The best paper I’ve ever reviewed is now out at nat comm:
https://www.nature.com/articles/s41467-024-55686-w
Let me tell why I loved it so much:
https://www.nature.com/articles/s41467-024-55686-w
Let me tell why I loved it so much:
Comments
I think you saw our STM paper that dropped 2 weeks ago that shows DBS affecting evidence accumulation in the DDM.
Sure feels like these two are related under the hood.
TS theories historically distinguish between ‘goal setting’ and ‘task prep’.
I bet sorting taps more of the former (tho both), whereas cued TS taps more into the latter (especially at long prep time)
Wondering if this has implications for task-switch professions like air traffic controllers?
Why is switching between tasks difficult, even when people have time to prepare (‘residual switch costs’)?
In particular, how does this play out in the dynamics of attentional adjustments — what is the process!?
The authors’ approach was to use ‘reverse correlation’, a powerful method from attention research.
By randomly roving both stimulus features over time, you can infer how people tend to weight each feature at each time bin.
By modelling the *weighted* evidence accumulation process, they show that early differences in sensitivity can reproduce the full the pattern of behavior. (Right)
(1) other model parameterizations don’t produce this pattern
(2) the model makes novel predictions for time constraints, which they confirm in a study
(3) interesting differences across two experiments using different stimuli
- switching requires rapid ramping-up of target processing
- this depends on targets >> distractors (re: task set intertia)
- this process depends on the pair of features you have to switch between (with striking symmetry)
- how does this play out with shorter preparation times?
- why does only target processing change, if the pair of features are critical? How does this depend on conflict between features?
- how is this reconfiguration realized in neural circuits?