Following Barrett, I contend that our sensations and beliefs play an important role in our emotional self-attributions. In contrast to Barrett, however, I propose that we focus on discovery, in addition to construction, while also recognising the role that self-interpretation plays in our emotional self-attributions. This view can be expressed as follows:

Partial First-Person Authority (FPA) of Emotions: Our first-person avowals about what emotions we are currently undergoing are grounded in our introspected sensations and beliefs. In order to gain awareness of our emotions we need to self-interpret those (introspected) sensations and beliefs, as well as other propositional attitudes. The FPA we have of our emotions is partially asymmetrical, but not wholly. Self-interpretation is involved partially in the process. We have FPA with respect to our sensations and propositional attitudes, but knowledge of emotions requires us to self-interpret those mental states.

According to this view, there is a partial epistemological asymmetry associated with our emotion self-attributions. We have ‘direct access’ to our sensations and beliefs, since we do not need to draw upon any behavioural evidence to know them, like other people need to do (cf. Bortolotti 2009, p. 210).Footnote 14 Evidence drawing is not wholly absent however, as we still need to self-interpret our sensations and beliefs to know our emotions. I call this partial FPA because introspective processes alone will not give you knowledge of your emotions: self-interpretation is also required.

I have not, to be sure, accounted for how we do authoritatively know our sensations and beliefs in this paper, and so remain neutral here about how best to do that. Even though I have been critical of TM and neo-expressionism, as ways of explaining FPA of emotions, it remains possible that either of these views can explain how we can know our sensations or beliefs in a non-interpretative way. The acquaintance view or inner sense view, further, may even be able to explain how we have FPA of our sensations or propositional attitudes (e.g., our beliefs and intentions). FPA, then, need not be explained uniformly, and it need not be an all or nothing phenomenon: it can be partial. An acquaintance theorist who thinks that we can come to know our sensations by acquaintance could accept that we know our emotions by self-interpreting those sensations.

As I mentioned in the previous section, focusing on self-interpretation, in addition to construction, allows us to say that we sometimes do discover the emotions we are undergoing; and it also suggests the possibility of certain types of errors. Where there is interpretation, there is also the possibility of misinterpretation. It is important to account for this phenomenon since emotions have a dispositional component. Focusing predominantly on construction, as Barrett does, makes it difficult to account for cases where a person comes to know that they are undergoing an emotion on the basis of another person’s testimonial evidence. If I come to know that I am jealous for example, because someone informs me that I am acting jealously, then the emotion has already existed for some time, as it was my jealous behaviour that caused the person to believe that I am jealous. Such a causal sequence would seem to suggest that the emotion was present before I became aware of it, so discovery seems appropriate here.

What evidence or justification is there for the partial FPA of emotions view? First, I think the view makes good sense of some of the psychological cases we looked at earlier, and also the cases that Barrett mentions. It is important, however, that I clarify how I think this evidence supports the partial FPA view of emotions, as it may seem like I am simply advancing the following two uncontroversial claims:

(1)

Knowledge of our emotions is partially based on self-interpretation; and

(2)

That self-interpretation is prone to error, i.e., to misinterpretation.

Suppose an inner sense theorist, who thinks that we can introspect our emotions via an internal scanner, considers these two theses. They may state that while we sometimes use self-interpretation to know our emotions, and that we sometimes make misinterpretations, this is not enough to undermine the thesis that we can know our emotions in a purely introspective way (i.e., via an internal scanner). Such a theorist could, following Carruthers’ (2011) terminology, be a ‘dual-method’ theorist who holds that we can know our emotions via self-interpretation, and we can also introspect them.

In my view, the kinds of data that we looked at in Sections 4 and 5, put pressure on such a view. An important question can be raised here: why think that we possess an introspective method in addition to an inferential method for discerning knowledge of our emotions, when a (partially) interpretative method alone (hereafter, the ‘single-method’) will suffice? If we can know our emotions by self-interpretation, then, it might be that positing the existence of a dual-method would be superfluous, given that a single-method appears to be sufficient.Footnote 15

The partial FPA of emotions view should be seen as a challenge to the dual method position with respect to emotions. If correct, then introspection cannot wholly give us knowledge of our emotions—we always need to partially rely on self-interpretation. If I am right that self-interpretation is always required, then the neo-expressivist view, the transparency method, and the inner sense view, when applied to emotions will be undermined. These views stand in contrast to the claim that we must always partially interpret our sensations to know what emotion we have.

One thing that is notable, with respect to the FPA view of emotions, is that emotional self-attributions appear to be correlated with subjective feelings. Someone who is nervous may also feel their stomach flutter. Further, someone who is presented with sensations, such as a feeling in their stomach, could interpret the feeling as romantic love, even though they are not really undergoing the emotion. This looks like a paradigmatic self-interpretation error, which may be illustrative of the way we learn our emotions. The cases of error are not enlightening because I think introspective views are committed to a kind of infallibility thesis. The cases of error are interesting because I think they help to reveal the normal way we come to know our emotions. When one is on a date, and one becomes aware of a sensation in one’s stomach it could be quite natural for one to interpret that feeling as romantic attraction to their date, even if one was not, as a matter of fact, romantically attracted to their date, but say sick. If self-interpretation is the normal way that we come to know our emotions, we would expect to make such mistakes when the context and other cues are misleading.

