Interview Eric-Jan Wagenmakers, Nov 12, 2021, During the symposium Nothing but the Truth, November 1 in Groningen, Jan-Willem Mantel and I, Mariëtte Oosterwegel, led a workshop What to Ask? following a lecture by Eric-Jan Wagenmakers, professor of Mathematical Psychology at the University of Amsterdam’s Faculty of Social and Behavioral Sciences. Afterwards, I talk to Eric-Jan about this workshop Learning to ask questions.
Who are you and what are you researching?
I am originally trained in experimental psychology and subsequently moved into mathematical psychology. There I spent a lot of time making formal models for cognition and behaviour. Gradually, I became more and more methodologically interested, especially in Bayesian statistics. Pierre-Simon Laplace (1749 – 1827) was actually the first true Bayesian and described that statistic as ‘common sense expressed in numbers’. And that’s exactly what is it. Bayesian thinking involves an enormous amount of work, both experimental research and model making. It also deals with questions such as: How do you draw conclusions and how do you conduct good research? And that in turn has to do with my interest in Open Science and transparent scientific practice. We recently wrote an open access article in Nature Human Behavior about this: Seven steps toward more transparency in statistical practice. The nice thing about methodology is that you can apply it very broadly to all kinds of different scientific fields. I am particularly interested in application to medical sciences, especially neurobiology, and to the philosophical side of Bayesian statistics.
Workshop What to Ask?
During the workshop What to Ask? we challenged the 30 participants to ask Eric-Jan questions about his lecture Truth and Simplicity. The participants received a form in advance in which 3 goals (gain insight, check credibility and explore relevance) and 3 reasons (incompleteness, inconsistency, association) were described. There were a lot of questions and we noticed that the participants really appreciated that there was enough time to ask questions and pursue deeper questions about some answers. And that those good questions were ‘rewarded’ because Eric-Jan carefully answered them.
This workshop has given me a lot of insight, op top of the fact that it strikes your ego when people ask so many questions about your lecture. The questions were of a high level and prompted me to rethink certain things. Some issues I had unconsciously insufficiently brought forward and others I had completely missed. The feedback from the question session has helped me to describe this research. I therefore found it very useful and also a lot of fun to do. The workshop not only made the participants think, but also myself and the workshop format worked very positively. When you’re writing a piece and also when you’re talking to other people, that’s a good way to think. While you are communicating you think about what you are doing. Questioning is a very underexposed skill.
Bayesian Thinking for Toddlers
You have written a children’s book about probability and probability, Bayesian Thinking for Toddlers. What prompted you to write that?
It started with a review I wrote for a book Bayesian Probability for Babies. When I went through that with my son, he missed the dinosaurs. Then I was challenged by one of my PhD students to write a book on statistics with dinosaurs. I worked with a great illustrator, Viktor Beekman, and together we came up with Bayesian Thinking for Toddlers. It took me a long time. Such a book forces you to tell something as simple as possible and other researchers should do the same. It is very instructive to bring your research back to the core. You have to leave out many important concepts and it should also remain fun to read. It has become a book with many layers, so that it is interesting even for experts of statistics. The nice illustrations also make it a fun children’s book. Chris Ferrie has written a whole series of children’s books on complex topics, including Quantum Physics for Babies, for example. These are good examples for mathematicians and other researchers of how you can make your work attractively visible.
Dutch Pancakes
On the internet I came across Dutch Pancakes when I googled you. What about those pancakes
I am currently writing a course book introducing Bayesian Statistics together with Dora Matzke. One of the standard examples I use is whether pancakes are fried with bacon or not. You have to predict: will the next pancake also be with bacon or not? And as you see more pancakes, you can say something about the chance whether the next 10 pancakes also have bacon. You can also say something about the underlying probability that I generally bake pancakes with bacon etc. The textbook is not finished yet, but I hope to make it publicly available very soon. The use of a metaphor, such as the probability of a pancake with or without bacon, also helps to make complex matters, such as Bayesian statistics, accessible.