Evidence for unconscious priming using a Bayesian single-subject approach

Although many studies have claimed the existence of unconscious priming, a valid statistical procedure for substantiating such claims is often lacking. For instance, the absence of prime awareness is often erroneously inferred from non-significant p-values, and priming and awareness are typically tested by comparing the outcomes of two separate t-tests. In a classic study by Vorberg (2003), observers were slower to identify the direction of an arrow mask when it followed an arrow prime pointing in the opposite direction (priming effect), despite being unable to identify the direction of the arrow prime (prime awareness). Using that observation as a starting point, we address several methodological shortcomings that we identified in that study (as in many other studies in the field). To overcome these shortcomings, we implemented three critical changes. First, we employed a Bayesian methodological-statistical framework to test priming effects and prime awareness at the single-subject level over multiple experimental sessions. Second, participants underwent extensive training in the prime awareness task to ensure they understood and executed the task as intended. Third, we directly compared prime awareness and the priming effect. By using this approach, we show unconscious processing at short prime-mask stimulus onset asynchronies (SOAs) in three out of six participants, replicating Vorberg et al. (2003). However, we also show that – unlike in the original study – unconscious priming was not consistently present across all SOAs, and awareness increased with increasing prime-mask SOA. Notably, priming effects were highly consistent across participants, whereas awareness showed substantial variability across participants.