To say that there is a connection between sensations and emotions is not a new observation, as shown in the works of poets, as Damasio (2018, p. 106) points out, and sometimes portrayed in fiction. For example, in a famous episode of the Simpsons, ‘Sweet Seymour Skinner's Baadasssss Song’ (Oakley et al. 1994), Bart Simpson is sitting at the breakfast table wondering why he is not jumping for joy after finding out that his school principal, and rival, Seymour Skinner was fired because of an incident he was the cause of. He claims that all he has is a: “weird hot feeling in the back of [his] head.” His sister Lisa, who is also at the table says, “That’s guilt, you feel guilty because your stunt wound up costing a man his job.” Bart replies, “Yeah I guess it is guilt.” The viewer momentary accepts that Bart feels guilt, until the very next moment, where we see a small spider on the back of Bart’s neck, that is biting him. This raises a question for the viewer about whether Bart really does feel guilt. But it also raises the question about the route Bart has followed to acquire knowledge of his guilt. It may be that this fictional account mirrors a real psychological process where humans self-interpret their sensations to know their emotions.

Why think that such an account is true of human psychology? In two recent studies by Nummenmaa et al. (2014, 2018) it was shown that people locate different areas of the body as feeling a certain way when they undergo basic emotions (e.g., anger, fear, disgust) and non-basic emotions (e.g., anxiety, love, depression, contempt). In the 2014 study, they asked Western Europeans (Finish and Swedish), and East Asian (Taiwanese) participants to point to the specific region on a map of the body where they experienced feelings when they were undergoing certain emotions. Participants used a computer based self-report method called ‘emBODY’, which presented them with blank silhouettes of a body. After viewing stories, movies or facial expressions, they were instructed to note the emotions they were undergoing, and also colour in the parts of the silhouettes that they felt represented the places in the body that those feelings were occurring. Participants who felt anger, for example, noted the presence of associated feelings coming from the head, stomach, and shoulder area. Happiness, on the other hand, was reported as having a feeling that was present in the whole body. And envy was associated with feelings that were in the head region and upper chest. In the follow up study, Nummenmaa et al. (2018), tested for an even greater number of emotions, and other states (100 in total). In addition to anger and fear, they looked at guilt, sadness, and disgust, amongst many others. As with their earlier study, it was found that various emotional experiences were paired with different feelings in the body, indicated by the subjects in similar regions on the bodily maps. Guilt tended to be felt higher up on the body, whereas nervousness around the middle of the body. The authors concluded that these bodily sensations ‘could be at the core of the emotional experience’ (2014, 650).

These results are relevant to our current discussion because they suggest that our awareness of our emotional experiences involve bodily sensations. This is just what one would expect if self-knowledge of one’s emotions required one to self-interpret one’s bodily sensations. And since the experimenters also found that certain areas of the body were active during various emotion experiences, this would also seem to suggest that we might sometimes mistake one emotion for another since certain emotions may have similar bodily feelings. I may for example see that I am at a doctor’s office and interpret a certain feeling as anxiety; but if I see that I am in a crowded lunch bar, I may self-interpret the same feeling as hunger. The best explanation of all of this, in my view, is that self-interpretation of our sensations is our normal route to self-knowledge of emotions.

I am not the first to focus on self-interpretation with respect to the way we know our mental states. Carruthers (2011) and Cassam (2014) both argue that self-knowledge of most of our mental states is self-interpretative in nature. They both deny that we have any introspective way of knowing our propositional attitudes, for example.Footnote 16 Carruthers, for instance, argues that in order to know whether I believe there is going to be a third world war, I would need to draw upon sensory evidence. He supports this claim by looking at the data from split-brain studies, choice blindness experiments, and priming studies. Problematically for Carruthers’ view, his interpretation of these data is controversial (see Andreotta 2021b) and not widely accepted; and further, there is notable lack of sensory data present when we acquire knowledge of our beliefs, as well as other propositional attitudes. Nummenmaa et al. (2014, 2018), for instance, do not note of any sensations that are paired with our experiences of believing or intending to do something. Even though some of our propositional attitudes are associated with certain feelings, it is implausible to suggest that we always need to interpret our sensations to know these states. One critic of Carruthers’ work, for example, Georges Rey, has pointed out, ‘[d]esire, wonder, doubt, pretence, curiosity, for example, don’t seem to be linked to any specific sensations’ (2013, p. 274). Self-interpretation as the only way we can know our propositional attitudes, thus, seems implausible. If we focus on partial self-interpretation and limit this approach to our knowledge of our emotions, on the other hand, then we are left with a much more plausible and empirically supportable position.

