Skip to main content

Jenefer Robinson

University of Cincinnati

I would like to start my responses with grateful acknowledgments to all those who responded to the invitation by ‘Debates in Aesthetics’ to comment on my work, and especially to those who submitted the very accomplished papers that were accepted and which I will be discussing in this commentary. I am also very grateful to the editors of ‘Debates in Aesthetics’, Ryan Doran and Shelby Moser, for inviting the critics to write about my work and for giving me the opportunity to respond. It has been both fun and stimulating to grapple with these papers. I think I have learned a lot. I have certainly been pressed to defend my positions on various issues and to justify some of the more outrageous claims I have made.


Vanessa Brassey[1] gives a rich and thoughtful response to a paper of mine called ‘The Missing Person Found: Expressing Emotions in Pictures’[2], in which I develop what she calls a ‘personalist’ theory of pictorial expression, in contrast to the ‘impersonal’ theory defended by Dominic Lopes in ‘Sight and Sensibility’.[3] Brassey’s paper is full of insights and I cannot address all of them here, so I will focus on three points which are the most interesting and important from my point of view: (1) the difference between ‘pictorial expression’, meaning expression by (whole) pictures, and ‘skilled depictions of emotions’, including expression by figures, scenes and designs; (2) an ambiguity in the notion of ‘design expression’; and (3) the idea of an implied painter as the locus of expression.

(1) Brassey characterises my view of pictorial expression as a ‘personalist’ view, meaning that for me ‘pictorial expression’ is always transitive, i.e., pictorial expression is always an expression of someone’s actual mental state or disposition, whereas according to Lopes’ ‘impersonalist’ view, expression in pictures takes the form of what he calls figure, scene, or design expression. Figure expression is “an expression that is wholly attributable to a depicted person or persons”[4]; scene expression is “an expression that is attributable at least in part to a depicted scene and is not wholly attributable to any depicted person”[5]; and design expression is “an expression that is wholly attributable to a picture’s design or surface and not to any figure or scene it depicts”.[6] In contrast, I argue that although figure and scene expression can contribute to the expression in a (whole) picture, they are not themselves expressions of emotion, but merely, in Brassey’s phrase, “skilled depictions” of figures and scenes as (looking) sad, anxious, light-hearted etc.[7] A genuine pictorial expression of emotion, on the other hand, is a genuine expression of emotion in a person, normally the painter.

Brassey comments that Lopes seems to think that I “[assume] that viewers sustain an FE figure expression line of thought in the absence of figures”, and that once we drop this assumption—which we should, on Lopes’ view—we can “adopt an impersonal theory”, such as Lopes endorses.[8] It is not entirely clear to me what Brassey means by a “figure expression line of thought”[9], but I think she is accusing Lopes of interpreting me ‘qua personalist’, to be taking ‘figure expression’, i.e. the expression of emotion by depicted figures, as in Daumier’s ‘Fatherly Discipline’, to be “the paradigm case of pictorial expression”[10], and that’s why for me there is a missing person problem when it comes to scene expression (and presumably design expression too). Figure expression can be transitive insofar as a depicted character in a painting wears an expression of emotion that reflects the emotion the character is depicted as actually experiencing: if, in Daumier’s ‘Fatherly Discipline’, the father is not expressing his own actual state of irritation but simply putting on an irritated face, the painting loses its point.

But Brassey perspicaciously points out that pictorial expression on my view is not modeled on figure expression. On the contrary: figure, scene, and design expression are all what I call ‘expressive elements’ in a picture: they contribute to what the picture expresses as a whole, which is what I mean by ‘pictorial expression’. As she writes, “Lopes’ error is to conflate his categories of expression”—figure, scene, and design expression—“with a picture’s expression of an emotion E”.[11] Pictorial expression is indeed transitive on my view but not because it is a version of figure expression. As she rightly points out, “personalists do not argue that simply seeing depicted figures expressing emotions is sufficient for seeing a picture as an act of expression”.[12] She notes that a row of smiling faces generated by the Flickr algorithm is a good example of what’s wrong with this claim. What a picture expresses as a whole is a function of what the figures in the picture express (in Lopes’ sense), what the scene or scenes in the picture express (in his sense) and, above all, how the depicted figures and the depicted scenes are unified into an expressive whole by the overall design or composition of the picture. In the Kokoschka self-portrait that I discuss, for example, there is a semi-realistic depiction of Kokoschka’s anxious facial expression, posture and gesture (holding his hand to his mouth). This is what Lopes calls figure expression, but that’s not what gives the picture its expression. There are also design elements that reinforce the expression of anxiety in the depicted Kokoschka: The “agitated passages of paint, the awkward perspective and the vague, unstable use of space” are what link up the depicted figure with the expression of anxiety by the picture as a whole. [13]

This is a particularly interesting example insofar as the depicted figure is the same person as the painter: in painting his own self-portrait in this fashion, Kokoschka paints his gesture, his posture, his facial expression and so on in such a way that the picture as a whole expresses his own feelings of insecurity and anxiety. In R. G. Collingwood’s terms, his painting explores and makes concrete the emotions he is experiencing and thereby allows the viewer to feel what it is like to be the person depicted in the picture.[14] Lopes’ three types of (what he calls) ‘expression’ come together in the picture to express the emotion of the painter, in the rich Romantic way that I espouse, which in this case is also the emotion of the person depicted.

