Article of the Day:
The Musical Stroop Effect – Opening a New Avenue to Research on Automatisms
by Grégoire, L., Perruchet, P., & Poulin-Charronnat, B. (2013)
(Experimental Psychology, 60(4). 269–278. DOI: 10.1027/1618-3169/a000197)
Background & Research
The Stroop effect, that naming colours is slower when the coloured stimulus consists of an incongruent colour word (e.g., the word “green”, printed in red), is one of the most famous findings in research on cognitive processes. This phenomenon has been very useful in studying automatisms, in particular the fact that automated processes related to well-practised tasks can create interference in other tasks where they are as such irrelevant.
Many variants of the standard version exist, including for instance picture-word and position-word variants. There are however limitations to the use of any variant employing word reading, mainly because it is quite difficult to manipulate variables related to reading in research settings. Reading practice cannot be restrained for practical and ethical concerns, and, further, reading ability is acquired at an age when other very significant cognitive changes are occurring. Any observations regarding changes in performance (and interference effects) over time may be due to many other factors.
This article reports on two experiments to study the possibility of using the automaticity of note naming in musicians in place of reading as a source of a Stroop-like effect. Such a “musical Stroop effect” would share most of the advantages of the classic word-color version in investigating automatisms in that note naming is a “naturally” occurring practice that is intensively practiced over many years by musicians, and is also a simple component of more complex activities and likely to become automated with practice. In contrast to word reading, the practice potential research subjects have had in note naming can be more easily manipulated (we may select those with 1 year of practice, or 5 or 25, for instance). The amount of musical training is also not coupled with age but can occur at any age, and breaks from practice of any length can occur and be studied, which is usually not the case with reading.
In the first experiment, musically trained (N = 14) and non-trained (N = 14) students were asked to press a button whenever the printed word they were presented with was the name of a note (DO, RE, MI, FA, SOL, LA or SI), and refrain from pressing if it was not (other common two- and three-letter French words were used). They were then presented, for a maximum of 1200 ms each, with words, some of which were names of notes and some of which were not, placed within a picture of a note. Some of the pictures were in congruent positions on a musical stave (a MI on the lowest line, for instance), while others were in incongruent positions (a DO on the lowest line, for instance) and some were presented without an accompanying stave at all. The last case, a baseline condition where note names and other words were printed inside a picture of a note without any contextual stave was included to explore whether any Stroop-like effect would be due to interference in the incongruent condition, facilitation in the congruent one, or a mix of both.
In the second experiment, 34 new participants, again half musically trained and half not, were presented with stimuli in the same manner as in Experiment 1. This time they had to read aloud the printed word, while ignoring its position. In this experiment, the trials involving words other than note names could be used to indicate whether the presence of the stave would by itself slow down responses (suggesting an effect of perceptual complexity).
Results & Discussion
In the first experiment, for musicians only, response times were significantly longer in incongruent conditions as compared to congruent conditions (M = 9.91 ms, SD = 11.56, t(13)=3.21, p = .007). An ANOVA showed no main effect of congruity, a significant effect of musical expertise (F(1, 26) = 33.11, p < .001, η² = .560) and a significant congruity X musical expertise interaction (F(1, 26) = 6.64, p = .016, η²p = .203).
Including the “baseline” condition in analysis showed that both musicians and non-musicians reacted faster in the out-of-context condition than in either the incongruent or congruent conditions, initially suggesting that even the congruent condition created some interference. However, as musical expertise played no role in this effect, the likely explanation lies in the perceptual complexity of each condition (that is, the existence of the musical stave as such increases the complexity of the stimulus and slows down responses). As the speed advantage of musicians over non-musicians was similar in both out-of-context and congruent conditions (about 125-127 ms), but differed in the incongruent condition (about 112.8 ms), the authors conclude that the musical Stroop effect observed here is mainly due to interference in the incongruent condition.
Clear evidence was thus obtained for a musical Stroop effect in a go/no-go procedure in the first epxeriment. Congruity had a reliable effect on reaction times, limited to musicians. In addition, an unexpected effect of musicians recognizing note names faster in all conditions was observed, which the authors attribute to the exhaustive set of note names being more accessible for musicians than for non-musicians. Musicians may be able to retrieve the entire set of names in parallel, whereas others need to perform a serial memory search to determine whether the observed word belongs to the category of note names.
