This poster drew my attention while I was working in the library of Cambridge University a couple of weeks ago:
For a while now, I have been fascinated with the way in which the language of emotions, or affect, has penetrated public discourse. People ‘love’ all sorts of things: the way a film uses interior light, the icing on a cake, their friend’s new hairstyle. They ‘hate’ Donald Trump, the weather, next door neighbours’ music. More often than not, conversations involving emotions would not be complete without mentioning online expressions of affect, such as ‘likes’ or ‘loves’ on Facebook or on Twitter.
Of course, the presence of emotions in human communication is nothing new. Even ‘ordinary’ statements – such as, for instance, “it’s going to rain tomorrow” – frequently entail an affective dimension (most people would tend to get at least slightly disappointed at the announcement). Yet, what I find peculiar is that the language of affect is becoming increasingly present not only in non-human-mediated communication, but also in relation to non-human entities. Can you really ‘love’ a library? Or be ‘friends’ with your local coffee place?
This isn’t to in any way concede ground to techno-pessimists who blame social media for ‘declining’ standards in human communication, nor even to express concern over the ways in which affective ‘reaction’ buttons allow tracking online behaviour (privacy is always a problem, and ‘unmediated’ communication largely a fiction). Even if face-to-face is qualitatively different from online interaction, there is nothing to support the claim that makes it inherently more valuable, or, indeed, ‘real’ (see: “IRL fetish” [i]). It is the social and cultural framing of these emotions, and, especially, the way social sciences think about it – the social theory of affect, if you wish – that concerns me here.
Fetishism and feeling
So what is different about ‘loving’ your library as opposed to, say, ‘loving’ another human being? One possible way of going about this is to interpret expressions of emotion directed at or through non-human entities as ‘shorthand’ for those aimed at other human beings. The kernel of this idea is contained in Marx’s concept of commodity fetishism: emotion, or affect, directed at an object obscures the all-too-human (in his case, capital) relationship behind it. In this sense, ‘liking’ your local coffee place would be an expression of appreciation for the people who work there, for the way they make double macchiato, or just for the times you spent there with friends or other significant others. In human-to-human communication, things would be even more straightforward: generally speaking, ‘liking’ someone’s status updates, photos, or Tweets would signify appreciation of/for the person, agreement with, or general interest in, what they’re saying.
But what if it is actually the inverse? What if, in ‘liking’ something on Facebook or on Twitter, the human-to-human relationship is, in fact, epiphenomenal to the act? The prime currency of online communication is thus the expenditure of (emotional) energy, not the relationship that it may (or may not) establish or signify. In this sense, it is entirely irrelevant whether one is liking an inanimate object (or concept), or a person. Likes or other forms of affective engagement do not constitute any sort of human relationship; the only thing they ‘feed’ is the network itself. The network, at the same time, is not an expression, reflection, or (even) simulation of human relationships: it is the primary structure of feeling.
“Human nature will be transformed in the 21st century because intelligence is uncoupling from consciousness. We are not going to build machines any time soon that have feelings like we have feelings: that’s consciousness. Robots won’t be falling in love with each other (which doesn’t mean we are incapable of falling in love with robots). But we have already built machines – vast data-processing networks – that can know our feelings better than we know them ourselves: that’s intelligence. Google – the search engine, not the company – doesn’t have beliefs and desires of its own. It doesn’t care what we search for and it won’t feel hurt by our behaviour. But it can process our behaviour to know what we want before we know it ourselves. That fact has the potential to change what it means to be human.”
On the surface level, this makes sense. Algorithms can measure our ‘likes’ and other emotional reactions and combine them into ‘meaningful’ patterns – e.g., correlate them with specific background data (age, gender, location), time of day, etc., and, on the basis of this, predict how you will act (click, shop) in specific situations. However, does this amount to ‘knowledge’? In other words, if machines cannot have feelings – and Harari seems adamant that they cannot – how can they actually ‘know’ them?
Frege on Facebook
This comes close to a philosophical problem I’ve been trying to get a grip on recently: the Frege-Geach (alternatively, the embedding, or Frege-Geach-Searle) problem. It is comprised of two steps. The first is to claim that there is a qualitative difference between moral and descriptive statements – for instance, between saying “It is wrong to kill” and “It is raining”. Most humans, I believe, would agree with this. The second is to observe that there is no basis for claiming this sort of difference based on sentence structure alone, which then leads to the problem of explaining its source – how do we know there is one? In other words, how it could be that moral and descriptive terms have exactly the same sort of semantic properties in complex sentences, even though they have different kinds of meaning? Where does this difference stem from?
The argument can be extended to feelings: how do we know that there is a qualitative difference between statements such as “I love you” and “I eat apples”? Or loving someone and ‘liking’ an online status? From a formal (syntactic) perspective, there isn’t. More interestingly, however, there is no reason why machines should not be capable of such a form of expression. In this sense, there is no way to reliably establish that likes coming from a ‘real’ person and, say, a Twitterbot, are qualitatively different. As humans, of course, we would claim to know the difference, or at least be able to spot it. But machines cannot. There is nothing inherent in the expression of online affect that would allow algorithms to distinguish between, say, the act of ‘loving’ the library and the act of loving a person. Knowledge of emotions, in other words, is not reducible to counting, even if counting takes increasingly sophisticated forms.
How do you know what you do not know?
