Professor Kuniyoshi L. Sakai, Graduate Student Reiya Horisawa (at the time of the research), and Assistant Professor Keita Umejima from the University of Tokyo Graduate School of Arts and Sciences, in joint research with the Talent Education Research Institute, have revealed that when intermediate players begin learning a new piano piece, practicing for about one week while first listening to recordings leads to more accurate understanding of the piece compared with starting by reading sheet music. Sakai explains, "It has previously been said that one should focus on piano rather than trying various instruments, but we found that experience with multiple instruments is actually effective for music acquisition. We obtained similar results to what we've previously established in English-language learning—that listening to recordings and learning multiple languages simultaneously is effective. I believe these findings can be applied to future music and language learning." The research was published in Cerebral Cortex.

(A) An example of training for seven days on four music pieces, grouped into Sets I (pieces I-1 and I-2) and II (pieces II-1 and II-2). During the first five days, the participants listened to a CD of recorded pieces for one Set (Set I in this case), and read sheet music for the other Set. These were designated the "Listen" and "Read" training conditions, respectively. During the last two days, the participants trained by playing the piano while reading the sheet music for both Sets. Participants placed a check mark in the bottom-right corner of each box to self-report her/his fulfillment of training.
(B) An example of a normal stimulus (✓), which was always auditorily presented in the scanner. In an error-detection task under the "Context" condition, participants listened to an excerpt of recorded pieces, and judged whether there was an unnatural portion in the excerpt. The initial section from piece II-2 (Entrée in A minor) is shown.
(C) The unnatural stimulus (✘) formed from the normal one shown in (B). Two phrases surrounded by two boxes were swapped (denoted by a double-headed arrow) to construct an unnatural stimulus. Swapped phrases preserved the major rules of counterpoint, as well as natural flows in harmony, but produced structural changes. Note that the swapped phrases were normal in themselves, but they were regarded as errors in the task based on the context of preceding and following phrases. These auditory stimuli were presented for 18 s, including a few more bars with fade-out.
Horisawa, R., Umejima, K., Azuma, S., Miyamae, T., Hayano, R. & Sakai, K. L.: Brain activation patterns reflecting differences in music training: listening by ear vs. reading sheet music for the recognition of contexts and structures in a composition. Cereb. Cortex 35, bhaf072 (2025). DOI: 10.1093/cercor/bhaf072
Music, like language, is based on humans' unique ability to understand complex hierarchical structures, but how the neural foundations of music are acquired through practice has not been well understood. In particular, there have been no comparative studies on whether practice methods that emphasize listening to recordings first (the Suzuki Method) or those emphasizing reading sheet music are superior, and there is no established theory about which brain regions are responsible for the effects of practice. It also remained unclear how brain functions related to music change with experience in playing instruments. In this study, 38 participants ranging from junior high and high school students to university students and working adults with piano experience were divided into two groups: 19 participants with experience playing instruments other than piano (the Multi group, including 11 Suzuki Method students) and 19 participants with only piano experience (the Mono group, including 12 Suzuki Method students). Additionally, 23 Suzuki Method students were designated as the Suzuki group.
The researchers used recordings of four piano solo pieces that participants did not know. Participants underwent training for one week, with the first five days consisting of listening to CD recordings of two pieces five times each day (Listen condition), while for the other two pieces, they were instructed to read the sheet music for a similar amount of time each day (Read condition). The sets of two pieces were swapped for half of the participants. During the last two days of training, participants confirmed their understanding of all four pieces by playing them on the piano while looking at the sheet music.
The day after training was completed, participants performed a music judgment task in an MRI machine. In each trial, participants were presented with an 18-second excerpt (half performed by different musicians than in the training recordings) and asked to determine whether there were any unnatural portions (where 1-2 measures had been swapped) by pressing a button. While the music flow around the unnatural portions still sounded natural, participants had to judge whether the structure was natural within the context of the piece (Context condition). As a control condition (Direction condition), participants were also asked to judge whether there were changes in sound localization (changes in left-right volume balance) in the same piece.
Results showed that the index calculated by dividing accuracy rate by response time was higher for the Multi group with experience in multiple instruments than for the Mono group, and for both groups, the Listen condition yielded better results than the Read condition, indicating better comprehension of the pieces. This superiority of the Listen condition was notable in the Suzuki group. Since this advantage was not observed in non-Suzuki participants, it was not simply that using sound in the task favored the Listen condition. Additionally, the difference between the Multi and Mono groups could not be explained by the simple quantitative measure of total practice time summed across multiple instruments, suggesting it is a qualitative effect of having at least one year of musical experience with different instruments.
When measuring brain activity using fMRI during this task (Context condition), increased activity was observed in the left language areas and auditory input-related regions only in the Multi group under the Listen condition. However, under the Read condition, increased activity was observed in both frontal lobes and anterior superior temporal gyri in both the Multi and Mono groups. The training effect that was more pronounced in the Read condition than in the Listen condition appeared in parts of the right frontal and temporal lobes in both groups, but in the Suzuki group specifically, the left language area was also included.
Further quantitative analysis of brain activity revealed that these activity levels changed in correlation with performance in the Read condition. Activity in the right anterior superior temporal gyrus showed a positive correlation with response time and a negative correlation with the index calculated by dividing accuracy rate by response time. In other words, the higher the performance (higher accuracy and shorter response time), the more efficient the activity in this region becomes.
These results simultaneously demonstrated the effects of both short-term and long-term musical experience for the first time. Furthermore, these effects were reflected in the activity of regions including the right frontal lobe, and particularly in the Listen condition of the Multi group, the left language area was effectively utilized. This suggests that the interpretation of context and structure in music is supported by brain functions common to language interpretation.
Chairman Seizo Azuma of the Talent Education Research Institute (Professor at Tokyo University of the Arts) commented: "Part of what I've intuitively felt over many years has been revealed in this experiment. Music contains various types of information that we receive unconsciously. I believe that from the beginning stages of learning, listening helps develop these sensibilities."
Journal Information
Publication: Cerebral Cortex
Title: Brain activation patterns reflecting differences in music training: listening by ear vs. reading sheet music for the recognition of contexts and structures in a composition
DOI: 10.1093/cercor/bhaf072
This article has been translated by JST with permission from The Science News Ltd. (https://sci-news.co.jp/). Unauthorized reproduction of the article and photographs is prohibited.