CFP: Bayesian learning and reasoning

The deadline for abstract submissions is extended until 15th February 2022. Please share widely.

Keynote speakers

Ulrike Hahn (BBK)

Josh Tenenbaum (MIT)

Nico Orlandi (UCSC)

Description

Everyday learning and reasoning challenges us to make inferences and decisions given limited evidence, time and resources, and under constantly changing environmental conditions. Bayesian inference is a widely used method in cognitive science for modelling learning and reasoning by capturing the probabilistic relations between incoming data and prior knowledge structures and the way they should be combined to generate better predictions. These predictions are increasingly often characterised by proponents of the Bayesian Brain Hypothesis on the subpersonal level of cognition as the ‘brain’s guesses’ about the world. However, this characterisation leaves many open questions to be addressed. For example, how exactly do these guesses interact and integrate with existing knowledge to form novel predictions? In what sense can changes in these predictions over time be considered optimal solutions to everyday learning and reasoning problems? Finally, how does this way of looking at cognition stand to more traditional philosophical and psychological approaches to learning and reasoning? The aim of this 2-day international workshop is to foster a greater understanding of how Bayesian models can support cognitive scientists to form new explanations of the interactions between knowledge, learning, and reasoning under uncertainty. Example topics include, but are not limited to:

• To what extent do Bayesian models explain causal and explanatory inferences and their development in (young) humans and other animals? How should changes in learners’ underlying representations of possible causes and explanations precisely be understood?
• How can the notions of inductive biases, intuitive theories and meta-learning be understood formally? How do they stand to traditional conceptions of theory and induction in the philosophy of science?
• To what extent is Bayesian learning with inductive biases and intuitive theories optimal? What factors does the optimality depend on?
• What kind of content do Bayesian predictions and hypotheses have?
• How do Bayesian models ultimately relate to the debates regarding empiricism and nativism when used to explain the acquisition of knowledge and belief given sparse data?
• What can we learn from the more recent applications of Bayesian models and theory theories to understand the development of creativity, arts, and game-playing?
• What is the scope of Bayesian models of learning and reasoning? Can all aspects of learning and reasoning be accurately described using probabilistic methods?
• To what extent does Bayesian reverse-engineering offer a platform to unify answers to these questions within the cognitive sciences?

Abstracts should be 500-750 words in length and should summarise a paper that can be presented in no more than 25 minutes. Please prepare your abstract for anonymous review and submit it to [email protected] by 15th February 2022. Abstracts should include up to 5 key references. Members of underrepresented groups and graduate students are especially invited to apply.

The event is planned to be held in a hybrid format (partly online, partly in person) at the campus of the Ruhr-Universität Bochum. 

If you have any questions, please contact Nina Poth ([email protected]).

Organisers

Corina Strößner

Krzystof Dolega

Nina Poth