Humans are social beings. If our day-to-day interaction is to run smoothly, it is important that we understand why other people react as they do. People with an autistic disorder, for example, do not have this skill. PhD candidate Ben Meijering carried out research in the Department of Artificial Intelligence of the University of Groningen, examining how people learn to understand each other. He will defend his thesis on 6 June 2014.
Knowing when your opponent is bluffing is an essential skill for poker players. A raised eyebrow can mean either a good card or a bad one. Or it could mean that your opponent knows that you are observing him and consciously raises an eyebrow to throw you off the scent.
Two schools of thought
‘Academics have spent many years studying the way we try to second guess each other’, says Ben Meijering. The underlying principle is called Theory of Mind. There are two main schools of thought. The first claims that Theory of Mind evolves from basic brain functions. Being able to think logically enables us to reconstruct another person’s thought patterns. The second school of thought sees this as a separate quality of the brain, supplementary to our powers of logical thought.
Meijering conducted experiments to test both schools of thought. He devised a game in which players had to choose between two alternatives before every move. Two players took turns to move, after trying to anticipate their opponent’s next move. ‘This turned out to be quite tricky in practice’, explains Meijering. His research involved a trial person playing against a computer. ‘The test subjects invariably thought just one move ahead.’ But after Meijering advised the test subject to look at things from his opponent’s perspective, the subject took more notice of the countermove. ‘This goes to show that social skills, such as negotiation, can be trained.’
In a second experiment, Meijering built a mechanism into the game that would automatically execute the last move in a logical manner. ‘The human players responded to the situation differently, depending whether the mechanism had been added or not.’ He thinks that this shows that there is more to Theory of Mind (which people need when trying to second guess someone) than logical reasoning.
The ability to look at things from another person’s perspective does not appear to be a separate module. As well as saying something about how our brains work, this is also useful information for designers of smart software. Meijering: ‘If you make a mistake in a programme, you see the message: ERROR. A programme that can put itself into the user’s shoes would say: “You’re nearly there. Just think about …”. This would make things so much easier.’
The notion that empathy may be trainable is an important conclusion. He is not yet sure whether this could benefit children with autism. ‘But just imagine; you get a new job that involves a lot of negotiation, but you’re not very good at it. Your initial reaction could be “I can’t do it” before simply giving up. My research shows that targeted training could help you to improve this skill.’
Meijering recently started working as an Artificial Intelligence Specialist at CrowdyNews, a young Groningen software company that develops intelligent software to link messages from the social media to news items on newspaper websites, for example. The software filters messages from social media such as Twitter and Facebook, and Instagram pictures, and posts them next to the relevant article via a widget.
Mini libraries filled with free, second-hand books sometimes seem a bit lonely, lost in the no man’s land between good intentions and uselessness. But nothing could be further from the truth, explains PhD student Anouk Schippers.