13.00-13.30: Kevin Zollman, Epistemic Social Dilemmas
13.30-14.00: Michiel van Lambalgen, “Going beyond the information given” and rationality
14.00-14.30: Robert van Rooij, Lying with a kernel of truth: Propaganda with generics
15.00-15.30: Andrés Perea, Incomplete information and common belief in rationality
15.30-16.00: Jan van Eijck, Public Lies and How to Recover From Them
16.00-16.30: Richard Bradley, Learning from Experts
17.00-17.30: Jakub Szymanik, Cognition and Decision Making
17.30-18.00: Paolo Galeazzi, On (context-dependent) rationality
Epistemic Social Dilemmas: Social dilemmas like the Prisoner’s Dilemma and Tragedy of the Commons are well known problems for the social organization of economic life. In these complex social situations, individuals who pursue their own self interest will make the group as a whole worse off. While well studied in situations of instrumental rationality, they have been less studied in the context of epistemology. In this talk I will explore some epistemic social dilemmas and discuss how various solutions might work in the epistemic setting.
“Going beyond the information given” and rationality: Jerome Bruner famously argued that the incompleteness of the information given by our senses requires cognitive ‘coding systems’ to enable information processing.
In essence, this observation is what led Kant to introduce his ‘synthetic a priori’ principles for human cognition. We will discuss the resulting concept of ‘constitutive rationality.’
Lying with a kernel of truth: Propaganda with generics: Game theory seems to be the natural theory to study persuasion and propaganda. However, the idealisations made in this theory predict that persuasion is hardly ever possible. Instead, persuasion and propaganda are possible only because of our cognitive limitations and biases (cf. Tversky and Kahneman). The representativity-bias, in particular, is responsible for the successful use of generic sentences in propaganda. A formal model of this bias will be presented, and its (mis)use for generic sentences about social groups will be discussed.
Incomplete information and common belief in rationality: In games with incomplete information, players face uncertainty about the opponents’ utility functions. In this talk I will discuss how to formalize common belief in rationality for this class of games, by extending the standard epistemic model with types in an appropriate way. I will also present an algorithmic characterization of common belief in rationality, which we call generalized iterated strict dominance. If belief hierarchies on utilities are fixed, it turns out that our concept is behaviourally equivalent to interim correlated rationalizability. (Joint work with Christian Bach.)
Public Lies and How to Recover From Them: The talk gives a formal analysis of public lies, explains how public lying is related to public announcement, and describes the process of recovery from public lies. The aim is to give a formal picture of the effects of brainwashing by a repeated stream of public lies.
Learning from Experts: In the face of disagreement in the expressed probabilistic opinions of one or more experts on some proposition, it has been suggested that we should revise our beliefs by adopting a linear average of them. Is such belief revision compatible with Bayesian conditionalisation? In this paper I look at situations in which full or partial deference to the expressed opinions of others is warranted to consider what Bayesianism and linear averaging respectively require of us. I will conclude that only in trivial circumstances are the requirements imposed by the two compatible.
On (context-dependent) rationality: In most of economics, rationality is defined as expected utility maximization, and agents are assumed to maximize expected utility. This view traces back to the works by de Finetti, von Neumann and Morgenstern, and Savage, and it still represents the dominant approach in decision theory, game theory, etc. In the talk I will criticize the identification of rationality with expected utility maximization, and I will argue in favour of a more permissive perspective, based on the notion of context dependence.