Institut Jean Nicod

Accueil > Membres > Visiteur·euse·s > RESCORLA Michael > RESCORLA Michael



RESCORLA Michael

Professeur, UCLA

Bayes and the Ocean

             

Parrain au sein de l’Institut

Roberto Casati

Équipe d’accueil

PERCEPTION, MEMOIRE, REPRESENTATIONS

          

              

SITE WEB ǀ CONTACT

                       

Professeur, invité de l’EHESS à l’Institut Jean Nicod du 31 mai au 1er juillet 2022

           

                  

"Ocean state estimation"

Séminaires PhilOcean (Roberto Casati)

2 Juin 2022, 15h30-17h30

Salle Assia Djebar, RdC, 29 rue d’ULM, 75005 Paris

Estimating the state of the ocean is vital for numerous scientific and commercial purposes. The ocean state estimation problem is quite difficult for several reasons, including the enormous dimensionality of the state space along with the relative paucity of data. I highlight some key challenges posed by the estimation problem and analyze how it pervasively impacts our dealings with the ocean.

             

                

"Bayes and the Ocean : Foundations"

Séminaires PhilOcean (Roberto Casati)

9 Juin 2022, 15h30-17h30

Salle Assia Djebar, RdC, 29 rue d’ULM, 75005 Paris

Bayesian decision theory, a mathematical framework for modeling reasoning and decision-making under uncertain conditions, is widely used to tackle the ocean state estimation problem. I elucidate foundational aspects of the Bayesian approach. I then argue for its normative and practical superiority to rival frameworks.

             

             

"The Kolmogorovian Representational Theory of Mind"

Colloquium de philosophie de Jean Nicod

10 juin 2022, 11h

Salle de réunion de l’Insitut Jean Nicod

The core notion of Bayesian decision theory is credence --- a quantitative measure of the degree to which an agent believes a hypothesis. What are the “hypotheses” ? In other words, what are the objects of credence ? One popular view holds that the hypotheses are sets of worlds. Another popular view holds that the hypotheses are mental representations. I argue that neither view fits well with empirically successful Bayesian models of perception, motor control, navigation, and other low-level psychological domains. I argue that these low-level domains are best handled by a novel view that assigns credences to sets of mental representations.

             

                

"Bayes and the Ocean : Applications"

Séminaires PhilOcean (Roberto Casati)

16 Juin 2022, 15h30-17h30

Salle Assia Djebar, RdC, 29 rue d’ULM, 75005 Paris

I survey how contemporary oceanography uses Bayesian decision theory to estimate ocean state. I focus on the ensemble Kalman filter as an exemplary application of the Bayesian framework. I discuss how the Bayesian framework enables us to fuse available data with prior knowledge (e.g. prior knowledge of ocean dynamics). I argue that the Bayesian framework underwrites empirically and pragmatically fruitful probabilistic representations of the ocean.


EHESSCNRSENS