(2) Brassey thinks there is “an unresolved tension” in my response to Lopes, which is illustrated by my description of Kokoschka’s ‘Self Portrait with Hand to Mouth’.[15] I say that the painting depicts Kokoschka as “looking anxious and insecure” (‘figure expression’ à la Lopes), but that the picture as a whole also “conveys anxiety and insecurity in the agitated passages of paint, the awkward perspective and the vague, unstable use of space”, which I describe as ‘design expression’ (à la Lopes).[16] Brassey suggests this means that I think that design expression is “solely responsible for giving us” expression by the implied painter of the whole painting), whereas, as we just saw, I also say that figure, scene and design expression are only ‘expressive elements’, not genuine artistic expressions of emotion.[17]

Brassey has picked out an important ambiguity in my view. I should have been more careful when I talked about how a painting is unified by its ‘design’. The trouble stems from Lopes’ narrow definition of design expression as “an expression that is wholly attributable to a picture’s design or surface and not to any figure or scene it depicts”.[18] He cites Jackson Pollock’s non-representational drip paintings as an example. On Lopes’ view, it seems that figure, scene, and design expression are independent modes of what he calls ‘pictorial expression’. In ‘Fatherly Discipline’, for example, he claims that “the narrative comprises nothing but what the figures express”[19] but that certain of its design elements—notably the “coiled lines used to render the faces and hands of father and child”[20]—reinforce figure expression. Likewise, it is true that the Kokoschka painting has ‘design elements’, namely colour, brushwork, and use of perspective, that reinforce its ‘figure expression’. But my notion of ‘design’ is not as a set of design or formal elements, such as colour, line and use of space, but rather the way the whole painting expresses an emotion by means of its composition. By this I mean, using Lopes’ terms, that it is the way that the figures, scenes, and design—the ‘expressive elements’—in a painting work together to express an overall emotional state that unifies the painting.

It is only in this sense that design expression is “solely responsible for giving us” expression by the implied painter of the whole painting. [21] I am grateful to Brassey for pointing out this interesting ambiguity which has large consequences for my view and helps to indicate why I think it is preferable to Lopes’ view. Indeed, I am reminded of my long ago paper on ‘Style and Personality in the Literary work’, in which I argued that artistic expression is expression in an artist or implied artist, and that ‘style elements’—like expressive elements—cannot even be detected as such until we know what the overall artwork expresses. The rough brushwork in Kokoschka’s self-portrait is part of its overall expression, for example, whereas the rough brushwork in my portrait of myself is simply the result of incompetence.

Brassey’s point is at the heart of my view that pictorial works of art can sometimes express emotions in the way described by R.G. Collingwood: they do not merely depict or describe emotions; they explore and articulate the artist’s emotions in such a way that viewers can recreate for themselves the emotions expressed.[22] To conclude this section: at the beginning of her paper Brassey invites me to clarify “the relation between (a) what a picture expresses and (b) what is depicted as expressing in the picture” but she has done a good job of doing this herself.[23]

(3) Finally, I come to perhaps the most contentious issue raised by my view, which is the idea that, although the emotions expressed by a picture as a whole are typically those of the artist who expresses his or her emotions through the picture, in cases where this seems implausible, i.e., where the personality expressed in the work does not match the personality of the artist in daily life—Brassey cites “Tolstoy’s ability to write with compassion while lacking compassion in his personal relationships”—the emotions expressed should be attributed to an implied painter, the painter as he or she seems to be from the evidence of the painting in its historical and cultural context.[24] This idea raises hackles partly because the implied painter isn’t actually a painter at all and lacks any ‘psychological reality’, as Richard Wollheim once put the point.

My main goal in the fairly long discussion of this issue in ‘The Missing Person Found: Part One’[25] was to question those who put so much weight on the possibility of the implied author being different from the actual author. Broadly speaking, I wanted to point out that there is a much more intimate connection between actual and implied author than this objection assumes. On the other hand, it is no easy matter to describe exactly what the relationship is and in any case my suspicion is that different paintings and different painters will need to be treated differently in this respect. I acknowledge that implied painters may be suspect in some cases, because the apparent author of a painting seems to have so little in common psychologically with the actual author, but I guess I would want to insist that in most cases extreme skepticism is not warranted. Even if the implied author is not identical to the actual author, they typically have a great deal in common psychologically.

Lopes, of course, wants nothing to do with implied authors (or even authors!), since his is an ‘impersonalist’ account of expression, but Brassey points out that this has unfortunate consequences for him insofar as his “categories [of expression] can only account for meaning derived from the formal (visible) qualities of the picture surface”, so that “meaning is restricted to what goes on inside the represented pictorial world”.[26] In contrast, Brassey points out that if we accept the idea of an implied painter, we can make sense of the “different levels at which we understand a picture”.[27] Brassey explains her idea by reference to Edward Hopper’s “’Nighthawks’ which she says illustrates how “a naïve painterly style” is “put to use by a mature, skilled artist who adopts a naïve style”.[28] She cites Kendall Walton who “suggests that this kind of multi-level impression of an implied painter operates at a ‘deeper level’ than just taking it that the work is painted by a naïve illustrator and can lead theoretically to the viewer experiencing a ‘richness and elusiveness of style’ in some paintings that ‘derives from an obscure partial awareness of a multi-level situation of this kind’”.[29]  The point is a nice one and interestingly it once again links the discussion of expression to the notion of style. I am very happy to accept this idea.[30]


The rest of the papers are about literature and film. I turn first to the paper by Mary Beth Willard on the paradox of fiction, which she describes as a commitment to the following three mutually inconsistent propositions.[31]

  1. We have genuine and rational emotional responses towards fictional characters and situations.
  2. We believe that fictional characters and situations do not exist in reality.
  3. In order for us to have genuine and rational emotions towards a given character or situation, we must believe that they exist in reality.[32]

