On the influence of informational content and key-response effect mapping on implicit learning and error monitoring in the serial reaction time (SRT) task

摘要

The present experiment was designed to enhance our understanding of how response effects with varying amounts of useful information influence implicit sequence learning. We recorded event-related brain potentials, while participants performed a modified version of the serial reaction time task (SRTT). In this task, participants have to press one of four keys corresponding to four letters on a computer screen. Unknown to participants, in some parts of the experimental blocks, the stimuli appear in a repetitive (structured) deterministic sequence, whereas in other parts, stimuli were determined randomly. Four groups of participants differing in the presentation of tones after each response performed the SRTT. In the no tone group, no tones were presented after a response. The other three groups differed with respect to the melody generated by the key presses: in the unmelodic group, one out of four different tones was chosen randomly and presented immediately after a response. In the consistent melody group, the press of a response key always resulted in the production of the same tone, resulting in a repetitive melody during structured parts of the sequence (consistent redundant effect). In the inconsistent melody group, the "melody" produced in the sequenced parts of the blocks was identical to the consistent melody group, but the same response could produce two different tones depending on the actual position in the stimulus sequence. Thus, during structured sequences, subjects heard the same melody as in the consistent melody group, but every key press could be followed by one out of two different tones. To disentangle effects of sequence awareness from our experimental manipulations, all analyses were restricted to implicit learners. All four groups showed sequence learning, but to a different degree: in general, every kind of tone improved sequence learning relative to the no tone group. However, unmelodic tones were less beneficial for learning than tones forming a melody. Tones mapped consistently to response keys improved learning faster than tones producing the same melody, but not mapped consistently to keys. However, at the end of the learning phase, the two melody groups did not differ in the amount of sequence learning. The error-related negativity (ERN) increased with sequence learning (larger ERN at the end of the experiment for trials following the sequence compared to random trials) and this effect was more pronounced for the groups that showed more learning. These findings indicate that response effects containing useful information foster sequence learning even if the same response can produce different effects. Furthermore, we replicated earlier results showing that the importance of an error with respect to the task at hand modulates the activity of the human performance monitoring system.