Gervais & Fessler (G&F) characterize contempt as a “sentiment” to account for inconsistent findings on contempt as a basic emotion. They claim that constructionism, an alternative to basic emotions approaches, cannot account for contempt findings. We suggest that “sentiments” sound a lot like basic emotions as natural kinds, a theoretical approach that has been heavily criticized. Moreover, G&F misunderstand constructionism, which parsimoniously accounts for the messy literature on contempt.

Despite claiming that contempt is not a basic emotion, G&F use basic emotion theory terms (e.g., Ekman & Cordaro Reference Ekman and Cordaro2011; Izard Reference Izard2011; Panksepp Reference Panksepp2011) to define sentiments: “As with emotions, each sentiment likely has a distinct evolutionary history and taxonomic distribution […], as well as partially dissociable neural bases” (sect. 4.3, para. 1). Similarly, when they suggest “a provisional set of sentiments – social attitude dimensions, corresponding to distinct social-relational affordances – whose states potentiate unique constellations of emotions” (sect. 4.3, para. 3).

As in basic emotion approaches, G&F define contempt as a natural kind. A natural kind is a non-arbitrary collection of natural phenomena or properties existing independent of human observation (e.g., chemical elements; Mill Reference Mill1884). However, growing evidence suggests that emotions are not natural kinds. Emotion categories have neither consistent nor specific outcomes making them biologically distinct from one another (Barrett Reference Barrett2006a; Kreibig Reference Kreibig2010; Lindquist et al. Reference Lindquist, Wager, Kober, Bliss-Moreau and Barrett2012; Mauss & Robinson Reference Mauss and Robinson2009; Vytal & Hamann Reference Vytal and Hamann2010; Wager et al. Reference Wager, Kang, Johnson, Nichols, Satpute and Barrett2015). Contempt is no exception.

Contempt lacks consistency and specificity. People fail to consistently identify facial expressions as contempt; the label “contempt” is used to categorize posed facial portrayals of contempt at or below chance (Izard & Haynes Reference Izard and Haynes1988; Wagner Reference Wagner2000). Instead, facial muscle movements are not specific to contempt – people categorize them as disgust (Haidt & Keltner Reference Haidt and Keltner1999; Russell Reference Russell1991d; Russell et al. Reference Russell, Suzuki and Ishida1993) or annoyance (Alvarado & Jameson Reference Alvarado and Jameson1996) depending on context. Although some studies find that people associate a unilateral lip-curl with contempt (Matsumoto & Ekman Reference Matsumoto and Ekman2004), this occurs only in forced choice designs involving direct comparisons between prescribed categories. In fact, prototypically contemptuous facial expressions are not universally perceived as contemptuous (Heuer et al. Reference Heuer, Lange, Isaac, Rinck and Becker2010; Russell Reference Russell1991d). Additionally, predicted correspondences between specific antecedent events (e.g., violations of community norms) and contempt are not upheld (Rozin et al. Reference Rozin, Lowery, Imada and Haidt1999). The evidence for the existence of contempt as a natural kind is so in question that even proponents of natural kind views of emotions admit contempt is less likely to qualify as such (Haidt & Graham Reference Haidt and Graham2016; Rosenberg & Ekman Reference Rosenberg and Ekman1995).

If contempt is not a natural kind, then what is it? We suggest it is a constructed experience, like all emotions and mental states (Barrett Reference Barrett2009; Clore & Ortony Reference Clore and Ortony2013; Cunningham et al. Reference Cunningham, Dunfield and Stillman2013; Lindquist Reference Lindquist2013; Russell Reference Russell2003). Rather than arising from discrete mechanisms with domain-specific functions, constructionism suggests that distinct mental states are the emergent product of domain-general ingredients, including core affect and conceptual knowledge (Barrett Reference Barrett2013; Cameron et al. Reference Cameron, Lindquist and Gray2015; Lindquist Reference Lindquist2013; Russell Reference Russell2003). These ingredients combine in different ways to produce different mental products. For example, just as the same combination of ingredients can create a sugary cake or a savory biscuit, different combinations of core affect and conceptual knowledge can construct different emotions.