In ‘Deeper than Reason’ I gave a somewhat dismissive account of the paradox of fiction: given my view of what emotions are and how they operate, I argued that the so-called problem of the paradox of fiction dissolves. If emotions are triggered by people, events and situations which are of great significance to our survival and/or well-being, then our emotion systems respond to anything that seems significant in this way. The trick to getting readers to respond emotionally to works of literary fiction is to get readers interested in, curious about, and often sympathetic towards the characters and to get personally involved in their problems and difficulties. We feel sympathy for Anna Karenina and that’s why we weep for her even though we know she doesn’t exist. Similarly, we respond emotionally to what happens to characters on-screen in fictional movies.[33] What I say is that “[i]t does not matter to my emotion systems (fear, sadness, etc.) whether I am responding to the real, the merely imagined, the possible or the impossible”.[34] Always I respond very fast and ‘instinctively’ to something fearsome, offensive, contaminated or whatever, and subsequent cognitive monitoring—including realising that the object of my emotion is imaginary, for example—modifies my response so that my incipient action tendency may not issue in any actual action.

Willard’s main focus is on whether emotional responses to fiction are irrational.[35] My own view is that, strictly speaking, feeling emotions for fictional characters and events is irrational, even though it happens all the time. But Willard claims that I “shouldn’t concede irrationality, or at least not on the grounds that the characters do not exist”, because “existence turns out not to matter all that much to whether […] our emotional responses toward fiction are rational”.[36] This may be true in the sense that the existence of fictional characters and events is not germane to how we respond emotionally to them qua fictions. Often we respond to them as if they existed, even if we know they don’t (and didn’t). But as a philosopher, rather than as a reader of fictions, we can still ask whether such emotional responses are rational or not, and on the face of it, it does seem cognitively irrational to believe at one and the same time that ‘Anna Karenina is having a hard time’ and that ‘Anna Karenina does not exist’.[37]

The 2010 paper that Willard cites was ‘adapted and abbreviated’ from ‘Deeper than Reason’, chapters 4-5.[38] One section of chapter 5, which I omitted from the 2010 paper, is headed ‘Is it Irrational to Respond Emotionally to Fictions?’ and there I argue that, although from “a strictly cognitive point of view it is irrational to have wants and goals with respect to Anna Karenina and her ilk, it is not ‘emotionally irrational’ and it is certainly not maladaptive”.[39] Here I am drawing on the work of Patricia Greenspan, who argues that in a case of mixed feelings, when my good friend has won a prize that I would have liked to win myself, it is in her terms ‘basically rational’ or ‘emotionally rational’ to feel both happiness and unhappiness, and that feeling both emotions, even though this would be inconsistent on strictly cognitive grounds (given the way Greenspan has set up the case), “on a standard of rationality that evaluates emotions according to their behavioral consequences—which takes into account, for instance, the social value of identification with others—ambivalence might sometimes be more rational than forming an ‘all things considered’ emotion that resolves the conflict”.[40] I grant that emotions can be rational in Greenspan’s sense, which is more-or-less equivalent to ‘adaptive’, without being cognitively rational.

According to my view of emotion, emotions are processes. Emotional responses are bodily responses triggered by crude appraisals [e.g. This is GOOD or BAD for me] that sometimes operate on simple stimuli, at other times on cognitively complex stimuli.  Suppose I am frightened by the possibility of another recession. In this case, my fear presupposes that I have some cognitively complex beliefs or thoughts about economics. However, what makes my reaction a fear reaction is its registration of THREAT in bodily changes of a fairly specific sort (including physiological and behavioural changes, states of action readiness and so on), which fasten my attention on the fact of a threat. If this is right, then the bodily registration of THREAT occurs prior to any cognitive assessment of the economic situation, and cognitively it is arational. However, after the crude ‘bodily appraisal’ of THREAT, there is invariably ‘cognitive monitoring’ of the situation, which is itself subject to norms of rationality.[41]

However, none of this is really germane to Willard’s main point, which is that “whether our emotional responses towards fiction are rational depends on our compliance with […] the norms of aesthetic engagement” and that “the paradox of fictional emotions has always, at its core, been a question of aesthetic appraisal”.[42] Interestingly, when Colin Radford published the paper that started the paradox of fiction industry, ‘How Can We be Moved by the Fate of Anna Karenina?’[43] the companion piece to his, in a supplementary volume of the ‘Proceedings of the Aristotelian Society’ for 1975, was a paper by Michael Weston in which he points out that “it is the recognition of Mercutio as part of a work of art that I find missing in Doctor Radford’s treatment”.[44] The idea is similar in spirit to Dr Willard’s. As she says, most discussions of the paradox of fiction have followed Radford in adopting the assumption that “if it is irrational to weep over purely imagined states of affairs, then it is also irrational to respond emotionally to fictional characters and situations, which after all are imagined states of affairs, the production of which has been outsourced to an artist”.[45] My weeping about a possible but non-actual situation in which my fit and healthy husband has died of a wasting disease is ‘on a par’ with my weeping for Anna Karenina’s tragic suicide.