The problem, however, is that humans do not have superior knowledge of emotions, their own or other people’s. I am not referring to situations in which people are unsure or ‘confused’ about how they feel [ii], but rather to the limited language – forms of expression – available to us. The documentary “One More Time With Feeling”, which I saw last week, engages with this issue in a way I found incredibly resonant. Reflecting on the loss of his son, Nick Cave relates how the words that he or people around him could use to describe the emotions seemed equally misplaced, maladjusted and superfluous (until the film comes back into circulation, Amanda Palmer’s review which addresses a similar question is here) – not because they couldn’t reflect it accurately, but because there was no necessary link between them and the structure of feeling at all.
Clearly, the idea that language does not reflect, but rather constructs – and thus also constrains – human reality is hardly new: Wittgenstein, Lacan, and Rorty (to name but a few) have offered different interpretations of how and why this is the case. What I found particularly poignant about the way Cave frames it in the film is that it questions the whole ontology of emotional expression. It’s not just that language acts as a ‘barrier’ to the expression of grief; it is the idea of the continuity of the ‘self’ supposed to ‘have’ those feelings that’s shattered as well.
Love’s labour’s lost (?): between practice and theory
This brings back some of my fieldwork experiences from 2007 and 2008, when I was doing a PhD in anthropology, writing on the concept of romantic relationships. Whereas most of my ‘informants’ – research participants – could engage in lengthy elaboration of the criteria they use in choosing (‘romantic’) partners (as well as, frequently, the reasons why they wouldn’t designate someone as a partner), when it came to emotions their narratives could frequently be reduced to one word: love (it wasn’t for lack of expressive skills: most were highly educated). It was framed as a binary phenomenon: either there or not there. At the time, I was more interested in the way their (elaborated) narratives reflected or coded markers of social inequality – for instance, class or status. Recently, however, I have been going back more to their inability (or unwillingness) to elaborate on the emotion that supposedly underpins, or at least buttresses, those choices.
Theoretical language is not immune to these limitations. For instance, whereas social sciences have made significant steps in deconstructing notions such as ‘man’, ‘woman, ‘happiness’, ‘family’, we are still miles away from seriously examining concepts such as ‘love’, ‘hate’, or ‘fear’. Moira Weigel’s and Eva Illouz’ work are welcome exceptions to the rule: Weigel uses the feminist concept of emotional labour to show how the responsibility for maintaining relationships tends to be unequally distributed between men and women, and Illouz demonstrates how modern notions of dating come to define subjectivity and agency of persons in ways conducive to the reproduction of capitalism. Yet, while both do a great job in highlighting social aspects of love, they avoid engaging with its ontological basis. This leaves the back door open for an old-school dualism that either assumes there is an (a- or pre-social?) ‘basis’ to human emotions, which can be exploited or ‘harvested’ through relationships of power; or, conversely, that all emotional expression is defined by language, and thus its social construction the only thing worth studying. It’s almost as if ‘love’ is the last construct left standing, and we’re all too afraid to disenchant it.
For a relational ontology
A relational ontology of human emotions could, in principle, aspire to de-throne this nominalist (or, possibly worse, truth-proceduralist) notion of love in favour of one that sees it as a by-product of relationality. This isn’t claiming that ‘love’ is epiphenomenal: to the degree to which it is framed as a motivating force, it becomes part and parcel of the relationship itself. However, not seeing it as central to this inquiry would hopefully allow us to work on the diversification of the language of emotions. Instead of using a single marker (even as polysemic as ‘love’) for the relationship with one’s library and one’s significant other, we could start thinking about ways in which they are (or are not) the same thing. This isn’t, of course, to sanctify ‘live’ human-to-human emotion: I am certain that people can feel ‘love’ for pets, places, or deceased ones. Yet, calling it all ‘love’ and leaving it at that is a pretty shoddy way of going about feelings.
Furthermore, a relational ontology of human emotions would mean treating all relationships as unique. This isn’t, to be clear, a pseudoanarchist attempt to deny standards of or responsibility for (inter)personal decency; and even less a default glorification of long-lasting relationships. Most relationships change over time (as do people inside them), and this frequently means they can no longer exist; some relationships cannot coexist with other relationships; some relationships are detrimental to those involved in them, which hopefully means they cease to exist. Equally, some relationships are superficial, trivial, or barely worth a mention. However, this does not make them, analytically speaking, any less special.
This also means they cannot be reduced to the same standard, nor measured against each other. This, of course, runs against one of capitalism’s dearly-held assumptions: that all humans are comparable and, thus, mutually replaceable. This assumption is vital not only for the reproduction of labour power, but also, for instance, for the practice of dating [iii], whether online or offline. Moving towards a relational concept of emotions would allow us to challenge this notion. In this sense, ‘loving’ a library is problematic not because the library is not a human being, but because ‘love’, just like other human concepts, is a relatively bad proxy. Contrary to what pop songs would have us believe, it’s never an answer, and, quite possibly, neither the question.
Some Twitter wisdom for the end….
[i] Thanks go to Mark Carrigan who sent this to me.
[iii] Over the past couple of years, I’ve had numerous discussions on the topic of dating with friends, colleagues, but also acquaintances and (almost) strangers (the combination of having a theoretical interest in the topic and not being in a relationship seem to be particularly conducive to becoming involved in such conversations, regardless of whether one wants it or not). I feel compelled to say that my critique of dating (and the concomitant refusal to engage in it, at least as far as its dominant social forms go) does not, in any way, imply a criticism of people who do. There is quite a long list of people whom I should thank for helping me clarify this, but instead I promise to write another longer post on the topic, as well as, finally, develop that app :).