What the effect actually looks like in the task
In these experiments, participants first learned through instrumental conditioning that one stimulus went with one action and another stimulus went with a different action, building a sensorimotor prediction—a learned link between a sight and a movement [1]. They then did a breaking continuous flash suppression task, in which a picture is hidden from awareness by a constantly changing mask and the person presses a key as soon as they become aware of it; the measure is how long the picture stays suppressed before it is reported [1]. The comparison of interest was between trials where the prepared action matched the trained stimulus–action link (congruent) and trials where it did not (incongruent).
The result is a timing difference: congruent pairings produced shorter breakthrough times than incongruent ones, meaning the stimulus was reported as consciously seen sooner when it fit the learned action prediction [1]. The first experiment showed only a numerical trend in that direction, and a Bayesian analysis there supported the null hypothesis of no influence; the second experiment, which tried to make the actions more predictively salient by not repeating the same action within blocks, found a significant speed advantage for congruent pairings, and a meta-analytic Bayes factor across both experiments confirmed the effect [1].
Why a prediction about your own action could change what you report seeing
The motivating idea is that conscious experience may be shaped not only by predictions about the outside world but also by predictions about how the world changes when we act—sensorimotor contingencies [1]. If the brain anticipates the sensory consequences of a prepared movement, then a stimulus that matches that anticipation may be processed more readily and reach report sooner. The study frames its result as support for this action-oriented predictive processing account, but the design measures the speed of reported awareness, not the underlying mechanism, so the faster breakthrough is consistent with such a mechanism rather than proof of it [1].
Other kinds of learned relevance also speed up breakthrough
Sensorimotor links are not the only learned association that can shift access to awareness. In a separate breaking continuous flash suppression study, gratings that had been associated with high monetary reward during a short Pavlovian learning session were reported as conscious sooner than gratings linked to low reward, and the effect replicated in a second experiment with a detection task orthogonal to the reward learning [2]. That suggests value-based learning can also accelerate access, putatively by boosting early visual representation [2].
A third study looked at whether suppressed natural scenes are processed at a categorical level, using suppressed indoor or outdoor scene primes followed by visible targets. Responses were faster and fast responses more accurate when prime and target shared a superordinate category, but the priming during suppression was small—up to about 10 ms—and depended on some residual prime visibility and on the timing between prime and target; without suppression the same priming was more robust, around 24 ms [3]. So category-level processing can occur under suppression, but the effect is modest and tied to how visible the suppressed prime remains [3].
How much weight the sensorimotor result can bear
The sensorimotor finding rests on two experiments with a split outcome: the first did not show a reliable effect, and the second did, with the combined Bayesian analysis favoring an effect [1]. That pattern means the influence of learned stimulus–action links on breakthrough time is probably real but not large or robust enough to appear in every version of the task. The reward and scene-priming studies show that other learned and categorical factors also modulate breakthrough timing, with the scene-priming effects in particular being small and dependent on residual visibility [2][3]. Taken together, the evidence supports the claim that learned associations—including action-linked ones—can bias when a stimulus is reported as consciously seen, while the size and reliability of any single association's contribution remain modest.
Sources used in this answer
Sensorimotor predictions shape reported conscious visual experience in a breaking continuous flash suppression task
Learned stimulus–action associations shortened breakthrough times for congruent versus incongruent pairings in a breaking continuous flash suppression task, significantly in Experiment 2 and confirmed by a meta-analytic Bayes factor across both experiments, though Experiment 1 supported the null.
Learned value modulates the access to visual awareness during continuous flash suppression
Gratings associated with high monetary reward during Pavlovian learning had shorter suppression durations than low-reward gratings in breaking continuous flash suppression, and the effect replicated with an orthogonal detection task.
Priming of natural scene categorization during continuous flash suppression
Suppressed indoor/outdoor scene primes produced small (up to about 10 ms) category-congruent priming effects that depended on residual prime visibility and prime–target timing, compared with more robust priming (about 24 ms) without suppression.