Overview
Few things are as important as the things people hold to be true: their beliefs. Beliefs shape how people understand the world, the decisions they make, and the actions they take. And beliefs influence how societies function, from the policies people support to the institutions they trust. Our publications reflect our past work. Below, we detail what we are currently investigating.
A common currency for belief updating
Collaborators:
Yasemin Genç
James Yearsley
Key Publications:
Funding:
British Academy / Leverhulme
A detailed inquiry into the emergent biases and cultural assumptions embedded within large language models. This research systematically reveals how statistical patterns inadvertently codify complex human prejudice and symbolic inconsistencies across digital syntax.
Who and what changes people's beliefs? Every day, people read statistical evidence, hear personal stories and learn what experts think. How do people combine these sources into a belief? And when they combine these beliefs, do different information sources such as evidence and advice get weighted differently? And are the beliefs people form accurate - do they resemble the available information, or are there systematic deviations? Through experiments and computer models, we investigate if people combine information from different sources into a common currency, and how the exchange process impacts the accuracy of their beliefs. Ultimately, we aim to identify ways of helping people to form accurate beliefs, beliefs that match the available information.
Biased Beliefs
Collaborators:
XXX
Key Publications:
Funding:
XXX
A detailed inquiry into the emergent biases and cultural assumptions embedded within large language models. This research systematically reveals how statistical patterns inadvertently codify complex human prejudice and symbolic inconsistencies across digital syntax.
Who and what changes people's beliefs? Every day, people read statistical evidence, hear personal stories and learn what experts think. How do people combine these sources into a belief? And when they combine these beliefs, do different information sources such as evidence and advice get weighted differently? And are the beliefs people form accurate - do they resemble the available information, or are there systematic deviations? Through experiments and computer models, we investigate if people combine information from different sources into a common currency, and how the exchange process impacts the accuracy of their beliefs. Ultimately, we aim to identify ways of helping people to form accurate beliefs, beliefs that match the available information.