G&F dismiss constructionism as a theoretical framework for understanding contempt, but their argument is based on a misunderstanding of constructionism. The authors wrongly claim that a constructionist view predicts that “a word such as ‘contempt’ is necessary to anchor … features categorized as a specific emotion” (sect. 3.3, para. 2), pointing to evidence in which people experience contempt without linguistic prompts (Fridhandler & Averill Reference Fridhandler, Averill and Averill1982; Matsumoto & Ekman Reference Matsumoto and Ekman2004; Rozin et al. Reference Rozin, Lowery, Imada and Haidt1999). However, this is a misunderstanding; constructionism hypothesizes that most instances of emotion are experienced in the absence of an explicit linguistic prompt – little of daily life involves explicitly labeling experiences. Instead, a constructionist view predicts that language plays a covert role in emotion insofar as it implicitly helps people acquire, organize, and use emotion concept knowledge during online categorization (Lindquist & Gendron Reference Lindquist and Gendron2013; Lindquist et al. Reference Lindquist, MacCormack and Shablack2015a; Reference Lindquist, Satpute and Gendron2015b).

Constructionism predicts that people experience a specific emotion concept (e.g., contempt) when they draw on their rich cache of conceptual knowledge about that category. Conceptual knowledge of “contempt” consists of past internal feelings in situations categorized as contempt, as well as past motor representations of behaviors, sensory representations of situations, and cultural knowledge about what it means to experience contempt. These diverse sensorimotor representations are partly united by the word contempt because contempt is not a natural kind with strong perceptual regularities uniting members of the category (Lindquist et al. Reference Lindquist, MacCormack and Shablack2015a; Reference Lindquist, Satpute and Gendron2015b). Unbeknownst to human observers, words cohere this category information and facilitate its accessibility during online perception (Lindquist et al. Reference Lindquist, MacCormack and Shablack2015a; Reference Lindquist, Satpute and Gendron2015b; Lupyan Reference Lupyan2012; Vigliocco et al. Reference Vigliocco, Meteyard, Andrews and Kousta2009). People can still experience contempt in the absence of explicit emotion words, but emotions are disrupted when implicit access to emotion words is impaired (Gendron et al. Reference Gendron, Lindquist, Barsalou and Barrett2012; Lindquist et al. Reference Lindquist, Barrett, Bliss-Moreau and Russell2006; Reference Lindquist, Gendron, Barrett and Dickerson2014).

In sum, constructionism accounts for the “messy” data on contempt more parsimoniously than the authors' model, suggesting domain-general processes underlie emotion rather than many discrete, local mechanisms. This converges with neuroscientific evidence suggesting domain-general neural networks are implicated in many different mental states besides the emotional (Barrett & Satpute Reference Barrett and Satpute2013; Cushman & Young Reference Cushman and Young2011; Lindquist & Barrett Reference Lindquist and Barrett2012; Shenhav & Greene Reference Shenhav and Greene2010). Additionally, constructionism generates novel predictions about contempt: People with more fine-grained conceptual knowledge about emotions (Lindquist & Barrett Reference Lindquist, Barrett, Lewis, Haviland-Jones and Barrett2008) may be more likely to construct contempt as opposed to anger or disgust out of diffuse core affect. This suggests that, contrary to G&F's claims, the experience of contempt may vary across persons within the same situation and within the same person across situations.

If the authors make a mistake, it is placing too much emphasis on the meaning of words. The lack of a verbal label in an experiment does not invalidate constructionism. And a new label of contempt – as a “sentiment” – does not make this argument different from old natural kinds claims about emotions. Words have power, but we should not confuse our labels with the essence underneath, especially when that essence may not exist.