Willard argues that this is a mistake. She perceptively notes that the crucial difference between these two cases is that there are norms of aesthetic engagement governing our interactions with ‘Anna Karenina’—and hence also, presumably, with Anna Karenina, whereas there are no such norms governing our counterfactual imaginings about our friends and relations. This is an important and often neglected point, and Willard is quite right to emphasise the essential role that aesthetic norms play in the debate about the rationality of responding emotionally to fictions: what it’s rational to feel about Anna Karenina depends a very great deal on her role in the fiction in which she occurs, something that has no application when I’m musing sentimentally about my husband.[46]

However, I have some doubts about the details of Willard’s proposal. She describes norms of aesthetic engagement as norms governing “the responsibilities appreciators of art incur when they engage works of art”[47] and suggests that they include “norms that govern the range of licensed emotional responses to the artwork”.[48] “Roughly, [these norms] say, ‘to experience the work correctly, appreciators should respond in this way’”.[49] The two aesthetic norms most relevant to her discussion are, she says, “(1) Being open to engaging with fiction”, and “(2) Exhibiting the appropriate emotional response”.[50]

The first proposed norm seems innocuous, but I think there are problems with the second. The norm “Exhibit appropriate emotional responses [to a fiction]” is meant to rule out such inappropriate responses as laughing at ‘Schindler’s List’, failing to be shocked by the violence in a “gritty film noir”[51]; or responding with terror to any dramatic tension in a work (as in her young son’s reaction to some action movies). As she rightly points out, none of these responses will be blocked by the assumption that the characters and situations in question don’t exist. The problem, however, is a logical one. One of Willard’s main points is that “only emotions that are sanctioned by the norms of aesthetic engagement are appropriate to have toward fictional characters and situations”.[52] However, the norms she is referring to here include the norm “Exhibit appropriate emotional responses [to a fiction]”.[53] The claim is tautological: “only emotions appropriate to fictional characters and situations are sanctioned by the norm: ‘exhibit appropriate emotional responses to fictional characters and situations’”.

Another problem is that there is rarely one and only one appropriate emotional response to a particular fictional characters or situation.[54] I defend a ‘reader-response theory’ of literary interpretation which allows for diverse interpretations partially as a result of the different emotional responses different readers have to the characters and events in a novel.


Next I discuss the papers by Eva-Maria Konrad, Thomas Petraschka, and Christiana Werner (henceforth Konrad et al)[55]and by Quixote Vassilakis[56], both of which address my claim that appropriate emotional responses to realist novels are important to understanding and interpreting (those kinds of) novels. Konrad et al are happy to accept this relatively weak claim, but they reject my stronger and more controversial claim that in some cases a novel “cannot be adequately understood” without emotional involvement with the characters and events of the novel.[57] In their elegant and well-argued paper, they question three of my arguments, (1) ‘the trigger argument’, namely, that out of all the information that a novel provides, it is often our emotional responses that make salient the information most important to understanding the novel; (2) ‘the empathy argument’, namely, that “understanding character is relevantly like understanding real people, and that understanding real people is impossible without emotional engagement with them and their predicaments”[58]; and (3) ‘the data argument’, namely, that  “we use our emotional responses towards [characters] as data in arriving at an interpretation of [a] character”.[59]

(1) They agree with me that readers’s emotional responses are often ‘sources of salience’ that trigger recognition of particularly important events or scenes in a novel, but deny that “being triggered by [one’s] emotional reactions” is the only way for a reader to become aware of significant episodes in a text”.[60] They are surely right about this. In particular, as they point out, there are structural features of novels and plays that also help readers pick out the most significant events and scenes. They note that a crux in a play might occur in the very middle of the play, for example, as in the meeting between Mary and Elizabeth in Schiller’s ‘Maria Stuart’. And very often there is what has been called a ‘tension arc’ in a play or novel, making it more likely that tension will gather over the first part and reach a climax in the middle or towards the end of the play or novel. There is no need for our emotions to get involved.

On the other hand, a tension arc is designed to arouse our curiosity and feelings of suspense as well as the relief that follows the release of tension. And as for the meeting between Mary and Elizabeth we have been primed to expect it as an emotional climax in the play. So the emotions might be playing a role even when it appears that purely formal considerations are what determine salience. However, I agree with Konrad et al that it is unlikely that emotional triggers are the sole sources of salience in a literary work.

(2) More controversially, I argue that understanding characters in a realist novel is necessary to understanding the great realist novels of F. R. Leavis’ Great Tradition, that “understanding character is […] relevantly like understanding real people”, and that “understanding real people is impossible without emotional engagement with them and their predicaments”.[61] Konrad et al focus on readers’s empathic emotions for characters, and they have an ingenious argument designed to show that emotional responses to fictional characters are very different from emotional responses to people in real life. They suggest that on the one hand we might adopt a realist theory of character, in which case characters really exist but only as constructs of words, and on the other hand for anti-realists about characters, fictional characters are not real at all. Whichever alternative we adopt, however, it turns out that fictional characters lack mental states: constructs of words don’t have mental states and neither do phantom ‘characters’. Hence interacting with characters is radically unlike interacting with real people.

In response, I would point out that engaging with a realist novel involves treating characters as having two aspects: we are encouraged to treat them as real people and simultaneously as constructs of words fitting into a larger structure. In ‘The Philosophy of Literature’, Peter Lamarque talks about viewing characters from two perspectives, the internal perspective from which they are (mostly fictional) people in a set of situations, and the external perspective in which they are elements of the structure of the work.[62] What I’d like to emphasise is that if we did not respond to characters as if they were actual people, the point of a novel would be lost. If characters were nothing but constructs of words, they would not appeal to us emotionally nor would they get us engaged in issues of love, death, remorse, or any of the other great themes of literary novels. So although I would plump for the view that fictional characters are indeed ‘not real’ in a metaphysical sense, nevertheless, novel-reading, at least where the great realist works of fiction are concerned, requires that we treat them as if they were real.[63]

