Review of 

Music As Cognition by Mary Louise Serafine

Reviewer: Ethan A. Bayer

In a report on her studies of the cognitive processes of music, Mary Louise Serafine discards the idea that music gains its beauty purely from an abstract notion of beauty. Instead, she notes that the fundamental trait of music is its organization, unique to each piece. Here, she describes the experiments she headed whose ultimate goal is to define the internal mechanisms that are implemented in the perception and understanding of music.

One of Serafine's fundamental arguments is that, contrary to current models, music is not mentally comprehended in terms of notes, chords, and scales. This is a reduction for the purpose of ease of understanding, and many music forms today cannot be encompassed in these terms (for example, computer-generated and percussion music) (50-1). She says, rather, that this notation of music is not cognitively inherent, but was created by humans for the purpose of music preservation and communication. Scales and keys are not the "units by which [music] is heard," (56) but rather, she proposes a set of cognitive abilities that together allow one to make sense of such a flux of sounds.

Serafine's purpose in her experiments is to chart these abilities across ages to determine if and when such functions develop. Another key aspect of the research is to look at the differences in performance on these perception and cognition tasks between those with formal music training and those without. This data can hopefully provide evidence as to whether the development of music cognition skills are based more heavily on a biological timeline or on external experience.

Approximately 15 of each age group-5, 6, 8, 10, 11 year olds, and adults-were tested on sixteen tasks which involved temporal processes (like succession and simultaneity) and nontemporal processes (abstraction, transformation, and hierarchic levels). An additional 34 children ages 4 to 11 that were currently being trained on violin under the Suzuki method (an intense formal training that stresses musical memory and rote drills) also participated. Pretests for such knowledge as conservation for number and quantity were given, as well as a pitch discrimination task. (It is only at age 8 that half of the subjects become successful at this task. Later scores on these tests and the experiments were examined for correlations, of which few were found.) Each experiment was designed so as to be accessible to all ages present; 5, 6, and half of the 8 year-olds received the younger version (more detailed instructions and explanation of the task, often elaborated by the use of dolls and a story). The experimenter sat in a room with a subject, read prewritten directions, then (in most cases) played the music excerpts or fragments on a stereo. In many tasks, they were allowed to listen to the excerpts as many times as they wanted, but overall this did not have significant (if any) influence on performance. Subjects' responses were recorded on a sheet of paper, and they were not given feedback.

The first set of experiments dealt with the temporal aspect of music-the time-related elements, or those that highlight how one note fits in with another, such as rhythm and succession. One task tested comprehension of motivic chaining. Subjects heard two fragments of music-or motives-and then heard them put together, one after the other. They then listened to more patterns and had to decide if what they heard really was a combination of the first 2 fragments or not. Most of the 5 and 6 year-olds did not pass, and success increased with age, with adults achieving 100% success. 

Another significant experiment in this set examined the correlation between familiarity with the Western idiom (a sort of template for music in the Western world, which happens to be completely environmental) and the refinement of one's cognition of music. Subjects were asked to discriminate between three sets of nine tones each: one set that formed an "intact" melody (as we know it in this culture), one that formed a "partly random melody" (the same nine tones, with some displaced), and one that formed a wholly "random" melody (all of the same nine tones displaced). Again, younger children performed at the level of chance, and performance did not reach significant levels of success until older childhood. Children under ten displayed a better perception for intact melodies than for completely random ones. The results suggest that older subjects had acquired an understanding of the construction of the Western idiom. 

The other set of tasks charted nontemporal functioning-literally, those that did not test timing-related aspects, but rather they required subjects to consider a fragment or composition as a whole on a deeper level than in temporal functioning. The first of these experiments tested comprehension of closure-that a piece is "finished." Here, "finished" pieces ended on a tonic, while "unfinished" pieces ended on a supertonic or another non-tonic chord. As expected, adults and older children showed very accurate perception of closure, whereas younger children performed at the level of chance. The interesting results here come from the Suzuki-trained children. While untrained children only had consistent success by age 10, Suzuki children were displaying a clear understanding of closure by age 8. However, this gap was only apparent after the children had received a significant amount of training under the Suzuki method; 5 and 6 year-olds performed equivalently, regardless of formal training. Serafine suggests that it is unlikely that children at this age could grasp closure, even if they had been through years of formal training.

Another nontemporal experiment dealt with motivic abstraction. Subjects heard a "model" fragment, then listened to two others, one an abstraction (same motive put into a different melody) and one a foil (completely unrelated to the model) and had to decide which of these sounded the most like the model. The underlying aim of this task was to elucidate the ability to recognize a basic rhythmic and musicality structure in spite of the tones and rhythm details it is wrapped up in. 5 through 8 year-olds performed at chance, and 11 year-olds and adults had about an 84% success rate. Younger children could only recognize a motive when it was isolated and not when it was superimposed on a different surface rhythm and tone sequence.

One more key experiment dealt with hierarchic levels. Music can be broken down into different levels of complexity; this task worked with two of those levels, including the most detailed (the fragment itself-the "model") and one where the basic rhythmic and tone structure had been extracted-in simple terms, a "dumbed down" version (called the "reduction"). Subjects listened to the model, then to two other fragments, one being the reduction, the other a foil. 83% of adults, 70% of older children, and 50% of younger children chose correctly the fragment that went best with the model. Interestingly enough, subjects (even adults) had difficulty explaining how they made their decision, and Suzuki-children showed the least difference in performance in relation to their untrained peers on this task. An understanding of hierarchic levels appears to be a very abstract concept, and one that is best acquired through years of experience with everyday exposure to music. 

