Max van Duijn nominated for Discoverer of the Year 2019
Every person on earth can read another person’s mind. Not in the way psychics or witches do, but by putting themselves in the shoes of others and considering how they perceive the world. This kind of empathy greatly facilitates communication and interaction. Max van Duijn studies this phenomenon to increase our understanding of human communication and cognition, and to improve the capabilities of artificial intelligence.
Everyone is a mind reader – yes you too
Picture yourself at the breakfast table with one of your family members. You have just used the butter, which is still in your hands. The other person at the table has a slice of bread on their plate without anything on it. In anticipation, he or she looks at you. You decide to pass them the butter. ‘In a way you’ve just read someone’s mind’, says linguist and cognitive scientist Max van Duijn. ‘Probably without even giving it too much thought.’
The ability to read the minds of others is distinctly human. That is, even though some birds and mammals can do it as well to a certain degree, none of them can do it in such a complex fashion as humans. Consider for instance a more intricate situation than the one with the butter. Say someone is throwing a party and you are helping with the table setting. A thought might enter your mind that goes something like: ‘I believe that Jim thinks that Wim suspects that Pim does not like him’, and you may decide not to give Wim and Pim seats next to each other. ‘We call this multiple-order intentionality, a type of thought process in humans that I am trying to understand through observations and experiments,’ says Van Duijn.
During his PhD research, Van Duijn found that this skill works differently in models and laboratory settings than in the real world. ‘Models and theories usually portray the process in a very abstract manner,’ he explains. ‘They involve propositions such as A knows that B wants C to ask B to give something to A. In real life, we rarely consciously work with such propositions.’
Humans rather have a database of former experiences at the ready, that they use to interpret a new situation. This database starts to form very early in life, which Van Duijn illustrates with an example from the animal kingdom: ‘Some species of crows are very skilled in fooling their peers. They might bury some food in the ground when they know another crow is watching. Then, when the other crow is distracted, they dig up the food and bury it somewhere else – thus fooling their competitor,’ says Van Duijn. ‘However, the only crows that do this are the ones that have a history of stealing food themselves. They have stored this experience in their database, which they are able to use later on. We humans have the advantage that we can acquire such experiences not only from our own lives. Through language, and stories in particular, we can store the real or imagined experiences of countless of others in our database.’
‘If we want digital assistants to improve, we have to obtain a better understanding of the role of empathy in human communication,’ says Van Duijn. ‘We have to go beyond understanding language in a literal sense towards understanding the person you are communicating with.’
Together with fellow researcher Tessa Verhoef, Van Duijn founded the Creative Intelligence Lab (CIL) at Leiden University. This lab brings together researchers with an interest in both artificial and ‘natural’ intelligence who have a creative and out-of-the-box attitude towards science.
The first CIL research project was performed at Lowlands, a music festival that doubles as a make-shift research facility. The project was called ‘Lowlands Whispers’ and involved people viewing a story on film and then retelling the story in front of a camera. Other people then got to see the retelling of the story, had to retell it again in front of a camera, and so on. ‘This way, a chain of stories formed,’ Van Duijn says. Although a lot of coding and analysis is still required, he is keen to share one preliminary result: ‘We see signs of grammaticalization, meaning that along the chain certain rules pop up that govern how information is shared. As the chain moves on, communication increases in efficiency’, he says.
One goal of Van Duijn’s research is to improve our understanding of human communication. Another is using this knowledge to improve a type of artificial intelligence many people use on a daily basis: digital assistants such as Alexa, Cortana, and Siri. These are pretty good at converting speech into a written language they can work with, but they are terrible in terms of empathy. ‘If we want digital assistants to improve, we have to obtain a better understanding of the role of empathy in human communication,’ says Van Duijn. ‘We have to go beyond understanding language in a literal sense towards understanding the person you are communicating with.’
Max van Duijn (Leiden, 1984) is an assistant professor at Media Technology, part of the Leiden Institute of Advanced Computer Science (LIACS). His PhD research revolved around so-called multiple-order intentionality in human communication (A believes that B wants that C thinks, and so on). Van Duijn likes to perform language and cognition experiments – all very ethically, of course – on his two-and-a-half-year-old daughter. On an average day you may find them fooling around with candy which is hidden and rehidden (see the crow story above), or in serious conversation about who does or does not know if someone else does or does not know something.