San Anton, E. 1, 2, 3, 5 , Schmitz, R. 1, 2, 4, 5 , Cleeremans, A. 1, 2, 3, 5 & Destrebecqz, A. 1, 2, 3
1 Université Libre de Bruxelles (ULB)
2 Center for Research in Cognition and Neurosciences (CRCN)
3 Consciousness, Cognition & Computation Group (CO3)
4 Neuropsychology and Functional Imaging Research Group (UR2NF)
5 Fonds National de la Recherche Scientifique (FRS- FNRS)
The serial reaction time task (SRT) is viewed as a paradigmatic example of implicit or unconscious learning. Several reports, however, have shown that performance in SRT may also rest on conscious knowledge acquisition. Applying the process dissociation procedure, Destrebecqz and Cleeremans (2001) proposed a generation task with both inclusion (generate the sequence) and exclusion (avoid to generate the sequence) instructions to evaluate the extent to which learning takes place consciously. They also showed that the Response-to-Stimulus Interval (RSI) influenced the conscious nature of learning: a null RSI favored unconscious learning while a longer RSI induced explicit learning strategies and conscious knowledge acquisition. However, other authors have failed to replicate this result (e.g. Wilkinson and Shanks, 2004). Participants' difficulty to apply the exclusion instructions might explain the discrepancy between these studies. In a first experiment, the use of reversible second-order conditional sequences allowed participants to generate the sequence in a reverse order in the exclusion condition. The second experiment was an exact replication of the original study. Overall SRT and generation tasks revealed the presence of implicit sequence learning but failed to show any RSI modulation. To conclude, we speculate on the reasons why we failed to replicate our original findings.