We used recent tools to analyze the prevalence of the effect (Bayesian Prevalence elifesciences.org/articles/62461 developed by @robince.bsky.social) within our sample, and observed that some participants implictly associate a positive valence with agency, while others associate a negative one.

Our study suggest that we explicitly say yes, but implicitly we seem not to attribute a specific valence, or at least, it depends on who is experiencing agency.

The idea behind this measure is that voluntary action requires an effort that involuntary action does not.

Thanks to Elisabeth Pacherie, @emiliecaspar.bsky.social , @axc.bsky.social , Uri Maoz for their help to this work. 4/4

Moreover, we introduce another measure of volition that aligns with the measures mentioned above: the effort exerted in reporting the decisions. More specifically, participants selected their answers using handgrips that measured the effort exerted. 3/N

Our study indicates that self-reported volition, the Readiness Potential, and the Temporal Binding increase when making decisions that could increase our remuneration compared to decisions that did not. This was not observed when contrasting decisions that involved more or less deliberation. 2/N

No confidence, only opinion « I found » « I see » I just wanted to help with my understanding of it not as an expert but as someone interested in these questions :) I think that if you want to help people understanding your claims it’s valuable to have input of what are the peoples understanding

I will be (maybe wronngly) more confident if it’s just M differences rather than S differences 2/2

If there is 5 studies, 1 finding a significant effect, 4 that doesn’t, I have the impression that it give extra information to know the magnitude of the effects and their sign in order to evaluate if it is worth trying to pursue this line of research. 1/2

Ok, I thought that type S and M were from a bayesian framework, which do not adress the uncertainty about long run sampling, but rather the uncertainty of observed data in combination with prior beliefs.

For the S error, I just see that as a different philosophy, not thinking about the long run sampling. I found sometimes type I & II error confusing as we never conduct the same study in psychology.

I found the M error pretty convincing. For instance a study claiming 20% effect of a treatment but other studies replicating 5% effects might be considered as a failed replication even though all significant, because the first did a M error.

📌

But in this commentary, we argue that does Centaur does *not* provide a better account of any psychological findings compared to previous models. Centaur is model without a theory. osf.io/preprints/ps...