Social Media, Emotions and Deception [Part 2]

In my previous post about Facebook’s experiment with users’ newsfeeds I focused on how the outrage at recent publication of the data perhaps related to the more longstanding relationship that we have to companies as regards our emotions. In this post I want to think a little about why the study doesn’t tell us very much about emotions and how we might transmit, circulate or transfer affective or emotional states.

As regards how we conceptualise emotions and affect, it is notable that the research uses an impoverished account of emotions, specifically as regards its two primary concepts: ‘emotional state’ and ‘emotional contagion’. The paper’s ultimate conclusion is that they have found “the first experimental evidence for massive-scale emotional contagion via social networks.” So the concept of contagion is just as important to their work as is the idea of an emotional state itself.

In terms of emotional state the study uses a simple measure of whether the expressed emotional content was ‘positive’ or ‘negative’. They classified statements according to whether they contained ‘positive’ or ‘negative’ words defined by the Linguistic Inquiry and Word Count Software. This is a problem because much meaning derived from expressions is indexical and implicated. The (emotional) meaning of something is often irreducibly tied to its context of expression. So, during the week in January 2012 when the experiment on Facebook users took place, people could have decided to share and comment on the Daily Mail’s characteristically superficial report on some findings that 1 in 4 office workers is chronically bored.

when im bored i check facebookLet’s imagine someone makes a status update about the article saying “I’m too bored to even read it!” This could only be understood in context and might actually imply, contrary to the surface meaning, that the person was far from bored and in fact quite interested enough to have created a joke based on the article. Indeed, it being a joke would be impossible to capture through the automated system judging the statements according to positive and negative words. So the affective and emotional dimensions of its construction – which might involve laughter or anger, for instance – might be completely misclassified. The fact is that our emotional expressions are indexical too and they can be complicated, conflicting, ongoing or transient all at the same time. Indeed, the idea of a discrete emotional state, as implied by the article, is a problem. Rarely do we feel a single emotion for any length of time, save for in psychiatric conditions characterised by such persistence. Instead, we have multiple emotional states depending on what’s going on for us at the time. I might be miserable about something I’m reading on Facebook whilst only a moment later find myself laughing at something on the TV in the background, all the while  I’m bound up with feelings of stress and boredom from trying to plough through a difficult piece of work. So if we take seriously the way in which emotions are described through situated language and the complexity of their embodied and situated interrelations then there isn’t much hope for the measure of individual ‘emotional states’ as adopted in the paper.

The idea of emotional contagion is similarly problematic and needs critical engagement if we are going to begin to understand how emotions circulate. The article asserts that “Emotional states can be transferred to others via emotional contagion, leading them to experience the same emotions as those around them.” Some of the problems here are to do with the metaphor of ‘contagion’, which implies that we ‘catch’ or are infected by emotions from others and then spread those emotions to others. Other problems are about the assumption that when others behave in a similar way to us, for example by posting ‘negative’ comments on Facebook, that they are “experiencing the same emotions” as those around them.

The term emotional contagion is most famously used by scholars in the first decades of the 1900s to make sense of crowd and collective behaviour. Gustav Le Bon is most regularly attributed the questionable honour of its invention but others including sociologists such as Georg Simmel, Gabriel Tarde and Herbert Blumer have made use of it to examine the emotional behaviours they felt were exhibited in crowds. The concept was appealing to them because it helped to explain why seemingly rational people could be co-opted into irrational behaviour. In this regard, it was particularly the cases of racist, xenophobic, fascist and far-right politics that they hoped to explain away. So the concept has had a politics itself from its first uses in social science. Some of its use today is similarly political, as scholars have sought to understand rioting behaviour, terrorism and contemporary far-right politics through the assumption that emotional contagion is involved in the transition from rational to irrational behaviours in crowd and collective contexts.

