Postdoc at UC Berkeley. Interested in all things social reasoning, cognition, modelling, and philosophy of science 🤓
https://manikyaalister.github.io/
Manikya Alister
Loading...
But they struggled to adapt to situations where info was sampled randomly or to be deliberately misleading. Even when told the teachers intent, were given feedback, and could practice being deceptive teachers, they could not sufficiently calibrate their learning in non-helpful contexts.
People were very good at learning from those they thought were helpful and tended to default to this "helpful" assumption when they were given no explicit information about the teacher's intent.
We interpreted these findings as adding to a growing literature showing people have a bias to assume helpfulness in most contexts, enabling efficient learning most of the time, but that can also leave us vulnerable to exploitation if we do not sufficiently adapt when we should
that's what we need. anyhow, if i studied this, i'd think more carefully about how to choose a meaningful control group. and try to have a good theory of under what conditions people seek advice, from whom, with what expectations, and when they actually take it.
