear) is linked to its propensity to be expressed in the human face and body. Experiment 2 determined whether this association between embodiment and innateness is causal, such that people are more likely to believe that emotions are innate when these emotions are described as embodied (i.e., localized in specific brain regions) compared to when they are not (i.e., devoid of brain localization). We predicted that embodiment would promote the perception of emotions as innate. Experiment 3 investigated whether people maintain their conviction that emotions are innate even when informed that the emotions in question are acquired.

Following Samuels (Samuels, 2004), we define psychological innateness as the propensity for spontaneous emergence, without reliance on learning. To gauge innateness, our experiments thus invited participants to determine whether a given emotion will be recognized spontaneously by a hunter–gatherer who has had no interaction with Western people (i.e., no opportunity for learning). We reasoned that, if the emotion is considered innate, then it should be perceived as one that is likely to emerge spontaneously in the hunter–gatherer, hence, as potentially available to all humans (i.e., universal).

We chose this approach over explicit innateness ratings for two reasons. First, we believe this tacit measure is more likely to tap into laypeople’s tacit nativist intuitions. Second, this very same approach has been amply used in past research in affective science (Ekman et al., 1969; Sauter et al., 2010; Gendron et al., 2014). While our present research decidedly concerns laypeople’s intuitions, rather than scientists’, a convergence would open up the possibility that the relevant biases might apply more generally. Our present investigation seeks to determine whether laypeople’s nativist intuitions concerning emotions depend on their perceived embodiment.

Experiment 1: Emotions Are Viewed as Embodied and Innate

Participants in Experiment 1 rated how likely emotions are to (a) be expressed in the face, (b) elicit a physiological response in the body, and (c) be recognized spontaneously by hunter–gatherers who have had no previous contact with Westerners (suggesting that emotions are innate and universal). We hypothesized that people would consider emotions as embodied and innate and that beliefs about the innateness of emotions would be associated with beliefs about their embodiment.

Methods

Participants

Sixty participants took part in this experiment. In this and all subsequent experiments, participants were recruited from Amazon Mechanical Turk. They were all adult (18 years or older) native English speakers who resided in the United States. All participants were reportedly free of reading disorders and had reportedly taken no upper level courses (beyond an introductory level) in Psychology. In experiment 1, 37% of the sample had a high school education, 50% had a college education, and 13% had a postgraduate education.

To be included in the sample, participants in Experiment 1 must have further provided a coherent answer to a question presented at the end of the experiment (asking them to explain their response) and spent at least 200 s on the experiment. All participants signed a written informed consent form, approved by the Institutional Review Board of Northeastern University.

Sample sizes in Experiments 1–3 were determined by pilot work as well as by previous research (e.g., Lindquist et al., 2013). Accordingly, we determined that samples of at least N = 60 per group were optimal to yield modest to large effect sizes.

Materials and Procedures

The materials consisted of twenty emotion terms (anger, disgust, fear, happiness, surprise, love, contentment, excitement, joy, pride, sadness, shame, contempt, jealousy, relief, desire, embarrassment, pain, awe, and boredom). This word list was compiled from Lindquist et al. (2013) and Cordaro et al. (2018). The 20 emotion terms were presented in random order. For each emotion term, participants were asked to consider three questions (counterbalanced for order, see Supplementary Appendix 1).

One question asked participants “how likely is it that a person who feels this emotion will “show” it as a distinct facial expression? In other words, how likely is it that you could tell that a person is experiencing this particular emotion from the person’s facial expression?” A second question asked, “how likely is it that each emotion will elicit a physical bodily response (e.g., a change in blood pressure, heart rate, perspiration)?” To evaluate the universality of emotions (a gauge of innateness), a third question invited participants to consider an experiment involving people in a small-scale society living in a remote part of the world, following the methods of actual published experiments (e.g., Ekman et al., 1969; Ekman and Friesen, 1971). Participants were told that “these people are hunter-gatherers; they have no electricity, so they have no access to media, and most of them have had no interactions with Westerner people before.” They were then asked to reason “how likely is it that they would recognize an emotion from the list below in facial expression. For concreteness, suppose you presented them (with the help of an interpreter) with a short story, depicting an event (e.g., “a person encounters a threatening animal in the jungle and he is afraid for his life”). Next, you showed them two pictures, depicting two distinct facial expressions (“horror” vs. “euphoria”), and asked them to pick which picture corresponds to the person depicted in the previous story. How likely is it that their responses would match those of United States participants?” For each of the three questions, participants provided a rating for each of the 20 emotion words on a seven-point scale (1 = very unlikely; 4 = neither likely nor unlikely; 7 = very likely).

