Proceedings of the Annual Meeting of the Cognitive Science Society

Like scientists, children have a sharp sense of when and howto seek evidence, but when it comes to generating causal in-terventions, their performance often falls short of normativeinformation-theoretic metrics such as the expected informationgain (EIG). We looked at whether this deviation resulted frommixing discriminatory strategies such as maximizing EIG withconfirmatory strategies such as the positive test strategy (PTS).Thirty-nine 5- to -7-year-olds solved 6 puzzles where they hadone opportunity to intervene on a three-node causal system toidentify the correct structure from two possibilities. Children’sintervention choices were better fit by a Bayesian model thatincorporated EIG and PTS compared to alternative models thatonly considered a single strategy or selected interventions atrandom. Our findings suggest that children’s intervention strat-egy may be a combination of discrimination and confirmation.