In the second experiment, reaction times were again significantly longer in the incongruent than the congruent condition, with a mean difference of 8.69 ms (t(16) = 2.18, p = .045), for musicians only. An ANOVA revealed a marginally significant congruity effect (F(1, 32) = 3.29, p = .079, η²p = .093.) and a significant congruity X musical expertise interaction (F(1, 32) = 4.20, p = .049, η²p = .116). Again, clear evidence for a musical Stroop effect in musicians was attained.
Similar to the go/no-go task, reading note names was much faster in the out-of-context condition compared with the congruent and incongruent conditions for all participants, with no difference between musicians and non-musicians in this effect. Importantly, for non-note words, an ANOVA revelead a significant and very large effect for context vs. no context (F(1, 32) = 180.50, p < .001, η²p = .849), but no musical expertise effect or context X musical expertise interaction. These results support the idea of perceptual complexity affecting reaction times significantly.
In the second experiment, the speed advantage of musicians was again similar in non-context and congruent conditions as compared to non-musicians, and somewhat smaller in the incongruent condition, though this difference did not reach statistical significance. Overall, the speed advantage of the musicians was not statistically significant in this task, suggesting that the effect in Experiment 1 was primarily due to the nature of the task, and likely to be based on differences in accessibility to the set of note names affecting the categorical membership decision.
In sum, these experiments provide the first proper evidence for a musical Stroop effect, a Stroop-like effect due to “interference generated by the automaticity of note processing in musicians.”
Now, it is interesting and amusing that Stroop-like effects can appear with regards to musical notation in musicians. But could this effect be usefully employed in future investigations on automatisms? The authors answer in the affirmative. The benefits of the musical Stroop paradigm, as mentioned above, lie in that it provides much greater flexibility in manipulating the amount of practice and that practice level can be decoupled from age and reading skill abilities. This offers researchers possibilities to flexibly manipulate a host of new variables in further studies. Other Stroop variants also exist, e.g., using numbers instead of words, or employing second-language skills, but according to the authors, all those variants have their own problems that the musical Stroop effect avoids. The authors identify three prerequisites for the usefulness of the effect: a large enough study population needs to exist (this is estimated to be 2 % of the general population in this case), the potential study population must not differ in other significant ways from the potential control group pool, and the effect needs to be robust and reproducible. In the authors’ view, the musical Stroop effect does seem to fulfill these conditions.
As a final comment, the authors consider the reverse Stroop effect (whether reading could be impaired by an incongruent context), an effect that has been difficult and sometimes impossible to tease out, apparently occurring only in some very specific contexts. Difficulty in demonstrating such an effect has lead some to suggest that word reading would be so automatic as to suffer little or no congruity related interference. The results of the experiments here suggest, however, that even a process as automatic as word reading is not completely protected from interference, when the competing process is also highly automatic. A similar effect has been observed in those possessing perfect pitch (an incongruent note played to them interferes with reading the name of a note). The possibilities of pitting two automatisms against each other could be an additional, interesting new avenue of research.
The original research on The Stroop effect hails from the 1930s, and lots of studies and publications have been published on it and employing it for the study of automatisms. This article is, however, to my knowledge the first to demonstrate a musical Stroop effect, which makes it quite ground-breaking and fascinating in its own, admittedly quite limited, niche. The Stroop-like interference effects demonstrated here were temporally small (8-10 ms) but effect sizes were moderate to large, and the experiments appear methodologically sound and straightforward enough. Though irrelevant to the research and its merits, the clarity of the article suffers somewhat from the less-than-perfect English of its authors.
The practical or long-term value of the results of these experiments are difficult to assess at this point. For now, they just provide us with an interesting aside to the issue of interference by automatisms in seemingly unrelated tasks. But it is possible, as the authors suggest, that the musical Stroop effect could become a particularly useful tool for studying the development of automatisms related to task practice. Potentially, it could be used to study, e.g., individual differences in the processes of automatization and factors affecting them, the effects of extended breaks from practice on automatisms, as well as the relative strengths of different automated processes.