So cool!

If you read the paper and have any thoughts we'd love to hear them! Ps. the copy-editors accidentally changed the exact month at which infants succeed in a couple of papers. We'll post the fixed version to the same link as soon as we have it.

We think this is a major part of development that's partially learned through parent-child conversations! (and it connects to ideas that Lio Wong, Dan Greco, and others have made about adult cognition!).

the idea is that adults are constantly building restricted scope models--simplified context-specific models that make reasoning tractable. Infants don't know how to do this, so they only succeed in tasks where the experiment itself gives them the restricted model (e.g., through familiarization)

There's two puzzles we've been thinking about: Why are infants so smart in the lab and seemingly dumb in real life? And how can the Bayesian reasoning models that explain infant and adult social cognition scale up to the messiness of the real world? We think both puzzles have the same solution

The paper tries to unify a lot of studies on social development by breaking them down into basic computational primitives and processes. Beyond the review side we have a couple of new ideas in there that I'm excited about.