... does this now mean that young people are even LESS uncertainty tolerant, because they listen more to what others say? We have opinions :) Get in touch! Happy to hear yours!

Also, big shoutout of course to @connectedmindslab.bsky.social 🤗

It also means that notions of uncertainty tolerance require additional explanation. It may be that younger people feel unsure about what to choose, which masks as an increased propensity for risky choice at first glance. But at the same time they use social info more, reducing their uncertainty 🤺

This internal uncertainty explains age differences in social influence in our experiment. We manipulated social signals by showing the choices of previous participants to ours. The more uncertain our participants were, the more likely they were to follow others. 🌈 This was true irrespective of age.🌈

Bayesian modelling reveals that younger participants' choices were characterized by greater uncertainty about the utility of choice options, which was distinct from the increased randomness of participants' choices that is usually shown in higher softmax-temperature in younger participants 🤔

We hope to bring more clarity to the notion of uncertainty tolerance among youth 😶‍🌫️. Asking for decisions from description or experience, where uncertainty is low and high, respectively, we demonstrate that adolescents indeed make more risky decisions than adults when uncertainty is high vs low.

Did you manage to steal it?

Yay; andrea @andgr.bsky.social is here now too 🚀 :)

This was deliberate, because how often do we really see exactly what rewards others experience 🤷

But I wonder if the "surrogate" and joint social-nonsocial, probabilistic rewards are the same thing for the mind and shaping behaviour... 🤨

Anyway; exciting stuff 🚀

[...] In every trial, there were 63 alternatives to copying, and the rewards of demonstrators were invisible to our participants, so we needed this surrogate reward.[...]

[...] the policy biasing approach that was suggested here: journals.plos.org/plosbiology/...
We could not model social and nonsocial learning jointly, because the personal rewards were so salient in our task and learning about demonstrators related to the demonstrators' behaviour history only [...]

Uh, thank you for sharing this 😊!
I like this feature-learning approach a lot; it seems so elegant and resonates with us in moving away from "greedy" social-copy heuristics, as social learning is just more dynamic than such models can portray. Our experiment required us to adopt [...]