Studying automaticity and its development in music and language – MULANAU

Learning new complex activities takes a long time before reaching the full achievement of an expert. With intensive practice in a task under attentional control, the mode of processing changes in qualitative ways. In both everyday and scientific terminology, processing is said to become automatic. One of the main characteristics commonly attributed to automatic processing is the irrepressible, ballistic nature of the responses (which means that the initiation of a response can hardly be refrained, and, when engaged, is difficult to suspend). The planned study mainly investigates the development and the evolution through practice of the irrepressible nature of the responses. To investigate this point, Schadler and Thissen (1981) studied the evolution of the well-known Stroop effect as a function of reading skill. They showed that the mean response times for the incongruent stimuli (the word red printed in blue ink) increased as reading proficiency increased from nonreaders to about the fourth-grade equivalent and then decreased, with the pre-sixth-graders' mean response time being not faster than that of the poor readers. The theoretical challenge of the present project is to account for this inverse U-shaped relation, and notably the surprising decrease of interference with prolonged practice. Our general hypothesis is that this pattern is consistent with Logan's theory of automaticity, which posits that the emergence of automatic behavior results from the transition from algorithm-based performance to memory retrieval. Briefly, our proposal is that interference is maximal after a moderate amount of practice because memory retrieval is triggered by very coarse, non specific retrieval cues. Automatized behavior appears whenever a target is displayed, irrespective of whether more fine-grained elements of the situation make it adaptive. With further practice of the task, memory retrieval would be exclusively triggered when increasingly specific elements of context are present. In this perspective, what could be construed as an apparently paradoxical re-emergence of attentional control over behavior is all simply accounted for the fact that behavior becomes increasingly dependent on finer and finer contextual cues. Most of the planned experiments involve musical note naming instead of text reading, which has served as a nearly exclusive paradigm to investigate automatisms in the past. Word reading and note naming are two domains that have obvious communalities. Both of them initially involve supervised forms of learning, and in the two cases, the most elementary steps of processing need to be fully automatized to allow the learner to reach the ultimate stages of mature performance in the respective tasks, namely: comprehension for reading, instrument playing or singing for music. However, reading is a very complex, multifaceted, and tightly constrained activity, which considerably restrict the range of possible experimental manipulations. For instance, the level of practice in reading is largely confounded with learners' age, which makes it difficult to dissociate the genuine effects of practice from all the learners' characteristics linked to age differences (such as the understanding of instructions). By contrast, it is easy to find individuals of a given age, children or adults, but differing with regard to their earlier training on music reading. Likewise, people are exposed at least to some extent to written material all along their life, while it is again easy to find individuals having given up any musical practice for a while after having reached a substantial level of musical expertise. These possibilities will be exploited in the next years to test and refine the account of the irrepressibility of automatic behavior outlined above, and more generally, to explore the role of context in the formation and the expression of automatisms. In addition to standard behavioral methods, we plan to use EEG to disentangle several interpretations about the curvilinear evolution of the Stroop interference. Behavioral measures lead to similar patterns for both poor readers and skilled readers. However, even if behavioral results are similar, it is likely that they are linked to different neural process in beginning readers and skilled readers. Using EEG may help to learn more about these processes by showing that different neurophysiological markers and/or brain oscillation patterns occur. Overall, our research program is promising to shed new light on the formation of automatisms, with the prospect of widespread theoretical and practical implications.