Summary of round table discussions around three themes:
When discussing what the distinction
Risk vs. Uncertainty actually means, we found that mainstream economics doesn't really recognise it and that economists prefer to use the term
ambiguity instead of
uncertainty. Besides, we found that economists tend to ascribe specific phenomena to a person's risk attitude, neglecting the circularity in ascribing a behaviour to another presumably
underlying behaviour. But how to cope with the fact that we cannot tell if a specific behaviour arose from risk or uncertainty if we just observe its outcome? An important impulse in our discussion was the assumption that one can always ask if a particular outcome was shaped by natural selection. Since evolution adapts us to what was there before, there is probably some generalising mechanism in the organism, and this would suggest that evolution adapted us to risk, but not to uncertainty. The next important question is then: Where else than in gambles do we have the risk domain? In other words: We are mostly in the uncertainty realm, but most theories and frameworks tackle either the realm of certainty and logic or the realm of risk, as can be seen by the frequent attempt to apply Savage theory to anything other than small worlds where you can consider all possible events. We came to a point where we agreed that we would benefit from considering different kinds of uncertainty, e.g. the
meta-uncertainties that arise when distributions change over time.
We started our discussion about
Notions of Rationality with the observation that the public seems to be highly interested in if we're rational or not, while psychology has no definition yet for rational cognition. Accordingly, we agreed on that the usual tests of rationality are arbitrary and that assuming that we learn from people's behaviour something about the system would lead us astray. Likewise, we found that Kahneman and Tversky's methodology, that is, experiments that are totally based on choosing and ignore learning, do not account for the fact that rationality is about learning possibilities. Moreover, an important suggestion in our discussion was that seeming violations of logical rules often arise from social norms or from ecological rationality. Cognition has evolved in a certain environmental structure with certain properties that are violated by experimental conditions. We came to the inspiring idea that people perhaps need the concept of rationality in order to have a measure for functional success in the world.
Our discussion about
Heuristics was driven by the question what we get from modelling people's decision making and if such models should be based on researchers' intuition or rather on separate building blocks. While we agreed on the basic idea that heuristics as simple ways of coping with our environment are the ultimate economizing mechanisms, we identified problems when taking the tension between the individual and the aggregate level into account. Like with rationality, we found that heuristics have a huge learning component. Besides, we agreed on that the path-dependency of the decision situations and their social contexts have to be taken seriously.
Kenneth Binmore (London, GBR), Jean Czerlinski Whitmore (Chicago, USA), Alexander Djamali (Berlin, GER), Susann Fiedler (Bonn, GER), Gerd Gigerenzer (Berlin, GER), Daniel Goldstein (New York, USA), Linda Heimisch (Berlin, GER), Ralph Hertwig (Berlin, GER), John McNamara (Bristol, GBR), Hansjörg Neth (Konstanz, GER), Paul Robinson (London, GBR), Geeske Scholz (Osnabrück, GER), Shyam Sunder (New Haven, USA), Nassim Nicholas Taleb (Port Jefferson, USA), Martine Varda (Düsseldorf, GER)