Running a video recruitment company for almost 5 years has created a professional obsession about the correlation between image (video) and identity (personality). In marketing and communication, we have already recognised that the internal identity and external image have blurred into a holistic reputation. Recently I stumbled upon new research demonstrating that this is also the case with individuals and their visual features displayed on Facebook. Researchers from California Institute of Technology, University of Cambridge and Stanford University (among others), did research about profile pictures and aimed to find the answer to the following questions:
Do profile pictures contain information about their owners’ personality?
Which, if any, of the profile picture features are indicative of their owners’ personality traits?
Now, how cool is that? And when you realise that a video is basically a series of pictures in a sequence of frames per second, you can probably imagine that it did not take long for me to get excited. In our Cammio video recruiting platform we have already been using verbal reasoning (basically the extract of audio into text and analysis of the verbal content) patterns to create BIG5 personality profiles. We have also started R&D projects with selected assessment partners in our partner ecosystem to use micro-expressions or facial mimics as input for a personality profile. To date, this type of input is only used to measure emotions which is basically meaningless in recruitment since it only is a specific moment in time and therefore a lousy predictor of a job match. The research report I found does not even require movement in the face, but uses profile stills: WOW!
Back to the research by these brilliant researchers at these reputable academic institutes… They have actually found a correlation between pictures and personality!
Assessing personality from profile pictures: Computers beat humans
Previous literature in personality psychology reported that users can already make accurate personality impressions from the information displayed in social network user profiles. The current research actually demonstrates that computer-based classifications are significantly more accurate than averaged human-based classifications for assessing the personality traits of Neuroticism and Extraversion. This is also due to the fact that only Facebook profile pictures were used as input for the research. The three remaining BIG5 personality traits, agreeableness, openness to experience and conscientiousness (traditionally a great indicator of job performance) have a lower correlation with Facebook profile pictures.
This isn’t the first time where computers beat humans at their own game:
We’re only getting started…
The researchers have found that agreeable individuals and extroverts tend to have warm coloured profile pictures and to exhibit many faces in their portraits, mirroring their inclination to socialise. Neurotic people on the other hand have a prevalence of pictures of indoor places. There is still room for improvement, the research in this case only uses one profile picture per individual. It is also limited to Facebook profile pictures and the dataset of individual users is still limited. However, this can only mean that following research should be expected to find higher correlations and maybe match with more of the BIG5 traits.
Data privacy protection is key
This is obviously truly exciting and brings many possible applications in marketing or online dating, but also in human resources. Unconsciously we are sharing more information on social networks than we think. A simple upload of a profile may unveil some aspects of our personality. And although it is great that we can automatically analyse more and more data, it is also a good thing that data privacy legislation such as EU-GDPR will protect individuals from companies using our profile. Technology such as this should only be made available in the right context, with the right justifiable purpose and, most importantly, to the benefit of the candidate.