(3) Finally, the data argument.[64] Konrad et al object to my claim that certain data or information about characters can be gleaned only through our emotional reactions to those characters. My examples include gentle amusement for Lambert Strether, the main protagonist of James’ ‘The Ambassadors’ and compassion for Anna Karenina. As in their first argument, Konrad et al argue that there are other ways in which we can learn that Strether is mildly amusing and that Anna is in a desperate situation. This may well be true, but I think there is an important difference between ‘understanding’ in the abstract, and understanding with your gut, as it were. You can tell me that Strether is mildly amusing, and in the past maybe your views on such matters have been reliable, but unless I feel amused myself by Strether, why should I believe you that he is in fact amusing? Moreover, I won’t really understand either Strether or Anna as ‘people’ unless I approach them sympathetically. Otherwise, Strether may seem like nothing but a fuddy-duddy and Anna may come across as nothing but a foolish narcissist. In other words, treating characters as if they were real is not sufficient for understanding them as we understand our close friends: care and concern may also be necessary.         Konrad et al also argue that empathic responses to characters will provide new data to a reader only when that reader has never previously experienced the emotion in question. If I have suffered as Anna Karenina suffers in the eponymous novel, I can simply recall my own past experience in order to understand Anna’s present experience. They therefore argue that “empathic emotions are only necessary for those empathisers who have not felt the affective state in question before”.[65] I would like to respond in two ways to this claim. First, one of the things that novels can do is to broaden our emotional repertoires and to get us to feel in ways we have never felt before. Anna Karenina’s situation is described in vivid, painful detail in Tolstoy’s novel. Even if we remember having gone through a difficult divorce or being abandoned by a lover or having lost a child in a custody dispute, our remembered feeling cannot be identical to Anna’s because our situation in the contemporary Western world is so different from hers. Hence the data we glean from our own memories about what it’s like to be in a state of despair about such things may fail to illuminate Anna’s very individual situation. Secondly, as a matter of empirical fact, we have great difficulty remembering feeling a certain way in the past. It is much easier to remember that such and such happened to me. (Luckily, mothers typically forget what it felt like to give birth, for example.) Hence on more than one count it is unrealistic to think that recalling my own experiences will be a good source of data collection when I try to understand what a character is going through.

Quixote Vassilakis also takes me to task for overstating my case for the importance of emotional engagement to the understanding of literary works, especially given my understanding of what an emotional response is, namely, a response to a perception or thought of something of great significance to me or mine that is registered in bodily changes. But Vassilakis takes his argument in a different direction from Konrad et al. He is focused mainly on the possibility of variation among acceptable interpretations. He notes the ubiquity of individual differences in readers’s responses to narratives as well as the ever-present possibility of inappropriate emotional responses to narratives.

(1) Like Konrad et al, Vassilakis argues that it’s perfectly possible to understand another person’s emotional states—whether actual or fictional—without any kind of emotional bodily response, but, unlike Konrad et al, he bases his argument on the presumed fact that there are certain individuals—namely, psychopaths—who have a “severe emotional deficit” but also a “potentially ‘enhanced’ […] ability to understand the emotional cues of others”.[66] (2) He makes the perceptive point that “bias is built into emotion” and that it can lead to misunderstanding and even confabulation about how a narrative unfolds.[67] (3) Finally, he points out that readers can and do have inappropriate responses to fictional worlds and claims that I “cast reader error as incidental and largely correctable” by after-the-fact cognitive monitoring, whereas in fact the ubiquity and persistence of idiosyncratic emotional response is a central feature, not a rare anomaly, of emotions.[68] These are important and interesting criticisms which I will now attempt to answer.

(1) The argument from psychopathy strikes me as odd. First, psychopaths appear to figure out other people’s emotions by some kind of calculation. In this respect they are like autistic people who can tell that someone is (probably) happy by the fact that the sides of their mouth are pushed up, not because of any empathic emotional response to a smile. But, although psychopaths may be better than the rest of us at divining clues to a person’s mental states from the way they look and act (although I think the data are not clear on this point), in general they are hopeless at figuring out what other people feel, because they are very self-centered and totally lacking in empathy. More importantly, unlike (most) autistic people, psychopaths are stunningly lacking in care and concern for other people. Moreover, whatever the truth about psychopathy, it is clear that novels are not by and large written for psychopaths—a comparatively small group of very strange and scary people—and the existence of psychopaths does not seem to me to be relevant to principles about how to read a novel in a normal way.

(2) Vassilakis rightly points out that “bias is built into emotion” and he perceptively points out that one reason why is that “emotion necessarily elicits action tendencies, which can be difficult to reverse—or even to be aware of”, something he thinks I should emphasise more.[69] This is also true of other physiological reactions that comprise emotional reactions. He rightly realises that this creates difficulties for theories of interpretation that stress emotional engagement with a work, especially those that advocate immersing oneself in a novel and becoming so “deeply emotionally involved with a novel” that they “construe fictional worlds as real”, at least—I assume—for the period while they are reading.[70] If readers are too immersed—are completely absorbed in reading from the internal perspective—their interpretations may be warped. I agree. As I mentioned in the discussion of Konrad et al, interpretations of literary narratives require taking an external perspective as well as an internal perspective and going back and forth between the two. All this seems to me very sensible.

(3) Finally, Vassilakis presses me about how I deal with inappropriate emotional responses. He claims that on my view “readers must respond to certain great literary works in accordance with particularly-defined emotional patterns in order to properly value them”.[71] But this is not what I say. As I observed earlier, I defend a reader-response theory of literary interpretation, according to which different readers may arrive at different but equally coherent interpretations of a novel or other literary work. In chapter 4 of ‘Deeper Than Reason’, I remind my readers of Monroe Beardsley’s principles of congruence and plenitude, which provide criteria of correctness for interpretations i.e., “make your interpretation of the parts of the text consistent” and “make your interpretation fit as much of the text as possible”.[72] These are very broad criteria, however, and it is not hard to see how they can be met by quite different interpretations.

