However, these results are contingent on whether the task features can be decoded. Activity-silent, or 'bursts' of proactive processes sprinkled in the ITI and not time-locked to the stimulus, may not appear with our method. Therefore, the proactive unicorn is still galloping even in our task.

I'll plug this in: we run the Multi task with EEG and, adapting Kikumoto's decoding+RSA method, we can see a change in distractors' coding compatible with the distractor-specific adaptation reported in our recent NHB. This 'looks' like reactive, given it shows up 500ms post-stimulus.

This is awesome! Congratulations!!

Overall, this suggests that attentional gains for multiple sources of information can be adapted in parallel.

This opens the questions of how ‘multidimensional’ conflict is represented in the brain.

Check out the discussion of our preprint for some speculation.

With a neural network model, we show that you need multiple, distractor-specific conflict monitors.

In short, humans can simultaneously adapt attention towards multiple sources, based on their individual ‘compatibility’ history.

(Lasagne never helped writing a great epitaph. Life stories did)

We observe within-dimension adaptation, but absence of cross-dimension adaptation.

Put simpler, participants adapted attention to a given dimension based selectively on the congruency history of that dimension with the cued task.

a.k.a. ‘spaghetti plot’. It would be lame if it wasn’t how we actually refer to this effect.

Control adaptation in the MULTI produces parametric congruency sequence effects.

You are robust to congruency effects the more you have been previously exposed to incongruence (flatter, lighter lines).

(Check out @hritz.bsky.social’s new paper for a related, convergent, finding!)

The congruency of non-cued dimensions spans from 0 (all incongruent) to 3 (all congruent). This leads to parametric effects on responses to the cued dimension.
Also, interference comes from everywhere: Any non-cued dimension interfere with any cued task (as in the best families).

In our new MULTI paradigm, two stimuli differ across 4 dimensions. Your need to find the instructed feature for a given cued dimension (aka ‘write the epitaph’).
The three other dimensions can either facilitate or interfere (i.e., your sister and your aunt).

Families can be large.

Could you compose your cousin's epitaph while your sister shares his life stories, and also suppress your aunt’s push for more lasagne?