Results

Figure 1 plots participants’ rating of how likely each emotion is to be expressed in the face, manifest in the body, and emerge universally.

FIGURE 1

Figure 1. The mean rating of emotions with respect to their manifestation in the face and body, and their universal recognition. “Basic” emotions are marked by asterisks. Error bars are 95% CI for the difference between the means. The “neutral” midpoint of the rating scale (4) is indicated by the dashed line.

We first compared the mean response to each of these three questions against the scale’s neutral midpoint (4-“neither likely/unlikely”) using a paired t-test. To evaluate the generality of our conclusions across participants and items (specific emotions), we conducted these tests using both participants (t₁) and items (t₂) as random effects.

The results confirmed that people considered these emotions as likely to imprint their marks on both the face [M = 5.33, t₁(59) = 13.00, p < 0.0001, d = 1.68; t₂(19) = 7.04, p < 0.0001, d = 1.57] and the body [M = 5.06, t₁(59) = 9.91, p < 0.0001, d = 1.28; t₂(19) = 4.64, p < 0.0002, d = 1.04]. Similarly, people believed these emotions would be likely recognized spontaneously by hunter–gatherers, a result significant by participants [M = 4.45, t₁(59) = 3.48, p < 0.001, d = 0.45] and marginally so by items [t₂(19) = 1.97, p < 0.07, d = 0.44]. This finding suggests that participants tended to view these emotions as innate and universal.

Strikingly, the presumed innateness of these emotions (gauged by their universality) correlated with beliefs about their propensity to manifest in the face [r₁(58) = 0.44, p < 0.0005; r₂(18) = 0.94, p < 0.0001] and in the body [r₁(58) = 0.39, p < 0.003; r₂(18) = 0.81, p < 0.0001]. Additionally, beliefs about the face and body manifestations correlated strongly [r₁(58) = 0.71, p < 0.0001; r₂(18) = 0.83, p < 0.0001].

To compare laypeople’s intuition and distinctions made in affective science, we further examined people’s rating for certain emotion categories that are classically proposed as universal “basic emotions” (Anger, Disgust, Fear, Happiness, Sadness, and Surprise; classified following Ekman, 1992; Sauter et al., 2010) and “non-basic emotions.”

The 2 Emotion (Basic/Non-basic) × 3 Question (face/body/innateness) ANOVA yielded a significant effect of Question [F₁(2,118) = 25.81, p < 0.0001, η² = 0.09; F₂(2,36) = 21.49, p < 0.0001, η² = 0.10], as the universality question elicited a lower rating than both the facial and bodily questions (Tukey HSD, p < 0.0001; p < 0.0001, respectively, by both participants and items).

Critically, the main effect of Emotion type was highly significant [F₁(1,59) = 200.95, p < 0.0001, η² = 0.23; F₂(1,18) = 8.84, p < 0.009, η² = 0.27], as putatively basic emotions (M = 5.72) were rated higher than non-basic emotions (M = 4.61). The interaction between Emotion and Question was marginally significant [F₁(2,118) = 6.41, p < 0.003, η² = 0.01; F₂(2,36) = 1.82, p < 0.18, η² = 0.01], as the effect of Emotion type was largest for the innateness question Δ = 1.37 (for the face and body questions, Δ = 1.08 and Δ = 0.89, respectively). Thus, laypeople’s beliefs in the embodiment of emotions and their innateness were particularly pronounced for putatively basic emotions.

Summarizing, Experiment 1 shows that, overall, people tended to view emotions as both innate and embodied, especially for categories that are presumed to be biologically “basic.” The perception of innateness, moreover, correlated with the belief that emotions are expressed in both the face and the body. We hypothesized that when people view emotions as embodied, they also perceive them as innate precisely because they interpret embodiment as evidence for innateness. Experiment 2 tests this possibility.

Experiment 2: Embodiment Promotes the Perception of Emotions as Innate

To determine whether the perception of emotions as innate is caused by the belief that they are expressed in the material body, Experiment 2 manipulated beliefs about the embodiment of emotions. Embodiment, here, was operationalized by the instantiation of an emotion in a particular brain area, corresponding to a discrete piece of matter. Of interest is whether describing an emotion as embodied will promote its view as innate.

Embodiment was manipulated across two groups of participants. Both groups were presented with the same list of 20 emotion words from Experiment 1. The “material” group was informed that these emotions each are localized in a distinct region of the brain. Participants in the “immaterial” condition were informed that these emotions are devoid of a brain localization. If people believe that embodied emotions are innate, then the “innateness” rating should be higher in the “material” relative to the “immaterial” condition.