FEATURE13 March 2017

Tech support

x Sponsored content on Research Live and in Impact magazine is editorially independent.
Find out more about advertising and sponsorship.

Features Impact Technology

Social media has not only been identified as a cause of stress, but as a tool for diagnosing it. What’s more, an algorithm has been created that allows computers to offer emotional support. Are computers the new counsellors? By Bronwen Morgan


Millennials ( 18- to 34-year-olds) are 16% more narcissistic than any other group of adult consumers. That is according to a recent study by digital agency SYZYGY, which identified a link between narcissistic behaviour and time spent on social media.

A separate study, by The American Psychological Association, has also singled out millennials as the most stressed-out generation, a finding that has also been linked to the fact that the vast majority feel constantly exposed by social media.

More specifically, this has been tied to the idea that we curate our online lives to hide aspects of our personality that we don’t like, and enhance those that we do. 

And according to yet another study, this behaviour carries an emotional and mental toll; researchers at the University of Tasmania asked people (aged between 18 and 55 ) to complete a personality assessment – once as their true self, and once as the self that they present on Facebook (Facebook self). They also measured social connectedness (people’s feelings of belonging and relatedness to others), subjective well-being, depression, anxiety and stress. 

They found that closer alignment between the true self and the Facebook self was associated with better social connectedness and lower stress levels. That is, being true to yourself on Facebook has positive psychological outcomes. 

“If having a greater coherence between one’s Facebook persona [and one’s true self] is associated with lower stress and greater social connectedness, as suggested in the current study, it might be fruitful to consider the potential utility of Facebook in reducing stress and enhancing social connectedness,” the authors, Rachel Grieve and Jarrah Watkinson, wrote. 

“Similarly, if feelings of social connectedness can be obtained through authentic presentation of the self on Facebook, then encouraging sincere self-presentation on Facebook as a means to improve social connectedness should also be considered. 

“As such, it may be prudent for mental health professionals to consider the role of Facebook in their clients’ lives.”

This view is shared by the authors of another recently published study, which looked into “psychiatry in the digital age”. Four academics from the University of Cambridge and Stanford University argue that “data from social networking sites should become a high priority for psychiatry research and mental health care delivery”. 

Evidence suggests that 92% of adolescents go online daily, and disclose considerably more about themselves online than offline; all these details are stored as time-stamped digital records dating back to when the user first joined. Given the amount of information recorded, Facebook data is an “unprecedented resource”. 

“Working with Facebook data could further our understanding of the onset and early years of mental illness – a crucial period of interpersonal development,” the authors wrote. They also point out that the “diminishing digital divide” has allowed for a broader sociodemographic group to access Facebook, including homeless youth, immigrants, people with mental health problems and older people, therefore enabling greater contact with traditionally hard-to-reach populations. 

Along the same lines, but using visual rather than text data, another set of academics, from Harvard University and the University of Vermont, have applied machine learning tools to identify markers of depression in Instagram posts. 

Statistical features were extracted from nearly 44,000 Instagram posts (from 166 individuals) using colour analysis, metadata components and algorithmic face detection. Resulting models reportedly outperformed general practitioners’ average diagnostic success rates for depression, even when the analysis was restricted to posts made before depressed individuals were first diagnosed. 

More specifically, photographs posted by depressed individuals were more likely to be bluer, greyer, and darker. Human ratings of photo attributes (happy, sad, and so on) were weaker predictors of depression, and were uncorrelated with computationally generated features. 

At the other end of the patient pathway, a computer algorithm has been designed that can deliver supportive messages to people undergoing stress, or otherwise in need of emotional support. 

Judith Masthoff, a scientist at the University of Aberdeen, ran experiments where participants were asked to imagine different stressful situations and then choose the messages that they found most supportive. Their responses were built into an automatic system that could then combine the most appropriate types of support, according to what the person was going through. 

Masthoff, found there was a gap between what people say about receiving emotional support from a machine, and how they behave in practice; when they’re asked if they would accept that kind of support from a machine, they would refuse outright, she says. But once the system was established, people relied on it as if it were a person.