Kobe Desender

Finally, and most importantly, Robin wrote an excellent and accessible demo which should allow anyone (you!) to get started with hMFC: github.com/robinvloeber...

MANY figures in the paper showing that hMFC works, but highlighting this one: with as few as 500 trials per participant hMFC allows excellent recovery of single-trial criterion, look at panel C for a representative example participant - I'm (obviously biased) impressed by this!

We developed hMFC, a Bayesian hierarchical framework which allows estimating single-trial criterion states, by fitting data from different participants while taking into account of the nesting of data within participants.

Ignoring fluctuations in criterion is problematic: simulations show that criterion fluctuations induce apparent history biases (panel C), lead to underestimated psychometric slopes (panel D) and underestimated measures of sensitivity, such as d' (panel D)

Classic models of decision-making, like signal detection theory, assume that choices are made by comparing a decision variable (DV) to a criterion. Often this criterion is (implicitly) assumed to be constant; here we implement a fluctuating criterion following an autoregressive model.

Introducing hMFC: A Bayesian hierarchical model of trial-to-trial fluctuations in decision criterion! Now out in @plos.org Comp Bio.
led by Robin Vloeberghs with @anne-urai.bsky.social Scott Linderman

Paper: desenderlab.com/wp-content/u... Thread ↓↓↓

#PsychSciSky #Neuroscience #Neuroskyence

Full details, alternative valence-only models, and post-experiment questionnaires targeting awareness, etc. all in the paper!

At the group level, our learning model won over a non-learning alternative, but more participants were actually best fitted by the latter. Closer inspection revealed why: there was a dynamic group (showing a clear confidence learning effect) and a static group (showing, well, nothing)

At the group level, participants adapted their reporting of confidence to subtle changes in feedback (with no effects on accuracy or RTs). Panel E nicely shows how people adapt their confidence to feedback over time, panel D shows that our learning model closely captures this finding!

To experimentally test this, we provided participants with model-generated feedback, reflecting the probability that their choice was correct. Unbeknownst to them, we alternated between between blocks with subtly higher/lower feedback

We know (more or less) how humans compute confidence, but how do we learn to compute confidence? We propose that agents compute prediction errors (confidence-feedback) to update the weights underlying the computation of confidence

"Learning to be confident: How agents learn confidence based on prediction errors"! Now out in @cognitionjournal.bsky.social led by @pierreledenmat.bsky.social

Paper: desenderlab.com/wp-content/u... Thread ↓↓↓

#AcademicSky #PsychSciSky #Neuroscience #Neuroskyence

Common neural choice signals reflect accumulated evidence, not confidence! Now out in @cerebralcortex.bsky.social w @helenevanmarcke.bsky.social @pierreledenmat.bsky.social @yfvisser.bsky.social @denizerdil.bsky.social a.o.

Paper: desenderlab.com/wp-content/u... Thread ↓↓↓

Read the full paper for fancy time-frequency plots (replicating pre-stimulus occipital alpha & confidence), and multivariate decoding (showing cross-decoding between priors and confidence)!

We instead identified a frontal signal, which tracked confidence and was sensitive to prior beliefs. Although speculative, this might be the signal that integrates priors and evidence into a confidence judgment!

Critically, EEG measurements confirmed the key prediction: although the stimulus-locked CPP and response-locked Pe were sensitive to high vs low confidence (which happens because confidence is correlated with evidence), they were _not_ modulated by prior beliefs condition (panels B and C)!

Replicating previous work, our manipulations had clear and consistent effect on confidence: confidence integrates prior beliefs about performance with accumulated evidence.

This work makes a key prediction: evidence accumulation signals (such as CPP and Pe) reflect accumulated evidence which feeds into confidence, but do not directly reflect confidence. To test this, we trained people on easy/hard tasks and provided pos/neg feedback (i.e. to manipulate priors)

In 2024, @helenevanmarcke.bsky.social @pierreledenmat.bsky.social showed that confidence is computed _conditional_ on prior beliefs about task performance (journals.sagepub.com/doi/abs/10.1...), represented by the heat map in the figure.

Common neural choice signals reflect accumulated evidence, not confidence! Now out in @cerebralcortex.bsky.social w @helenevanmarcke.bsky.social @pierreledenmat.bsky.social @yfvisser.bsky.social @denizerdil.bsky.social a.o.

Paper: desenderlab.com/wp-content/u... Thread ↓↓↓

Can we use confidence-driven information-seeking as a tool to combat fake news!? Really cool study by really cool @helenevanmarcke.bsky.social ↓↓↓

Full-Force at #ccn2025 in Amsterdam. Come along for a chat if you're interested in metacognition, confidence, computational modelling, reasoning, etc. @yfvisser.bsky.social @jeremiebeucler.bsky.social @helenevanmarcke.bsky.social @alexandre-lietard.bsky.social @zoepurcell.bsky.social

Or if you want to make your life easier, simplify to SDT and see whether biased criterion alone suffices, or whether you also need biased confidence criteria (mapping onto v and v_s)

Manipulating Prior Beliefs Causally Induces Under- and Overconfidence - PubMed Humans differ vastly in the confidence they assign to decisions. Although such under- and overconfidence relate to fundamental life outcomes, a computational account specifying the underlying mechanis...

The model used here: pubmed.ncbi.nlm.nih.gov/38427319/ where you can test whether the prior affects confidence directly (via z or drift bias) or indirectly (via v_s), or both. These parameters are then updated based on the prediction error cue-stim weighted by a learning rate. Good luck :)

Impressive behavioral + modelling + eeg work, led by @lucvermeylen.bsky.social , relating time-on-task changes in comp parameters to changes in metacognitive reports!

Performance feedback counterintuitively _increases_ overconfidence in young children. Great Collab @dsotob.bsky.social

Performance feedback triggers liberal detection and perceptual confidence biases in early childhood: Implications for metacognitive training - PubMed Metacognition allows us to monitor our own mental processes and the quality of our decisions in order to promote adaptive behavior and learning across different domains. Despite its potential, the rol...

Performance feedback triggers liberal detection andperceptual confidence biases inearly childhood: Implications formetacognitive training @kobedesender.bsky.social

pubmed.ncbi.nlm.nih.gov/40555904/

Very cool new work led by the incredible @helenevanmarcke.bsky.social in @cognitionjournal.bsky.social Go and Read if you're into information seeking, metacognitieve confidence and (!) _beautiful_ figures :)

Impressive study led by @johnpgrogan1.bsky.social and @lucvermeylen.bsky.social combining evidence accumulation modeling and EEG to find a computatioanl account that explains the computations of confidence. So many answers in this manuscript :)

If you're in Dublin, go and talk to @yfvisser.bsky.social she's doing some really cool stuff on insight into our own biases!!