Turing Fellowships for psychologists

In a wonderful collision of the old and new a cognitive psychologist will use machine-learning techniques to trace psychological traits through history. Professor Thomas Hills (University of Warwick) has recently been elected a Fellow of the Alan Turing Institute where he will explore how generations of people thought and felt.

Founded in 2015 and based in the British Library, the Alan Turing Institute brings together data scientists from all fields to tackle challenges ranging from engineering, high-performance computing and cyber-security to smart cities, health, the economy and data ethics. Its Fellows are encouraged to work collaboratively and across disciplines, and Hills said his election to the Institute would give him academic freedom to explore an unusual area.

‘My work with the institute combines humanities, economics, history, psychology and statistics to create a historical view of psychological patterns over the last 200 years. It’s very exploratory, but the fellowship gives one a sense of academic and intellectual freedom to explore some pretty big ideas with the other Turing Fellows.’

Hills’s methods combine machine learning with psychological theories as well as economic theories and econometrics. ‘My short-term goals are to ask what historical patterns we can make sense of as cognitive psychologists. That could be about anything – we’ve got projects on risk, sex, death and money, and we’re trying to understand how people thought about these things in a given historical time. But the longer-term goal here is asking whether we can integrate these ideas together in ways that will give us a more coherent sense of what it was like to be a person in a particular time. What influenced the way people felt, such as government policies, epidemics, or economic production, and how did this in turn lead to different outcomes in the future, such as war and changes in ingroup/outgroup thinking.’

Hills has previously explored the evolution of American English using similar techniques to analyse all kinds of texts, including presidential speeches, newspapers and magazines. Describing this work, he said: ‘You can look at the history of American English and quantify it in relationship to how concrete the language is that people were using. That allows you to see how American English has changed dramatically over the last 200 years and has become a much more concrete language where people are talking about real, tangible stuff. The language has changed in all kinds of ways, and you can easily see it if you read something written by Henry James or Hawthorne. It’s a very different type of writing.’

Hills and his team are working towards validating their models and algorithms, using an index of subjective wellbeing created in an earlier project: ‘From the early 1970s there’s been the Eurobarometer, an indicator of people’s subjective wellbeing in different nations, so we can test that against our index. You can also say, OK, it would be really odd if people had really positive feelings during the world wars and you’d question your index, but in fact the lowest points and most dramatic changes in our index happen in world wars. In the 1920s, sure enough, you can see people’s subjective wellbeing by our index is extremely high, but of course this immediately tanks right after the Great Depression starts.

It makes sense by the numbers and it makes sense intuitively, so what can we do next? We can ask does it correlate with GDP or longevity or democratisation.’

Professor Zoe Kourtzi, Chair of Experimental Psychology at the University of Cambridge, is another of this year’s new Turing Fellows. She is working to develop predictive models of individual health based on machine-learning approaches that use multivariate and longitudinal studies.