On the Invalidity of Neta and Kim's Argument That Surprise is Always Valenced

In a recent challenge to theories of Basic Emotions (BET), Ortony (2022) proposed that being valenced (i.e., being either affectively positive or negative, as opposed to affect-free) is one of three minimal requirements for something to be an emotion and that in view of this the status of surprise (a stock candidate for a basic emotion in BET) as an emotion, let alone a basic emotion, is questionable. Responding in the same journal as Ortony's article, Perspectives on Psychological Science, Neta and Kim (2022) (hereafter, N&K), while accepting the necessity of valence for emotion, contended that surprise is always valenced, and therefore is an emotion and quite possibly a basic emotion. In this brief article, we devote but little attention to a substantive discussion of the main issues (for which, see the original articles), focusing instead on what we see as serious logical flaws in N&K's arguments.

We believe that a weakness of many emotion theories (e.g., Ekman, 1992; Frijda, 1986; Izard, 2007; Plutchik, 1962; Tomkins, 1962) is their inclusion of states such as interest and surprise as emotions (indeed, as basic emotions). But the failure of such states to meet one of the three proposed requirements (Ortony, 2022) for a state to be an emotion—that the state be necessarily, intrinsically, valenced—raises the possibility that they should not be considered emotions at all. And this is the case with the experience of surprise. Most of us have at one time or another experienced surprise while being affectively neutral. But if one can be surprised without any experienced affect then surprise is not necessarily valenced, and this leads to serious questions about the justifiability of treating surprise as an emotion. Like many others, N&K find the suggestion that surprise might not be an emotion rather unpalatable, and so they attempt to defend its status as an emotion by maintaining that there is no such thing as affect-free surprise and that in fact surprise is always valenced. In order to reach this conclusion, they propose the novel idea that surprise is an emotion whose intrinsic valence, rather than being fixed as it is for emotions in general, is variable, being sometimes positive and sometimes negative.

N&K's first argument in support of their position relies on definitions: affect-free surprise is, by their (implicit) definition, not really surprise at all. In essence they are stipulating that any experience that one might take to be affect-free surprise cannot, by definition, be surprise. For N&K, apparent cases of affect-free surprise are really cases of unexpectedness, which they view, correctly in our opinion, as a purely cognitive state as opposed to an emotion. This line of reasoning first emerges early on when N&K characterize the elicitation of surprise by noting that “. . . when a prediction is violated, one's response—and its associated emotional value—depends on the magnitude of the prediction error and its direction…” (p. 2, italics added). Never mind that surprise often occurs in the absence of any prediction (Ortony & Partridge, 1987), the real problem is that by inserting the “associated emotional value” into their characterization of surprise N&K are assuming precisely what is at stake, thereby committing the logical fallacy of petitio principii (begging the question). This gratuitous incorporation of emotional value into the definition then establishes the foundation of N&K's main point, which is that surprise is always valenced, a conclusion that could not have been reached had N&K not incorporated it into their implicit definition of surprise in the first place.

From our perspective, N&K's insistence on a sharp distinction between surprise and unexpectedness makes a qualitative distinction between two equivalent constructs. In contrast, N&K seem to think that labeling affect-free surprise as “unexpectedness” validates their position. In characterizing the framework of their argument, they quote Ortony approvingly: “surprise is always the result of the registration of a discrepancy between what is encountered and some reference point” (N&K, p. 2). This is basically the same idea as theirs, but without any assumptions about “associated emotional value” and the “direction” of a discrepancy. In fact there is nothing new about the idea that the feeling of surprise/unexpectedness and affect are independent, although this independence has usually been cast in different terms. Specifically, it is generally recognized that the mismatch triggered by the processing of unexpected events results in nonspecific arousal, which is itself affect-free (e.g., Grossberg, personal communication, August 9, 2023; Grossberg & Schmajuk, 1987, p. 209; Mandler, 1984), and in dimensional models of emotion, arousal and valence are always taken to be orthogonal (e.g., Russell, 1980; and see Rubin & Talarico, 2009 for a review). But N&K consider accounts of surprise based only on discrepancy alone to be inadequate because such accounts do not require that the experiencer attach subjective importance to the surprising fact or event (i.e., the object of the surprise). For N&K, if the object is of no subjective importance (as might be the case, for example, on hearing that China has only one time zone), the state that the experiencer is in “might better be referred to as unexpectedness” (p. 2). We, on the other hand, along with many other native speakers of English, call it “surprise” because we believe that experiencing something as unexpected is tantamount to experiencing it as surprising, and vice versa. N&K's position implies that despite our being native speakers of English who study emotions and language, we are routinely misusing the term “surprise” (which itself might be viewed as an example of something surprising but of no importance!). In fact, we suspect that most native speakers of English recognize affect-free surprises as genuine surprises. N&K are, of course, free to restrict their own use of the word “surprise” to exclude affect-free surprise, but doing so establishes nothing about the valence status of surprise for anyone else. We are reminded of Humpty Dumpty's pronouncement in Alice in Wonderland: “When I use a word . . . it means just what I choose it to mean—neither more nor less!” But this debate is not about N&K's linguistic idiosyncrasies, it is about the nature of experience as commonly described. N&K's opinion that the experience of surprise is always, necessarily, valenced—but ambiguous as to whether specifically positive or specifically negative—is totally reliant on their unimpeachable rejection of any possible counter-examples as legitimate instances of surprise. And why do they reject putative counter-examples? Because such counter-examples are not valenced! Petitio principii, again.

