Spy in Your Pocket

While smartphones have become intimately integrated into the daily lives of most affluent and semi-affluent people, few realize the technology is gaining the ability to analyze every aspect of personality, beliefs, and behavior.

Researchers at Australia's Royal Melbourne Institute of Technology say they are able to predict your personality type based on how your smartphone moves with you over time, the types of calls you make, and the texts you send.

The researchers matched data generated by a smartphone's accelerometer (which tracks movement of the phone) against the responses smartphone owners shared on a standard questionnaire assessing their personality type. Additional monitoring also provided data on how often the smartphone owner made voice calls on the phone, and how often the owner texted using the phone.

Some of the researchers' key results: people with consistent movements on weekday evenings were generally more introverted. In contrast, extroverts exhibited more random patterns, perhaps meeting up with different people and taking up unplanned options. Meanwhile, 'agreeable people' displayed more random activity patterns, and were busier on weekends and weekday evenings than others. Plus, friendly and compassionate females made more outgoing calls than anyone else.

The study's findings also revealed that the way people used smartphones also provides an indication of those who are more conscientious, more neurotic, or more inventive.

"Activity like how quickly or how far we walk, or when we pick up our phones up during the night, often follows patterns and these patterns say a lot about our personality type," says Flora Salim, deputy director of the RMIT Center for Information Discovery and Analytics at the Royal Melbourne Institute of Technology.

According to the research team, this study differs from most previous research in that it goes beyond simply analyzing what people post on social media or who they happen to be in contact with via voice or text. "Previous work using the same dataset has only focused on modeling the relationships between the callers/senders and recipients of voice calls and text messages to predict the personality traits," Salim says. "In our work, we predict human personality traits by combining physical activity intensity data with traditional phone activity, voice call and text data, and validate this with Big Five survey data taken by the users themselves."

Mirco Musolesi, a professor of data science at University College London who also works in smartphone use research, says the Australian research confirms a number of similar studies that have analyzed smartphone use to predict personality.

However, Musolesi believes that moving forward, "the next step for this body of research is to adopt rigorous experimental practices from psychology, psychiatry and neuroscience, for guaranteeing the reproducibility of the results."

"Another key problem of existing work is the generalizability of the results, since they have been obtained for specific populations," Musolesi said, noting  that the Australian researchers relied on a public dataset generated at the Massachusetts Institute of Technology, which tracked movement and use of an Android smartphone. Musolesi said future studies will need to utilize a dataset more representative of the world's population, to support greater generalization about the relationship between smartphones and personalities.

Victor Cornet, a Ph.D. Candidate in informatics at Indiana University – Purdue University who studies the use of smartphones as health and wellbeing sensors, agrees. Unfortunately, Cornet says, academic research rarely has access to "large samples to test research questions pertaining to passive sensing on smartphones.

"For example, the participants in the dataset used in the Australian study were, or had a relative affiliated with a major North American research university."

Perhaps most troubling is the ever-increasing sophistication of sensors, software, and processing power on smartphones, which could easily transform one of the most essential devices in the daily life of many into a relentless snitch on our personalities, politics, behaviors, associations – and more.

"The number and quality of smartphone sensors, both physical, for example, the accelerometer, and virtual, for example, application use, have been constantly improving since the smartphone's early days," says Cornet. "Back then, smartphone GPS sensors needed close to a minute to triangulate an accurate position," Cornet adds. "Today's smartphone geo-location is quasi-instantaneous. New smartphone sensors are also continuously being developed — the new Google Pixel 4's radar sensor being a telling example."

Musolesi fears people living under repressive governments may not fare as well as those in strong democracies. "I envisage serious implications related to the use and impact of these technologies in non-democratic regimes."

Joe Dysart is an Internet speaker and business consultant based in Manhattan, NY, USA.