Andrew Howes discusses the mechanics of human decision making to understanding human behaviour. More specifically on the potential of computational rationality as a means of linking theories of human information processing mechanisms to theories of rational behaviour.
Interview with Andrew Howes, Professor of Computer Science, following his inaugural lecture
Most of us use new technology regularly and in a variety of ways, such as to buy products online and speak to friends and acquaintances via social media. But how many of us stop to think about how our behaviour changes – or is manipulated – by our interaction with and through computers?
Andrew Howes has built an impressive career out of thinking about these sorts of questions. Professor of Human-Computer Interaction in the School of Computer Science, his work is shaped by a desire to investigate the idea that the mind is computational.
‘I’m interested in how people interact with computers, which is a topic that is increasingly relevant in today’s society,’ says Andrew, who recently delivered his Inaugural Lecture, entitled ‘Computational rationality: linking mechanism and behaviour through bounded utility maximisation.’
His research interests range from how technology shapes our social relationships and how we make purchasing and healthcare decisions through to how we control the movement of the mouse across the computer screen and how we use a touch screen.
‘Each of these interests is informed by theories in behavioural sciences, such as theories of human perception and motor control and theories of social interaction,’ explains Andrew.
‘Computer science has a big impact on the world; in particular through technology, which is all around us. The impact of the Internet, for example, is extraordinary, but computer science is also a means of explaining human behaviour. And that’s an aspect that is less well known. This is really my core interest – how theories of human behaviour can be formalised as computation. As a society we are now happy to talk about the brain as being an information processor: how the various functions work together as a system that generates behaviour can be informed by computer science.’
Much of human behaviour might be explained as a response to the question ‘what do I do next’? So, for example, if you’re offered the choice between an apple and a blueberry muffin, you have to make a decision, and that involves many aspects of neurological information processing, from the colour and shape of the two foods to your goals and desires in relation to the apple and the cake.
Where things get interesting is when – just as you’ve decided to choose the apple over the muffin because it’s healthier – a third choice is thrown into the mix in the form of a cake that’s even stodgier but not as visually attractive as the muffin. In this situation, we sometimes think: ‘Maybe the first cake wasn’t so unhealthy after all,’ and eschew the apple for the muffin.
This is known as a ‘preference reversal phenomenon’, which suggests that we don’t make independent evaluations of choices, but instead rely on context-dependent decision-making. In other words, we’re not rational!
Andrew disagrees, however. ‘This is something that has puzzled behavioural scientists for 40 years and lots of people use it as an example of how we’re irrational. But we think that’s wrong, because mathematical analysis has shown that if you are faced with choices about gambling odds, switching choice when the context changes can lead to higher returns.’
Yet businesses are increasingly aware customers’ behaviour can be manipulated, and this influences the way online retail sites are set up and how apps are developed.
Andrew, whose research group collaborates with colleagues in the UK, Europe and the US, has done some work around social media, including how people decide what – and how much – information they want to share with online ‘friends’ or ‘followers’ and, too often, with more people than intended.
‘Social media can give rise to social tensions,’ he explains.
Something that surprises people is the compression of social space. One of the primary intentions of social media technology is the idea that conversations can be broadcast to hundreds of people. In this situation, it’s very difficult to be constantly aware of who the audience is. In the early excitement of social media, we concentrated on all the good things about it, but there is also a potential for damage. It’s very difficult to maintain ring fences around conversations.
This, then, poses the question of how you redesign technology so that people can manage their audience effectively. ‘Some of the early work we did on this was before Google+ Circles (which enables users to organise people into groups or lists for sharing) was introduced. But more effective compression of social space is very hard to achieve.’
Andrew, whose work is funded by bodies including the European Commission and the US Air Force, is still excited by academia – nearly 30 years after he almost went into industry instead.
‘When I graduated from the University of Lancaster, I was offered a job with IBM, but was sacked before I even started! They let go their entire graduate recruitment for that year, 1986, because it was a difficult year for them. So a month before I was due to start, I found myself without a job.’
As is often the case, it was serendipitous. ‘I decided to do a PhD as a fill-in, with the intention of getting a “proper” job after that.’
Happily, Andrew never did get a ‘proper’ job. After working at institutions including the University of Manchester and NASA’s Ames Research Center in California’s Silicon Valley, Andrew came to Birmingham in September 2011.
‘I find it really fascinating and it’s a wonderful privilege to be able to study these things on a daily basis.’
‘I look back now and I can’t imagine not having been part of a research group,’ he says. ‘There are several reasons I find my work exciting: Firstly, many of these problems are very poorly understood, so there is a lot of new work to do. Secondly, there are lots of rapid advances being made at the moment, so there’s a real sense of progress across psychology, computer science and biology. And thirdly, computational sciences have a big role to play in explaining human behaviour.