N&K complain that Ortony “misrepresents the breadth of literature showing that surprise is in fact valenced” (N&K, p. 2). But their position that surprise is never valence-free cannot be strengthened by citing, as they do, numerous cases of surprise that are valenced. The existence of valenced cases (in the sense that people judge some unexpected event to be valenced) is not in dispute; nobody doubts that people often experience (what are often called) pleasant surprises and unpleasant surprises, and so in this sense, of course, surprise is often valenced. But often valenced does not mean always valenced, nor does it mean intrinsically valenced. If surprise is an emotion, then, as N&K grant, it must be always and intrinsically valenced. Our view that people can be surprised while not being in a valenced state is unaffected by N&K's litany of examples of people being surprised while also having a valenced response to the object of their surprise. What N&K would need to show is that non-valenced (affect-free) surprise never occurs, or that apparent cases of affect-laden surprise can occur in the absence of any other possible locus of affect. The examples of apparently valenced surprise that N&K cite come nowhere close to supporting such conclusions.

Much of N&K's attempt to bolster their case is based on empirical work on people's interpretations of the facial expressions of others, research whose irrelevance N&K almost acknowledge when they say that although such interpretations “[do] not necessarily capture the specificity of one's emotion experience . . . they are not entirely independent constructs” (p. 2). We, however, believe that in fact they are almost entirely independent. Findings about how people interpret the faces of others (surprised or not) provide no assurances about the emotions of the person whose facial expression is being judged (Russell, 2017). People can be surprised with no concomitant change in their facial expressions, and vice versa. To reiterate a point made earlier, the question at hand is about experience—about the nature of the state that a person is in when encountering an unexpected, but (to that person) unimportant, event, not about other people's judgments about such a person. But even if we were to allow the potential relevance of such findings, we would still have serious misgivings because of N&K's frequent reliance on the assumption that surprise is inevitably expressed in a recognizable facial signal. They seem unaware of the fact that the general idea that faces “express” emotions is very problematic (e.g., Fridlund, 1994), and unaware of the fact that the existence of a reliable relationship between experiences of surprise and facial expressions has been challenged on empirical grounds (e.g., Durán & Fernández-Dols, 2021; Reisenzein et al., 2006). In the case of surprise, what can (sometimes) be seen in the face is most parsimoniously interpreted not as “expressing” an emotion (and certainly not as signaling to a conspecific) but as part of the process of perception: upon encountering an unexpected event, one spontaneously widens the eyes to better take in information.

The core of our disagreement with N&K is their assumption that valence is an inherent quality of surprise. Our view is that surprise per se is never valenced and that when it appears to be valenced it is because there is a co-occurring positive or negative feeling whose (formal) object is the same as that of the surprise itself. It is in this co-occurring emotional state that the affect inheres, not in the surprise. What surprise (aka unexpectedness) does do is to modulate the intensity of emotions (as discussed in detail in Ortony et al. (1922, pp. 78–80)). When there is surprise with no co-occurring emotion, as typically happens when the experiencer attaches no subjective important to the object of the surprise (i.e., is emotionally indifferent), there is no affect at all. Accordingly, what makes a surprising experience pleasant or unpleasant cannot be the surprise; rather, it is the valence of a co-occurring emotion. So, for example, x is glossed as a pleasant surprise when one is both surprised that x and happy that x; y is an unpleasant surprise when one is both surprised that y and distressed that y. But z is simply a surprise, neither a pleasant nor an unpleasant surprise, when one is surprised that z with no co-occurring emotion about z.

