Importantly, ANNs trained on the same task showed folded-X patterns for both manipulations — suggesting the human double-increase pattern reflects metacognitive mechanisms beyond the statistical structure of the task. 4/4
Across 2 experiments (Gabor patches + dot motion), we found a clean dissociation: boundary distance → folded-X pattern (confidence up for correct, down for errors); stimulus reliability → double-increase pattern (confidence up for both correct and errors). 3/4
Two ways to manipulate task difficulty: (1) stimulus reliability (changing contrast/coherence to make the stimulus itself unclear); (2) boundary distance change tilt offset (keeping the stimulus clear but harder to categorize). 2/4
Our new paper is out in Cognition! What determines whether confidence follows the classic "folded-X" pattern vs. the "double-increase" pattern? The answer lies in the type of stimulus manipulation. Big thanks to my advisor Doby @dobyrahnev.bsky.social and co-first author @herrickfung.bsky.social !
Importantly, ANNs trained on the same task showed folded-X patterns for both manipulations — suggesting the human double-increase pattern reflects metacognitive mechanisms beyond the statistical structure of the task. 4/4
Across 2 experiments (Gabor patches + dot motion), we found a clean dissociation: boundary distance → folded-X pattern (confidence up for correct, down for errors); stimulus reliability → double-increase pattern (confidence up for BOTH correct and errors). 3/4
Two ways to manipulate task difficulty: (1) stimulus reliability (changing contrast/coherence to make the stimulus itself unclear); (2) boundary distance change tilt offset (keeping the stimulus clear but harder to categorize). 2/4
Huge thanks to Medha Shekhar, @kaixue98.bsky.social, @manurausch.bsky.social, & @dobyrahnev.bsky.social for their guidance and collaboration on this project.
Full article: www.sciencedirect.com/science/arti...
Data and code: github.com/herrickfung/...
n/n
My article "A comprehensive assessment of current methods for measuring metacognition" is finally out in Nature Communications 🎉 If you work on metacognition and think you know the psychometric properties of your favorite measure, you may be surprised.
www.nature.com/articles/s41...
Great to see our research, led by @johannakuci.bsky.social, covered in The Transmitter. One of the more surprising findings to have come out of the lab.
"We found that, when informed about task difficulty, humans barely shift their confidence criteria." Awesome new work by Xue, Shekhar and @dobyrahnev.bsky.social questioning the Bayesian confidence hypothesis in perception: www.pnas.org/doi/abs/10.1...