However, I think that the term is more of a hindrance than a help if the metaphor stops at its most immediate implications of a 1:many transmission, without critical scrutiny, from body to body. But this simplifies how people respond to the observation of emotions in others, how they feel emotions themselves, how emotions relate to spaces, and of how people are recruited to or resist recruitment to collective affective activity. Indeed, to take just one of these examples, people regularly do resist the emotions that appear to be spreading through a crowd. Take the riots in England in 2011. Plenty of people were ‘recruited’ the emotional behaviours involved in the rioting, but plenty of people nearby the riots were not. What made the difference between the people who were engaged by the spectrum of emotions of the various rioting actions across the country and those who were not?

Drawing on a number of more contemporary theories of affect and emotion, Margaret Wetherell (2012) has argued that:

“any particular instance of the circulation of affect, whether occurring in consulting rooms, parliamentary committees, football stadiums or in the message boards of the Internet, involves understanding a raft of processes: body capacities to re-enact the actions of others; the developmental infrastructure of inter-subjectivity; the power of words; the affective-discursive genres personal and social histories provide which channel communal affect; inter-subjective negotiations; consideration of the cultural and social limits on identification and empathy; and exploration of practices of authorisation, legitimation and resistance, not to mention analyses of the containing institutions, spaces and media circulation.”

It is this vast list of features that is implicated in how emotions and their circulation is situated. Consideration of how emotions might circulate from one person to another or to many others cannot be captured by the classification of emotional states according to ‘negative’ and ‘positive’. And even if emotions could be easily classified in such a way, the finding that a greater percentage of Facebook statements were negative when users were presented with more negative statements would tell us very little about how emotional states circulate through social networks.

We need to situate emotional circulations and not de-socialise the body in order to argue for a virus like process of dumb replication. Indeed, even a little expansion of the contagion metaphor would help show how inadequate to the task the emphasis on 1:many body-to-body transfer is. Human Immune SystemFor a start, people have immune systems and so regularly are able to resist transmission of infections. The immune system is made of various processes that interact in complex ways with each other, with other bodily systems and with other bodies and the environment, so that at any given point in time a person may be more or less susceptible to a particular infective agent than at any other. Repeated exposure to infectious agents often increases this resistance. Our immune systems are individual and change over time, they are inherited and evolved and can be supplemented or compromised by various technological interventions. And so on and so forth.

To understand how emotions are circulated we have to think more broadly about what emotions are and how we can know about them, and do so from within a socialised account of embodied, ongoing, habituated and yet spontaneous, contextual and situated intersubjective life. The concepts of negative/positive emotional states and emotional contagion will not do very much for this line of thinking.



Social Media, Emotions and Deception: The Facebook Experiment [Part 1]

The recent response to the Facebook experiment with user’s newsfeeds can tell us something about how people understand three topics that are of interest to me. First, the way in which social media figure in human relations; second, how emotions and affect are conceptualised; and third how we understand lies and deceptions as part of everyday life. In this post I’ll mostly be dealing with the social media issue and emotions, and will be getting back to lying and deception in future posts.

The Facebook experiment, conducted in collaboration with researchers from the University of California and Cornell, was designed in order to understand whether emotional states could be transmitted by ‘emotional contagion’. The site removed either positive or negative stories from a sample of user’s feeds, which meant that some people were shown an unusually high number of positive or negative stories. The site didn’t invent or add any material, they just changed how items were selected from the pool of possible newsfeed stories. They then monitored those user’s own posts and determined whether they were more likely to post positive or negative stories as a result of this subtraction from their feeds. The results show a small but significant effect of 0.001, meaning that an increase in positive newsfeed items led to an increase in user’s own positive statements on Facebook and vice-versa. The significance of the effect is perhaps misleading and when viewed in light of the scale of Facebook it certainly looks more note-worthy, as the authors point out: “…given the massive scale of social networks such as Facebook, even small effects can have large aggregated consequences… an effect size of d = 0.001 at Facebook’s scale is not negligible: In early 2013, this would have corresponded to hundreds of thousands of emotion expressions in status updates per day.” So it seems that the results do show something, though what exactly I’ll get to in a later post. For now, let’s focus on the way in which the experiment has been received and what this might tell us about how we feel about advertising and social media.

