Making sense of fallacies

An interview with Robert van Rooij

1 September 2014

Robert van Rooij is professor of Logic and Cognition at the ILLC. He is best known for his work in the formal semantics and pragmatics of natural language (e.g. comparatives, conversational implicatures, presuppositions) and philosophical logic (including vagueness, truth, and conditionals). He also has a keen interest in the evolution of language.

“Before I studied philosophy, I did horticulture. My father was a grower, and I thought I would step in his footsteps. During the weekends and the vacations I always worked in horticulture at home, until I was 20 or so. Many people were working there in the summer and on the weekends, and it was a lot of fun. But during that study I found out that I wanted to do something else. I was always interested in history and philosophy. I thought history was the worst study to get a job with, so I studied philosophy (which was maybe even worse). The two things I found most interesting there were philosophy of science and philosophy of language. But for me, studying philosophy of science did not make a lot of sense because I did not study a science before, so I decided to go for language and to study linguistics as well.”

“I try to change topics every once in a while. One feels freer if you are relatively new in the field. I like it that when you are still a bit naive on a new topic you dare to say things that somebody who has read all the literature on the topic would not dare to say. If you say something that is really absurd, you will find out soon enough. But if you try a lot of things, eventually something comes out that makes sense.”

“Causality is for me such a topic. I have never written anything about it, but I am going to teach about it, and hope to learn a lot from that. I will teach the course together with Katrin Schultz, my wife, who is more of a specialist on this topic. I can work together well with her. We know each other and we have the same academic interests. But she is actually also very critical, so if something I write is not good, she will be the first one to say so. I hope I do the same thing with her. The papers I wrote with her are my best quoted papers.”

“I started on vagueness because I was invited to write about it. I do not know why they asked me, but it was actually good. What is problematic about vagueness is that every natural language predicate seems to be vague. For instance, what is the meaning of ‘red’? Where does ‘red’ start and where does it end? To model that is difficult. If you use logic for the analysis of natural language, you try to translate each natural language sentence into a logical formula, and then you know what follows from which sentence. That is how you get the meaning of the sentences. The problem with vagueness is that you also want to hold on to certain intuitive principles. For instance, if somebody is small, and somebody else is one millimeter taller, then this second person should be small as well. But then you very soon obtain the result that everybody is small, and that everybody is not small. That is obviously a paradox. To get rid of that you have to think about what intuitive principles of logic to give up. That is what a lot of people in philosophical logic are thinking about. I have come up with one solution, and I am working that out. It turns out that this solution works for other paradoxes as well.”

“I am not only interested in how to model meanings, but I am also interested in how meaning evolved. I want to know how a language evolves so that it can compositionally account for meanings in communication. I am using evolutionary game theory, where you explain some of the phenomena that you find in language by saying that these properties have evolutionary advantages. You try to come up with a simulation on the computer and then the simulation shows that there is an advantage, for example of compositionality, so that compositional languages will dominate other languages.”

“With vagueness, the problem is that it is not helpful – you can use evolutionary game theory to prove that, under ideal circumstances, vagueness will never be beneficial. So, how can it be that language is vague? To explain that, you have to weaken certain idealistic assumptions standardly adopted in (evolutionary) game theory. What do you have to give up in order to have vague languages evolve? It turns out that the assumption that people make in behavioural economics to explain their data are very similar to the assumptions we need. When you put those in your simulations, vagueness will actually evolve.”

“When I did my PhD (which dealt with the analysis of propositional attitudes and anaphora) I was in an institute for computational linguistics, the Institut für Maschinelle Sprachverarbeitung in Stuttgart, Germany. I learned a lot of linguistics, and when I came to Amsterdam I actually did a lot of linguistics. But about 5 years ago, I realized that issues like finding out what the meaning of ‘even’ or ‘only is, were not really what I wanted to do. I went back a bit to philosophy, and a bit away from formal semantics. I am now using non-standard logics to investigate reasoning, and game theory and decision theory to investigate language. You might think that game theory and decision theory are really different from logic, because standard logic is qualitative, whereas game theory and decision theory use numbers. But I always thought that it is not just that you can use insights from game theory and decision theory to reason about what you should do and what you want others to do, but that game theory and decision theory are actually part of logic.”

“One prominent use of language is to convince others to do certain things. That is the area where probability theory and non-monotonic logic are more interesting than standard logic. In argumentation theory there is a lot of research about fallacies. You can use logic to show why a fallacy is not good reasoning. But of course we all do use these types of reasoning, and there must be a reason for that. One reason could be that we are all stupid, and that is, in a sense, true. But standard logic does not work well in such situations where you need to come up with alternative models. In decision and game theory there are more ways to slightly change things, to model people that are less smart, remember less and so on. But even more importantly, many fallacies do in fact seem to work. Using probabilistic reasoning you can actually show that, even though these fallacies do not yield logically valid conclusions, they often do make sense. I think that probability theory is going to be useful in pragmatics, because it can explain better than standard logic why arguments traditionally called fallacies are actually used so much.”

“In linguistics you only study pragmatics for cases in which the speaker and the listener agree with each other on what they want. But of course, that is the ideal case, and typically if you talk with someone, he or she does not share all your goals. Game theory was developed to study situations where people do not have the same interests. It seems only natural that you use this for communicative situations, as a way to enlarge the scope of pragmatics, which is also something that I try to do in my research.”

by Gisela Govaart, September 2014

Published by  SMART Cognitive Science