We give an analysis of the Monty Hall problem purely in terms of confirmation, without making any lottery assumptions about priors. Along the way, we show the Monty Hall problem is structurally identical to the Doomsday Argument.

"The first two lectures place the alternative I defend -- a kind of pragmatic realism -- in a historical and metaphysical context. Part of that context is provided by Husserl's remark that the history of modern philosophy begins with Galileo -- that is, modern philosophy has been hypnotized by the idea that scientific facts are all the facts there are. Another part is provided by the analysis of a very simple example of what I call 'contextual relativity'. The position I (...) defend holds that truth depends on conceptual scheme and it is nonetheless 'real truth'. "In my third lecture I turn to the Kantian antecedents of this view, explaining what I think should be retained of the Kantian idea of autonomy as the central theme of morality, and extracting from Kant's work a 'moral image of the world' that connects the ideals of equality and intellectual liberty. In this lecture I defend the idea that moral images are an indispensible part of our moral and cultural heritage. "In the final lecture I defend the idea of moral objectivity. I compare our epistemological positions in ethics, history, analysis of human character, and science, and I argue that in no area can we hope for a 'foundation' which is more ultimate than the beliefs that actually, at a given time, function as foundational in the area, the beliefs concerning which one has to say 'this is where my spade is turned'. In ethics such beliefs are represented in moral images of the world.". (shrink)

The well known Monty Hall-problem has a clear solution if one deals with a long enough series of individual games. However, the situation is different if one switches to probabilities in a single case. This paper presents an argument for Monty Hall situations with two players (not just one, as is usual). It leads to a quite general conclusion: One cannot apply probabilistic considerations (for or against any of the strategies) to isolated single cases. If one does that, one cannot (...) but violate a very plausible non-arbitrariness condition and is led into a Moore-paradoxical incoherence. Even though arguments for switching are correct as applied to series of games, they don’t say anything useful about what rationality demands in a single case. (shrink)