In general, I argue that when we are talking about great writers and their great works, the best interpretations are going to be those that adhere most closely to what the author probably intended. However, I was careful to point out (in chapter 6) that what sort of person the implied author is and what they probably intended is in the last resort up to the reader. ‘The meaning’ of a novel is determined by the interaction of author and reader. So I reject the charge that in my view ‘reader error’ is “incidental and largely correctable”.[73] A difference of opinion about a small passage can lead to major differences in the interpretations of two different but equally sensitive readers. Vassilakis is quite right to chide me for not emphasising this enough. He is also right to highlight “the ubiquity and persistence of idiosyncratic emotional response” as “a central feature, not a rare anomaly, of emotions”.[74] All I can say is that I heartily agree.


Finally, I come to Irene Martínez Marín’s[75] paper, on self-conscious emotions. Martínez Marín attacks my Jamesian theory of emotion, which claims that bodily changes are a sine qua non of episodes of emotion. She goes to the heart of the Jamesian position when she takes me to task for failing to distinguish between being emotional and having or experiencing an emotion. For me, all emotion episodes, whether initiated by a fleeting perception of a rattlesnake or a considered judgment that I’ll be bankrupt in six months, involve bodily changes. This is the crucial difference, it seems to me, between a mere judgment or thought and an emotion. However, I do acknowledge explicitly that emotion words are sometimes used to refer not to a genuine emotion but only to a related judgment or thought: “I am very proud of you” may simply report my approval of you, without any implication that I am currently in an emotional state of pride.[76] Similarly, if the shop assistant says “I regret that the jacket is not available in red”, we do not think that he is reporting on his current emotional state.

Martínez Marín wants to make a sharp distinction between “higher cognitive emotions” or “non-primary or intellectual emotions”, such as “guilt, shame, embarrassment, pride, or nostalgia”, and “basic emotions”, such as anger and fear, because she thinks that, unlike the latter, the former do not require any bodily involvement.[77]

Consequently, she thinks that according to my view “higher cognitive emotions [cannot have] the status of really being emotions”.[78] But this is not my view. For me, ‘higher cognitive emotions’ require bodily involvement in order to be genuine emotions, just like any other emotions. Here too I follow James, who argues that even in the case of ‘subtle emotions’ such as aesthetic emotions, ‘the bodily sounding-board’ is at work. There are no “pure cerebral emotions”, he says.

Unless in them there actually be coupled with the intellectual feeling a bodily reverberation of some kind, unless we actually laugh at the neatness of the mechanical device, thrill at the justice of the act, or tingle at the perfection of the musical form, our mental condition is more allied to a judgment of right than to anything else. And such a judgment is […] a cognitive act.[79]

He goes on:

But as a matter of fact the intellectual feeling hardly ever does exist thus unaccompanied. The bodily sounding-board is at work, as careful introspection will show, far more than we usually suppose.[80]

He thinks we should distinguish between “the keen perception of certain relations being right or not”, and “the emotional flush and thrill consequent there upon”, and claims that “these are two things, not one”.[81] This seems to me perfectly plausible.

However, this response, I think, ignores Martínez Marín’s most pressing concerns. To me her most interesting contribution is the challenge to my view she thinks is represented by the ‘film-diaries’ of Jonas Mekas. (1) She worries that “[my] model seems to be excluding from the process of art appreciation a set of emotions that are familiar to us in life and the arts” namely, “emotions that are partly constituted by cognitive processes that unfold over time and which require attention to the personal history of a person”, such as “nostalgia, […] grief, shame, envy, [and] pride”, and that have no obvious bodily signature.[82] She also argues that “the nostalgic view from which Mekas’ film-diaries are narrated cannot be grasped empathically because it is impossible to “[share] the particular object of the nostalgic emotion expressed in these experimental films”.[83] (2) Nostalgia, in Martínez Marín’s evocative phrase, is “an emotion concerned with the irrevocability of one’s own past”.[84] It involves not simply remembering my past life but reflecting upon it, so that in this sense nostalgia is indeed “partly constituted” by reflections.[85] But my question would be: why call nostalgia an emotion if it is primarily a set of reflections? What I have consistently claimed in my work is that long-term emotions—what Goldie might call ‘narrative’ emotions[86]—are not properly regarded as emotions, rather than mere judgments or thoughts, unless they involve episodes of emotion with bodily components. It seems to me that if I am in an emotional state of nostalgia my reflections about my past will themselves evoke emotions: very roughly, happiness at the memory of something loved, and sadness at the fact that this happiness is irrecoverable. Nostalgia is a bittersweet emotion.[87]

As for bodily symptoms of these emotions, as far as I know, it is not clear whether or not there are specific bodily changes in nostalgia,[88] but some of Martínez Marín’s other examples clearly do have bodily symptoms: shame and guilt involve withdrawal or hiding behaviours, for example, and grief is expressed in characteristic postures, gestures, vocal and facial expressions. Yet I am loath to say that the long-term emotion of nostalgia is nothing but a disposition to experience episodes of happiness about the past mixed with sadness about the past. As I am sure Martínez Marín would agree, this description does not seem to capture the phenomenology of long-term nostalgia.

My (untutored) impression is that Mekas is trying to create a mood of nostalgia, by creating films that are expressive of nostalgia and inviting spectators to feel nostalgia for their own pasts as they watch these flickering images. We can trace these effects to formal and expressive aspects of the films such as their ‘amateur’ appearance, characteristic of home movies, the unclear temporal sequencing, their impressionistic, dreamy quality, the lack of developed characters, and, more generally, the lack of narrative drive. Martínez Marín has identified a way in which art and emotion interrelate that is different from anything I have written about (although I have things to say about music that may be relevant) and I am grateful to her for introducing me to these haunting films. Clearly, they raise issues that I have not thought about enough.