Let us take an example of Cassam’s, involving emotions, to see why this is the case. In a discussion of his inferentialist position, Cassam examines a case of self-knowledge involving the emotion love from Marcel Proust’s Remembrance of Things Past. The case involves the character Marcel’s realisation that he loves Albertine, which comes about after her absence causes him to experience anguish (or suffering). Cassam suggests that Marcel can come to know that he loves Albertine by focusing on the suffering he is undergoing. Cassam suggests that the suffering is thus the basis of Marcel’s self-knowledge but stresses that it does not itself constitute Marcel’s love. This is because love might only be one of several possible explanations of Marcel’s suffering. Cassam explains further:

The inference is mediated by an interpretation of his suffering that is grounded in his understanding of the relationship between this kind of suffering and romantic love. His route to self-knowledge here is inference, whereas the basis of his self-knowledge is suffering (2014, p. 181 emphasis in original).

Cassam as an inferentialist would deny, then, that we can know our emotions, such as love, by an introspective, non-interpretative method.

One feature of this self-knowledge process that Cassam does not spend much time discussing, however, is the phenomenological set of sensations that make up the anguish or suffering that Marcel can draw upon. In terms of FPA, this is significant, because the suffering would involve felt experiences which can be introspected. Marcel can, further, attend to his introspected memories, beliefs, and intentions. Introspecting these mental states can give him an important epistemic advantage over other people trying to determine what emotion he is undergoing, even if self-interpretation is involved in part of the process. So to say, as Cassam does, that Marcel must rely on self-interpretation to know that he is in love is partially true; but not the whole story. Marcel can ground his self-attribution of love with his introspected experience of suffering, which is epistemically significant.

The FPA view of emotions, then, can be thought of as a middle ground position, between those who think we can have direct introspective access to our emotions on the one hand and those such as Cassam, who thinks that it is all self-interpretation, on the other. We do not need to abandon the idea of asymmetry completely, as Cassam seems to suggest, even if we do need to recognise that self-interpretation is involved in the self-attribution process. This may require us to abandon the kinds of epistemic confidence we feel we have when it comes to emotion self-attributions, compared to the confidence we have when we self-attribute a belief or a sensation; but again it is important to recognise that a kind of epistemic advantage is still there. Even in cases where we might misattribute emotions, we may get other mental states right. Consider Damasio, again, who says ‘precise feelings that comparable situations evoke may well be tuned by cultures’ (2018, p. 109). In one example, he suggests that the nervousness of students before an exam can be experienced by German students as butterflies in the stomach, whereas, in Chinese students it can be experienced as a headache. Both students may have FPA with respect to the sensations they are experiencing, but may interpret those sensations in very different ways, given the context, culture, background beliefs and so on. Some students may interpret those sensations ‘correctly’ and some may not. We need not say, along with Cassam, that it is ‘inferential all the way’ down (2014, p.161). In the case of emotions, our FPA is partial because some of the evidence we draw on to know them is available only to us; and it is stronger than the evidence we draw on regarding other people’s emotions, which the Cassam-style view denies.

One may object to the partial FPA of emotions view by saying that self-interpretation occurs ‘internally’, and thus is an introspective process; so we can still have FPA about our emotions. In a recent paper on the introspection of emotions, De Vlieger and Giustina (2022) give an account of how we introspect our emotions. They advance a three-stage model, which begins with primitive introspection. This involves becoming aware of the non-classificatory ‘information about the phenomenology of the introspected experience’ (2022, p. 561). The second stage, reflective introspection, is relevant to our focus here. They claim, ‘It consists in classifying the introspected experience as an instance of a known experience type’ (2022, p. 563). This coheres with what I have suggested here: the partial FPA of emotions view suggests that we need to self-interpret our sensations to know them. De Vlieger and Giustina add that ‘At this stage, the subject gives or attempts an interpretation of the introspected experience: they try to figure out what kind of experience’ (2022, p. 563 emphasis added) it is.

Although I agree with De Vlieger and Giustina that we need to interpret our sensations (as well as other mental states such as memories and beliefs) to know our emotions, I disagree that this process is best thought of as ‘reflective introspection’. Let us consider one of their examples to see why this is the case. Consider Caroline, who feels her heart pumping, notices a smile coming across her face, and experiences an urge to jump up into the air. According to De Vlieger and Giustina, Caroline can know that she feels the emotion joy by self-interpreting these experiences—what they refer to as reflective introspection.Footnote 17 Now, while Caroline clearly has epistemically significant evidence to draw on to self-ascribe her emotion—namely, her felt sensations—she still needs to do interpretive work, just like other people who are attempting to attribute an emotion to her. So, I would not call this interpretative process ‘introspective’ simply because it occurs internally. Caroline’s epistemic authority is thus only partial.