Given their commitment to the idea that surprise is always valenced, N&K embrace the curious, novel, and we believe unparsimonious, position that surprise is ambiguously, valenced—sometimes positive and sometimes negative, always one or the other, never neither. But this means that a specific valence is not a fixed feature of surprise and is not determined by the nature of surprise per se. In other words, the valence of surprise is determined by something other than the surprise itself and therefore cannot be constitutive of it. As already indicated, this is at odds with what is known about emotions in general, which is that their valence is fixed, and is intrinsic to and constitutive of them. Unlike N&K's model of surprise, the valence of each emotion type is always exclusively positive or exclusively negative—the feeling of happiness per se is never negative, and of sadness, never positive. The proposal that surprise is ambiguously valenced—maybe positive, maybe negative—needlessly attributes to surprise a property that emotions in general patently lack (a fact that N&K acknowledge). We, along with most other emotion researchers, believe that the valence of every emotion type is intrinsic to and constitutive of that emotion, being either always and only positive, or always and only negative. This is not the case for N&K's version of surprise.

N&K's section “The plausibility of an initial negativity” (p. 4) of surprise further illustrates our claim that the valence of surprise depends on an accompanying mental process. N&K surmise that an initial negativity of surprise occurs when one is startled because the state of being startled is intrinsically negative. We find it curious that while N&K are so ready to make a sharp and implausible distinction between surprise (which they consider to be an emotion) and (the registration of) unexpectedness (which they rightly, in our view, consider to be a non-emotional, cognitive, state), they are apparently unwilling to acknowledge the distinction between surprise (a mental state) and startle (a reflex). Of course, the word “startle” is often used to convey extreme surprise, but in its technical sense startle is “a characteristic sequence of muscular responses” (Davis, 1984). To be sure, muscular responses might be (un)pleasant, but they are not in and of themselves affective. Any affect results from analyzing the cause of the startle (or possibly its consequence) does not reside in the startle itself. Startle isn’t intentional (it isn’t about anything), surprise is. Thus, while it is of course true that one might dislike being startled just as one might dislike being hungry, that doesn’t make being startled or being hungry emotions (Clore et al., 1987).

Whereas N&K are unambiguous in their commitment to the idea that surprise is an emotion, they are more circumspect when it comes to the question of whether it is a basic emotion, saying, in conclusion, that given the wealth of evidence that they have provided to establish that surprise is an emotion, “. . . if it is an emotion, it very well could be a basic emotion” (p. 6). Ironically, this remark illustrates one of Ortony’s (2022) several criticisms of BET, which is that the criteria for membership of the basic emotions club are unacceptably vague. In support of this claim, Ortony cited the wide array of candidates that have been proposed, including states such as interest and surprise that are arguably not emotions at all, as well as the fact that from time to time new candidates are added (e.g., Cowen & Keltner, 2017; Ekman & Cordaro, 2011). The sentiment that if surprise is an emotion, maybe it's a basic emotion and maybe it isn’t, is a paradigmatic demonstration of the referential indeterminacy of the basic emotion construct, and indeed, of the emotion construct too.

In conclusion: We believe, as do N&K, that for a state to be an emotion it must always be valenced, but we believe that their notion of “ambiguous valence” reduces to a lack of any intrinsic valence at all and that, indeed, surprise per se is not intrinsically valenced, and therefore not an emotion. Renaming affect-free surprise as “unexpectedness” solves nothing. We also share with them the belief that subjective importance is necessary for emotion, but we think that it is perfectly possible to be surprised by something that is of no subjective importance. For example, lemons float, but limes sink, surprising perhaps, but most of us don’t care. Ergo, surprise is not an emotion. Our view is that N&K's need to ascribe a unique property of valence ambiguity to surprise is rendered unnecessary (thank you Occam's razor; Feldman, 2016) because what makes a so-called positive surprise positive or a negative surprise negative is a co-occurring positive or negative emotion, which is where the locus of the affect lies. Surprise remains affect-free—a cognitive state of unexpectedness, just as N&K say! Logically speaking, the question of how people feel when they experience surprise cannot be answered by defining away affect-free surprise. Nor do numerous examples of affect-laden surprise refute the existence of affect-free surprise.