Research has suggested that people can learn to link contextual cues, stimuli, or blocks of trials with different levels of control demand. For example, in a cued task-switching paradigm, people can learn to associate task switch probability with predictive cues so that canonical switch costs (cf. Monsell, 2003), namely, the increase in reaction time and error rate caused by task switching, are reduced following high vs. low switch probability cues (Dreisbach, Haider, & Kluwe, 2002). In the current study, we aim to replicate and extend the findings of Farooqui and Manly (2015), who raised the possibility that subliminal cues predictive of high vs. low switch probability could facilitate cued task switching more than supraliminal cues. Recent contextual learning studies also pointed out that the importance of consciousness in control learning might be overstated (Hommel, 2013). Specifically, people often implicitly acquire knowledge about the association between contextual manipulations and control demands. However, in Farooqui and Manly (2015), when the cues were made supraliminal (E9 & E10), the participants were explicitly told to ignore the cues. Therefore, the lack of cuing benefits in this condition may have stemmed less from the supraliminal nature of the cues than from the participants having been actively discouraged from using the cues. Considering this potential limitation of Farooqui and Manly (2015), we extended the design such that our experiments formed a 2 x 2 factorial design: we manipulated whether cues of control demand are consciously perceived or are presented subliminally (cue visibility: subliminal vs. supraliminal), and whether participants have explicit prior knowledge of the cue meaning or acquire that knowledge implicitly through experience (cue knowledge: explicit vs. non-explicit). By disentangling the effects of cue visibility and cue knowledge, this project aims to enhance our understanding of the conditions that promote or impede control learning. Our main experiment and the cue perception task was coded and administered in PsychoPy, while the post-test and demographics form was programmed in Qualtrics. These materials can be found under the Experimental Code folder. The datasets from all four experiments (AllSubs_E#.csv) are in the Data folder. The column headers of the csv files (experimental data) are defined in the R script and the html report files, which are posted in the Analysis Code folder. Datasets for the main and supplementary analysis are stored in AllExpts_MainSubs_RegData.csv AllExpts_AllSubs_RegData.csv, respectively. Post task questions were stored in AllExpts_MainSubs_posttest.csv and AllExpts_AllSubs_posttest.csv, respectively. Main Subs refers to the analysis reported in the paper, while All Subs refers to the supplemental analyses. We completed all analyses in R. We have both main and supplemental sets of analyses, because we added exclusion criteria that were not in the preregistration. Open the html files in your browser to view the full-length reports of both the main and the supplementary analysis. The stage 1 manuscript and final manuscript can be found under their corresponding folders. If you have any questions, please email the corresponding author, Christina Bejjani.