It is unsurprising to find social networking websites are interested in the degree to which they can shape human emotions and self-expression. Indeed, individuals regularly go about trying to do this to each other in everyday life. But it’s obviously a different issue when we’re talking about one of the most powerful corporations in the world experimenting with how we feel about ourselves and others.

Of course, the advertising presented on Facebook is already targeted to an individual’s expressed interests and the kinds of topics that they post about, like, share and so on. I suspect that most people already know this. One only has to note how having looked at some pages for upcoming gigs in an area suddenly increases the number of ads you are shown about bands you like. In one case I briefly flirted with the idea of buying a juicer, and then having decided they were too expensive and I was unlikely to really use it, found that Facebook was quite insistent that I’d made the wrong choice and proceeded to show me juicers of various kinds for at least a week afterwards. So people may not be familiar with the technical details or aware of the scale of data involved, but they probably do have the sense that ads are being targeted to them. I think that this is all part of the contemporary neoliberal contract online. We know that we get a plethora of information and can make choices about consumption in new ways at the expense of that information being shaped by our previous choices, expressed interests and social relationships. However, the experiment has caused significant consternation online.

I think this tells us that people placed some degree of trust in Facebook not to be manipulating how they feel in a negative way. Indeed, much of the backlash seems to have been about how Facebook could make us feel bad about ourselves and the world around us. Some have pointed out that the research didn’t take ethics terribly seriously, and I agree. Most significantly, people were selected to be part of the trial randomly, and so there is a strong chance that at least some of the ~155,000 people subject to the feeds skewed to negative statements were vulnerable in one way or another. Seeing a ‘negative’ news story on Facebook can seriously affect my day, particularly if some god-awful politician has said something vainglorious, homophobic, racist, sexist, and so on and so on. Which they do on a practically hourly basis. Fortunately, I’m reasonably stable as regards my emotions but some people may not be and the study should have at least reflected on the importance of this.

But why do we trust Facebook with our emotions at all? I think part of it has to do with what we’re already willing to allow companies to do and what we expect them not to be doing. Most of us notice that the newsfeed changes when you look at it repeatedly throughout the day, and so we understand that some kind of computer-based selection process and prioritisation is probably going on. But I think we largely assume that this is based on selecting content that is most interesting to us. So we all end up with cat videos clogging the page. And we’re mostly fine with that. But we clearly have been surprised to find that the algorithm is not only capable of doing more than this but that it has actively been used to test out how changes to our newsfeeds might affect us personally. In this regard, we have balked not at Facebook trying to alter the items we see on our newsfeeds but at their actively trying to alter how this makes us feel about ourselves. And, as I’ve said, it is the negative side of this manipulation that has caused most distress.

The ostensible neoliberal contract with advertisers is that they make us feel good about ourselves in order to try and get us to buy stuff. Of course, we recognise that in some ways advertising is bad for us, particularly when it comes to body image. But more often than not, and particularly with video advertisements, the message is about how wonderful we are, how great the world is, how brilliant this product is, and how we should buy it to make sure we keep feeling so wonderful all the time. People are savvy about this and we can generally see how advertisers are trying to manipulate us. For example, there’s nothing hidden in how Fosters tries to appeal to contemporary notions of masculinity and male comradery. But we don’t expect Fosters to try and make us feel shit about ourselves in order to sell us beer. That’s not on. And in this regard, Facebook broke the deal and now we’re angry about it. So the answer certainly seems to be that yes, changing our newsfeeds can change our emotions, but perhaps not in the way they expected. The backlash against Facebook offers an excellent case from which to begin to unpick how we understand our emotional lives and how this figures in the organisation of contemporary economic exchange.