One of the overarching, though now somewhat obvious, findings of this study is that musical cognitive processes, both temporal and nontemporal, do in fact exist. They develop throughout childhood, but not all at the same speed. The results can be broken down into developmental stages.

During early childhood (here, ages 5), performance on these tasks was poor across the board. The cognitive capacities have not blossomed yet, but there are signs that they are budding, notable in the succession and simultaneity tasks. In middle childhood (age 8), primitive forms of an awareness of hierarchic levels is apparent, and children show adult-like success in the understanding of timbre, idiomatic construction, and a few other musical features. They have trouble in tasks involving closure and the processes underlying the formation of patterns. The greatest developmental leap occurs between ages 8 and 10. By this time, 65-75% of subjects succeed on the tasks (adults showed 85-100% success rates). Children understand how to build phrases out of multiple motives and its parts, and also how to break them down into patterns, units, and hierarchic levels. They grasp the concepts of closure and transformation, even over longer compositions. However, they still show difficulty in identifying number of parts or voices in complex textures. One interesting and unexpected findings that these results yield is that nontemporal processes seem to develop noticeably before temporal processes. It can be proposed from this that children first make sense of the global features of music and then start to register the rhythmic nature of compositions as their memories improve.

Serafine also came to some conclusions on the role of formal training in music cognition. The few aspects of musical understanding that correlated with formal training were comprehension of closure, motivic chaining, and hierarchic levels. This connotes a better grasp of abstract concepts, as well as of the conventional tonal idiom. In early childhood, Suzuki-trained children exhibited larger memory capacities, which assisted them on some tasks, but aside from that they performed at approximately the same level as their untrained peers. As age increased, there was a wider divide between Suzuki-trained and untrained children, but even that was not a wildly dramatic difference in success. Serafine suggests that formal musical training must be intensive and extensive in time in order to have a significant impact on one's music intelligence. The increase of success on these tasks as a function of age shows that everyday exposure to music and possibly cognitive maturation in general assists in comprehension of music. She sums up her findings in this area by saying that formal training is "neither necessary nor sufficient" (229) for music cognition. 

By examining the musical conceptual abilities at different ages, Serafine is able to chart out a rough sketch of the development of the structures that make such conception possible. In highlighting the rich processing that must occur to understand music, she emphasizes her view that music is not merely an abstract art form that one can enjoy like a cup of tea, but it is a composite of cognitive and cultural entities that must be appreciated as much for its aesthetic mystery as for its structural and temporal organization.

Serafine's results have direct applicability to considerations in music, especially music education. The discrete areas of cognition that she tested help us to know what skills children are capable of mastering and when, so that training may be focused on building up strengths that are prepared to blossom rather than those that require more cognitive maturation before they can be developed. Hopefully this information can discount the cultural stigma that in order to fully grasp music one must undergo intense training. The experiments demonstrate that understanding of music is not some innate talent, but rather a compilation of processes that anyone can come by through either practice or simple exposure. If this data is affirmed by future research, the system of music education will call for significant restructuring. 

She also gives evidence to the notion that music is more than just notes. Currently, music is taught in theory classes as a series of logical formulas whose constituents can be classified according to seven letters, A through G. Serafine's findings suggest that a more abstract and holistic approach to learning the structure of music with less of an emphasis on its mathematical properties would benefit students just as much, and would perhaps be preferable. The acknowledgement of music as a universal cognitive entity may also encourage non-musicians that the pathway to musical understanding is not reserved for the prodigies, but rather it is something that everyone is capable of appreciating.

The applications of the knowledge about music cognition are not limited to one domain. Since Serafine poses it as a function of cognitive processing, it carries along with it the properties of all of our basic cognition. For example, this is yet another test of working memory; in her experiments that required subjects to listen to whole compositions, younger children showed failure primarily because they could not attend to such a large amount of information for an extended period of time. We can take this observation and extend it to teaching methods in a normal classroom setting. Just as the children in this experiment showed the ability to focus for only short spurts of time, they probably display the same concentration patterns in areas like math or reading. Having to read straight through a whole book (even a children's book) must be taxing for a young brain; perhaps a better approach would be to ask for the child's undivided attention for just a page or two at a time, and insert non-strenuous activity in between. Something as simple as giving the teacher a high five would allow the working memory a break and a chance to refocus at the beginning of the next segment.

Of course, one major area of interest here is expertise. The fact that Suzuki-trained children, some of whom could play twenty or thirty pieces by memory, showed little edge on music cognition tasks accentuates the strictly limited transfer between domains. We normally do not take into account that "music" encompasses many domains, such as perception, memory, and performance. 

This knowledge about the severity of domain specificity can transfer to our knowledge of other tasks. For instance, we should not expect that someone who writes beautiful poetry will write equally powerful persuasive speeches, or that someone whose theoretical apprehension of math will map onto proficiency in applied math or engineering. College admissions offices would benefit from this knowledge when weighting standardized tests, like that SAT, that assess only a few skills in the domains that they claim to analyze thoroughly. Perhaps domain specificity is one reason why these sorts of tests do not consistently predict academic success; hopefully colleges will be aware of this and look for other measures that can more accurately display someone's ability and potential.

Serafine's work is a significant step in deepening our understanding of the mechanisms that assist in the processing of something that can oftentimes appear to just be an orderless form of art. She shows that music is cognitive in nature, and our perception and comprehension of it require more than the feeling of enjoyment that we receive.