Long-term contingency learning (still) does not depend on contingency awareness

The present paper presents evidence against the notion that true contingency learning effects are only possible when participants are consciously aware of the pairings, as has been argued in a series of recent papers. We present a novel variant of the colour-word contingency learning paradigm designed to eliminate (or minimize) the influence of stimulus-response binding confounds. Within each of three sets of words, each word was presented most often in one colour. Participants cycled through the three sets of words in such a way that words never repeated within close proximity (between 6 and 10 intervening items). In a series of mixed models controlling for immediate colour repetitions, long-distance word-response repetitions, and item-level objective awareness, contingency learning effects were consistently observed. Most importantly, a robust contingency learning effect was not only observed for "aware items", but also for "unaware items" after controlling for the above-mentioned repetition effects. This represents clear evidence for implicit learning. That is, even when participants do not try to learn and do not develop conscious awareness of the pairings, a robust contingency learning effect is still observed.