We focus on prediction, not fitting: we make a priori predictions from independent data, and focus on across-task generalization, rather than just fitting a model to a dataset. We think it's a nice new model/approach that has something useful to add, and hope meta-cognition researchers agree.

We argue that if you use a model of memory representations, you can just directly read out the strength of this representation (how strong the memory match signal is for a target item) and predict how confidence relates to entire error distributions as well as novel confidence-accuracy dissociations

New preprint where @mmrobinson93.bsky.social and I jump into the literature on meta-cognition (hopefully in a useful way!): osf.io/preprints/os... We show that a simple memory model (TCC) can be straightforwardly adapted to make predictions about confidence #neuroscience #psychscisky

Excited to start building all the cool demos in @bjbalas.bsky.social’s book!