In our experiments, participants performed task-switching tests. On each trial, participants saw a colored rectangle that indicated which of two tasks should be performed on the numbers or letter inside it. A couple of seconds before these were presented, one of three subliminal cues appeared: One cue predicted a switch from the prior task, another cue predicted a repetition of the prior task, and a third cue was nonpredictive (and could therefore appear before either type of trial). The cues were not linked to any task set; rather, they predicted the switch/repeat status of the trial, and therefore the participants could use them to make proactive goal-directed changes in the currently active cognitive set.They found that
Despite participants’ being entirely unaware of subliminal cues in a series of challenging switch tasks, cues predicting switch trials were reliably associated with improved performance. This robust effect was observed across four variants of stimuli and tasks in four independent participant groups.From their summary abstract:
... This utilization of subliminal information was flexible and adapted to a change in cues predicting task switches and occurred only when switch trials were difficult and effortful. When cues were consciously visible, participants were unable to discern their relevance and could not use them to enhance switch performance. Our results show that unconscious cognition can implicitly use subliminal information in a goal-directed manner for anticipatory control, and they also suggest that subliminal representations may be more conducive to certain forms of associative learning.