Simon Stephan

(1/7) New paper in @cognitionjournal.bsky.social: "Manipulating Prior Causal Beliefs Affects the Formation of Apparent Causal Illusions" Why do people sometimes "see" causal links in zero-contingency data? We tested if this is rational updating, not bias. 🧵 www.sciencedirect.com/science/arti...

(6/7) Exp. 3 replicated with a different test query; Exp. 4 showed the same pattern using purely statistical (non-visual) causal structure priors. The source of the prior (visual vs. statistical information) didn't matter — only its size did.

(1/7) New paper in @cognitionjournal.bsky.social: "Manipulating Prior Causal Beliefs Affects the Formation of Apparent Causal Illusions" Why do people sometimes "see" causal links in zero-contingency data? We tested if this is rational updating, not bias. 🧵 www.sciencedirect.com/science/arti...

(4/7) Exp. 1: When the mechanism visibly ruled out a link, the "illusion" nearly vanished — ratings dropped to near zero. When hidden, the classic illusion appeared, scaling with outcome density exactly as Bayesian models predict.

(1/7) New paper in @cognitionjournal.bsky.social: "Manipulating Prior Causal Beliefs Affects the Formation of Apparent Causal Illusions" Why do people sometimes "see" causal links in zero-contingency data? We tested if this is rational updating, not bias. 🧵 www.sciencedirect.com/science/arti...

(2/7) This seeming "causal illusion" is often framed as irrational bias. Bayesian models say otherwise: it emerges when non-zero causal priors aren't fully overridden by the data. We manipulated participants' causal priors, which affected the formation of "causal illusions"(N=1,860, 4 experiments).