(3) Martínez Marín also suggests that ‘affective empathy’ for the narrator of Mekas’ diary-films may not be possible because we can never “[access] the particular object of another’s self-conscious emotion”.[89] For example, in nostalgia, “we fail to intentionally imagine the other’s recollections in an empathetic way because the yearning associated with this emotion is inseparable from the subject’s self-evaluation of her own past”.[90] This may be true, but it’s also true that empathy is always a matter of degree, and although it may be hard for me to recreate in imagination Mekas’ nostalgia for the village of his youth, other attempts at empathy for situations closer to home are equally difficult to achieve. William Ickes in particular has shown that empathic accuracy is, in general, just not very great.[91] Recall, too, my discussion of Konrad et al. on the relevance of my memories to my ability to empathise with another person.

Finally, I cannot help ending this comment on a plaintive note. Throughout ‘Deeper than Reason’ and most of my other work on art and emotion, I explicitly say that I am focusing on canonical and/or realist works in the various arts: ‘Anna Karenina’, ‘The Raft of the Medusa’, Beethoven’s symphonies etc., and that I am precisely not discussing avant-garde, experimental, post-modern or other non-realist works.[92] Mekas is a very interesting case, but not the sort of thing that I was aiming to analyse. What Martínez Marín’s discussion reveals is that there is a lacuna in my view—how should we think of long-term higher cognitive emotions? —that I might have recognised and tried to fill. But there again, the book is already 500 pages long, and, more importantly, my critics have demonstrated that they and no doubt others are well-equipped to take on some of these issues for themselves!

Thank you again to all my critics for really interesting and perceptive papers.

About the AuthorJenefer Robinson is professor emerita at the University of Cincinnati where she was full professor of philosophy from 1985 until 2016. She writes mainly on topics in aesthetics and philosophical psychology, especially the theory of emotion. Her book, Deeper than Reason (OUP 2005) applied recent advances in emotion theory to issues in aesthetics, such as the expression of emotion in the arts, how music arouses emotions and moods, and how the emotional experience of literature and music in particular can be a mode of understanding and appreciation. More recently she has written a series of papers on empathy and the arts, including papers on architecture, painting, literature [in progress] and music. Jenefer was President of the American Society for Aesthetics from 2009-2011.


Bedford, Errol. (1967). “Emotions”. In: Essays in Philosophical Psychology, Donald Gustafson (Ed.). London: Macmillan, pp. 77-98.

Collingwood, Robin George. (1963). The Principles of Art. Oxford: Clarendon.

De Sousa, Ronald. (1987).  The Rationality of Emotion. Cambridge, MA: MIT Press.

Greenspan, Patricia. (1980). “A Case of Mixed Feelings: Ambivalence and the Logic of Emotions”. In: Explaining Emotions, Amelie Rorty (Ed.). Berkeley: University of California Press, pp. 223-250.

Goldie, Peter. (2000). The Emotions: A Philosophical Exploration. New York: Oxford University Press.

Ickes, William. (1997). Empathic Accuracy. New York: Guilford Press.

James, William. (1884). “What is an Emotion?”. Mind 9(34), pp. 188-205.

Lamarque, Peter. (2009). The Philosophy of Literature. Malden, MA: Blackwell.

Lopes, Dominic. (2005). Sight and Sensibility: Evaluating Pictures. New York: Oxford University Press.

Radford, Colin. (1975). “How Can We Be Moved by the Fate of Anna Karenina?”. Proceedings of the Aristotelian Society. Supplementary Volumes 49, pp. 67-80.

Robinson, Jenefer. (2004). “Emotion: Biological Fact or Social Construction?”. In: Thinking about Feeling, Robert C. Solomon (Ed.). New York: Oxford University Press, pp. 28-43.

—.  (2005). Deeper than Reason: Emotion and its Role in Literature, Music, and Art. Oxford: Clarendon Press.

—. (2010). “Emotion and the Understanding of Narrative”. In: Routledge Companion to the Philosophy of Literature, Garry Hagberg & Walter Jost (Eds.). Oxford: Wiley-Blackwell, pp. 69-92.

—.  (2017). “The Missing Person Found, Part One”. British Journal of Aesthetics 57, pp. 249-267.

—.  (2018). “Emotion as Process”. In: The Ontology of Emotions, Hichem Naar & Fabrice Teroni (Eds.). Oxford: Oxford University Press, pp. 51-70.

Walton, Kendall. (1976). “Points of View in Narrative and Depictive Representation”. Noûs 10, pp. 49-61.

Weston, Michael. (1975). “How Can We Be Moved by the Fate of Anna Karenina?”. Proceedings of the Aristotelian Society, Supplementary Volumes 49, pp. 81-93.

[1] Brassey 2019, pp. 15-29.

[2] Robinson 2017.

[3] Lopes 2005.

[4] Lopes 2005, p. 51.

[5] Ibid., p. 52.

[6] Ibid., p. 57.

[7] Brassey 2019, p. 18.

[8] Ibid., p. 20.

[9] Ibid.

[10] Ibid.

[11] Ibid., p. 21.

[12] Ibid.

[13] Robinson 2017, p. 262.

[14] See Collingwood 1963.

[15] Brassey 2019, p. 22.

[16] Robinson 2017, p. 262.

[17] Brassey 2019, p. 22.

[18] Lopes 2005, p. 57.

[19] Ibid., p. 55.

[20] Ibid., p. 57.

[21] Brassey, 2019, p. 22.

[22] See Collingwood 1963.

[23] Brassey 2019, p. 15.

[24] Ibid., p. 25-26.

[25] Robinson 2017.

[26] Brassey 2019, p. 23.

[27] Ibid., p. 22.

