@earlkmiller.bsky.social @danieljamesyon.bsky.social @jdsitt.bsky.social @neddo.bsky.social @luciecharlesneuro.bsky.social @meganakpeters.bsky.social @hakwan.bsky.social @matthiasmichel.bsky.social @dobyrahnev.bsky.social @smfleming.bsky.social @timostein.bsky.social @axc.bsky.social

Comments welcomed:
@standehaene.bsky.social @annaeiserbeck.bsky.social @mesec-community.bsky.social @fahrenfort.bsky.social @johanneskleiner.bsky.social @predictivebrain.bsky.social @adriendoerig.bsky.social @nadinedijkstra.bsky.social @davidchalmers.bsky.social @svangaal.bsky.social

We argue that this might reflect recurrent interactions between the BU and TD networks, which lead to the behavioral interaction effects.

To summarize,

1 + 1 only equals 2 during unconscious processing.

If such behavioral interaction effects are indeed a good measure of recurrent interactions in the brain, these results imply the absence of feature binding during uncons. proc.: red color and apple shape of an apple are processed in respective networks without recurrent interactions inbetween.

Across four conditions of dual-variation tasks, when the prime is invisible, form and color incongruency yield additive effects on RT: (e.g., form incong effect = 24 ms, color incong effect = 22 ms, all incongruent effect = 44 ms). When prime visible, also underadditive (58 ms + 58 ms > 87 ms).

Surprisingly, one day after my paper was accepted. I came across this study: doi.org/10.1037/a001....

They embedded two features into prime and target under meta-contrast masking: Form and Color.

Therefore, prime and target can also be in 2*2= 4 congruency (cong) conditions.

We also argue that behavioral interaction of this kind could be a measure of recurrent interactions between different networks in the brain. If two representations emerge from two distinct networks that do not communicate, they should influence follow-up processing in an additive fashion.

This observation aligns with some postulates of Global Neuronal Workspace theory (GNW): the long-lasting simultaneous and reciprocal interaction between TD decision network and BU input network is preserved during conscious processing (T1 green) and disabled during unconscious processing (T3 red).

In contrast, when prime is partially visible, BU and TD congruency yield under-additive effect. (if only BU-incongruent effect is 30 ms and only TD-incongruent effect is 40 ms, both incongruent effect will be lower than their addition: 30 + 40 > 45 ms).

1 + 1 > 2 during conscious processing!

Across two Exp we found that when the prime is invisible, BU and TD congruency yield additive effect. (if only BU-incongruent effect is 20 ms and only TD-incongruent effect is 30 ms, both incongruent effect will be their addition: 20 + 30 =50 ms).

1 + 1 =2 during unconscious processing!

Here, I term the objective prime identity as a bottom-up (BU) representation of prime, whereas the subjective prime identity as a top-down (TD) representation of prime. The former is a BU input. The latter a TD guess. Both influence RT to target. This distinction yields 2*2=4 congruency conditions.

A few years ago, Sand and Nilsson rendered the prime invisible. They surprisingly found that participant's guess of the prime identity also influences RT. That means if they guess the prime is GREEN, but target is red, RT is also slower. doi.org/10.1177/0956...

In the classic Stroop priming paradigm, if a color word (prime) is incongruent with the target color, response time (RT) to target is delayed compared to when they are congruent.