Research Seminar in Decision and Action Theory: L.A. Paul (Yale) and Hayden Wilkinson (Oxford)

Dear All,

I’m delighted to announce a special double session of the Research Seminar in Decision and Action Theory this week. On Wednesday 21 May, we will have a special session with L.A. Paul (Yale) and Hayden Wilkinson (Oxford).

The event will take place in the ZEPP Seminar Room (M210, Geschwister-Scholl-Platz 1, LMU’s main building) and start at 9:30am. The format will be as follows:

9:30 – 11:00: Hayden Wilkinson (Oxford): Regret aversion: A new consideration for rational choice?
Break (going for coffee together)
11:30 – 13:00: L.A. Paul (Yale): Knowledge in LLMs
Lunch

I append abstracts at the bottom.

The seminar will be held in person, but here is a Zoom link for remote participation:
https://lmu-munich.zoom.us/j/93279453658?pwd=MGRkZ1lOYTVtK2tleWVHbFZ4UFNzZz09 

Title:

Hayden Wilkinson: Regret aversion: A new consideration for rational choice?

L.A. Paul: Knowledge in LLMs

Abstract:

Hayden Wilkinson: Regret aversion: A new consideration for rational choice?

I discuss preferences over gambles that seem intuitively rational, and that many real-world agents hold, but that deviate from orthodox normative decision theory. These preferences even deviate from the various less orthodox decision theories designed to accommodate risk aversion. This is because such preferences exhibit not risk aversion, but insteadregret aversion: a preference for one’s chosen option to be more likely to actually turn out better (or perhaps significantly better) than the alternatives. Is regret aversion rational? Plausibly, yes. Beyond mere intuitions about cases, I offer two further motivations. The first is that regret aversion is needed to reflect at least some concern for doing what is objectively best, and it is plausible that such a concern is rational (perhaps even rationally required), especially in moral decision-making. The second motivation is that regret aversion correctly diagnoses and treats what’s wrong with so-called ‘fanatical’ verdicts in cases of extremely low probabilities and extremely high stakes. There are also reasons to think regret aversion irrational: it leads to violations of several widely-held and seemingly plausible principles of rationality. Perhaps these violations constitute a decisive objection to regret aversion. Or, perhaps, regret aversion constitutes a decisive objection to all normative decision theories so far proposed.

L.A. Paul: Knowledge in LLMs

Drawing on philosophical theories of knowledge, I engineer a new concept of LLM knowledge and propose a taxonomy of knowledge that spans biological and artificial categories, from C. elegans to LLMs. The approach lays a foundation for a rigorous account of how knowledge can be realized across animals and machines while at the same time respecting the diverse biological and physical sources of such knowledge.

The seminar will be held in person, but here is a Zoom link for remote participation:
https://lmu-munich.zoom.us/j/93279453658?pwd=MGRkZ1lOYTVtK2tleWVHbFZ4UFNzZz09