[28] Ibid., pp. 22-23.

[29] Brassey 2019, p. 23; Walton 1976, p. 52.

[30] See Walton 1976.

[31] Willard 2019, pp. 30-44.

[32] Ibid., pp. 30-31.

[33] Willard says that “on Robinson’s view, becoming emotionally involved in a movie requires nothing more than instinctively responding to the visual stimuli” (p. 32.). This makes it sound as if I think movie-watching is nothing but paying attention to changing patterns of light and colour, which is not what I mean. The “visual stimuli” are the filmed characters, scenes and situations.

[34] Robinson 2005, p. 145; Robinson 2010, p. 86.

15 She also talks a little about whether they are genuine. I assume she is referring to the debate about whether the emotions evoked by fictions are genuine insofar as they meet the same conditions as ‘life’ emotions, as opposed to the quasi-emotions posited by Kendall Walton and others.

[36] Willard 2019, p. 32 (my italics).

[37] In deciding whether emotions in general or emotions for fictions in particular are rational or irrational, much depends on what your conception of rationality is. In what is still one of the best extended discussions of ‘the rationality of emotion’ Ronald de Sousa, in his eponymous book, distinguishes rationality of belief from rationality of desire and both from rationality of emotion. Cognitive rationality aims at truth. Strategic rationality—or rationality of desire—aims at goodness. Emotional (or axiological) rationality depends upon the ‘paradigm scenarios’ within which a given emotion term was learned. Fear is rational if the situation I am in is sufficiently like the fearsome situations in which I learned the meaning of ‘fear’.

[38] Robinson 2005.

[39] Ibid., p. 147.

[40] Greenspan 1980, p. 238.

[41] For an explication and defence of my evolving view of emotion, see Robinson 2004, Robinson 2005, and Robinson 2018.

[42] Willard 2019, p. 32.

[43] Radford 1975.

[44] Weston 1975, p. 84.

[45] Willard 2019, p. 33.

[46] In the discussion of Konrad et al I note the double nature of characters as both human beings and elements in a design.

[47] Willard 2019, p. 34.

[48] Ibid., p. 32.

[49] Ibid., p. 34.

[50] Ibid.

[51] Ibid., p.33.

[52] Ibid., p. 41.

[53] Ibid., p. 35.

[54] What counts as an ‘appropriate’ response is an issue I return to in the discussion of Quixote Vassilakis’ paper.

[55] Konrad et al 2019, pp. 45-59.

[56] Vassilakis 2019, pp. 60-73.

[57] Robinson 2005, p. 107; Konrad et al 2019, pp. 46 & 49.

[58] Robinson 2005, p. 126; Robinson 2010, p. 78.

[59] Robinson 2005, p. 116; Konrad et al 2019, p. 55.

[60] Konrad et al 2019, p. 55.

[61] Robinson 2005, p. 126.

[62] Lamarque 2009, pp. 201-202.

[63] In Robinson 2005, chapter 4, I defend the view that we should respond to fictional characters as if they were human beings, and in chapter 7 I draw attention to the ‘double function’ of both plot and character, e.g. p. 215. Both, I argue, are also “important [principles] of structure or form” (ibid.).

[64] Konrad et al 2019, pp. 54-57.

[65] Ibid., p. 52.

[66] Vassilakis 2019, pp. 61 & 65.

[67] Ibid., p. 68.

[68] Ibid., p. 69 (my italics).

[69] Ibid., p. 68.

[70] Ibid.

[71] Ibid., p. 61.

[72] Robinson 2005, pp. 428-9.

[73] Vassilakis 2019, p. 69.

[74] Ibid.

[75] Martínez Marín 2019, pp. 74-94.

[76] Compare Bedford 1956-7.

[77] Ibid., p. 73.

[78] Ibid., p. 80.

[79] James 1884, pp. 201-202.

[80] Ibid.

[81] Ibid. I suspect that Martínez Marín has misread James and perhaps also me on this point.  James does not claim, as she accuses him of doing, that the main difference between ‘standard emotions’ and ‘non-standard’ or ‘intellectual’ emotions, is that “the bodily sounding-board, vibrating in the one case, is in the other mute” (p. 80). He is claiming that the bodily sounding-board is always “vibrating” in a state of emotion, whether “standard” or “non-standard”. It is ‘mute’, however, for the critic who has become jaded and incapable of the “flush and thrill” of genuine aesthetic delight. On the one hand there is the critic who is delighted by the elegance of the proof or ‘flushed and thrilled’ by the beauty of the music, and on the other hand the jaded critic who recognises the beauty perhaps, but does not feel the emotional flush or thrill that he did in his youth.

[82] Martínez Marín 2019, p. 80 (my italics). She thinks that such emotions are better accounted for on Peter Goldie’s theory of emotions. See Goldie 2000. I do not have space to defend this claim, but I have always found Goldie’s key concept of ‘feelings towards’ as very unclear.

[83] Ibid., p. 76.

[84] Ibid., p. 75.

[85] Ibid., p. 84.

[86] For Goldie’s view of emotions as having a narrative character, see Goldie 2000, especially chapters 1 and 2.

[87] Notice that a long-term emotion is not the same as a higher cognitive emotion: I can have a brief surge of pride for my son that lasts a few seconds as well as a long-term emotional state of fear or anger such as fear of snakes or anger about Brexit.

[88] Both happiness and sadness have been studied, but it seems unlikely that nostalgia would get half its bodily symptoms (say) from happiness and the other half from sadness: this would pose a problem if only because the two sets of bodily changes seem to be incompatible.

[89] Martínez Marín 2019, p. 86.

[90] Ibid.

[91] See e.g. Ickes 1997.

[92] And I barely mention movies!