Arts & Sciences

Newsletter
Spring 1996 Vol. 17 No. 2

Mozart and Mind: A Perspective from Music Cognition


Carol L. Krumhansl


As a complex human behavior, music is a natural focus for the disciplines that make up the cognitive sciences. The experimental methods of cognitive psychology, theoretical proposals in linguistics, aesthetic questions in philosophy, and computational techniques in computer science have all recently yielded new insights. Music theory has increasingly addressed the nature of human capacities for perception, memory, and skilled performance. Recent technology has provided powerful tools for music analysis, synthesis, composition, and performance.

As a cognitive psychologist, my primary interest has been in such questions as: How do we perceive and remember music? What are the structural properties of music that affect these processes? To what degrees do experience, musical training, enculturation, and development influence how we understand music?

The research methods my colleagues and I employ are representative of those in experimental psychology. Listeners hear music in controlled situations and respond according to the instructions given for the experimental task. These will depend on the particular research question. Some questions can be answered with simple tonal and rhythmic patterns written to vary only certain local properties of interest. Others require more musically complex materials, such as excerpts from compositions. Listeners in the experiments range from young infants to professional musicians.

In keeping with a theme of this issue of the newsletter, I will describe briefly two experimental studies using pieces by Mozart. The first study was done in collaboration with a developmental psycholinguist, Peter Jusczyk at SUNY Buffalo. He and his colleagues had shown that young infants are sensitive to segmentation in language, and we wondered whether they would also show this sensitivity in music.

Our method, called the preferential looking paradigm, presented two kinds of musical excerpts from Mozart minuets. In the appropriately segmented excerpts, we added a short pause at the end of each musical phrase. In the inappropriately segmented excerpts, we added an equivalent number of pauses in the middle of phrases. Otherwise the music, played by a synthesizer, was metronomic. The infant sat on its parent+s lap in the center of a three-sided test booth. Each trial began by blinking a light on the center panel. An observer (who was hidden behind the center panel) detected when the infant looked toward center, extinguished the center light, and illuminated a flashing light on one of the side panels. When the infant had turned its head in the direction of the side panel, the music started and continued until the infant looked away for two seconds.

During a familiarization period, the infant learned that it could stop the music by looking away from the side panel. Then, during the experiment itself, the two kinds of excerpts were played and the observer recorded the number of seconds the infant looked in the direction of the loudspeaker before turning away. (Incidentally, the observer and the parent heard other music over headphones so they could not hear the excerpt.) We found that 4 1/2 and 6 month old infants oriented longer when the pauses appeared at the ends of phrases than when they appeared mid-phrase. This pattern, which was remarkably consistent, indicates the infants could differentiate between appropriately and inappropriately segmented music.

Figure 1 Click on thumbnail for full image

To what cues in the music might the infants be responding? Analysis of the excerpts showed two differences that correlated with the orientation time data. First, a drop in pitch tended to precede the pauses in the appropriately segmented excerpts, but not in the inappropriately segmented excerpts. Second, the last notes before the pauses tended to have longer durations in the appropriately segmented excerpts than in the others. Interestingly, these are some of the cues that infants appear to use in segmenting language. This convergence suggests that infants use these acoustic contrasts in early development to parse both music and language.

The second study grew out of a collaboration between a group of psychologists and music theorists who convened as a special project group at the Center for Advanced Study in the Behavioral Sciences (Stanford, California) in 1993-94. We planned a collection of papers on Mozart+s Piano Sonata in Eb, K. 282. Music theory provides a valuable resource to psychologists for suggesting testable proposals, selecting musical materials, and interpreting experimental results. In exchange, the experimental results can serve to refine theoretical proposals, offer complementary techniques of analysis, and describe the psychological processes engaged by music. Our focus on a single piece of music resulted in a significant, and somewhat surprising, degree of convergence between theoretical and experimental approaches.

Figure 2 Click on thumbnail for full image

We chose this particular Mozart sonata, in part, because a virtuoso pianist had performed it on a computer-monitored Bosendorfer acoustic grand piano. The piano records precisely the timing and velocity of each note event. Caroline Palmer (psychology, Ohio State University; Ph.D. '88, Cornell) statistically analyzed this information for correspondences with theoretical proposals. She found that the expressive timing variations in the performance correlated with Fred Lerdahl's (music, Columbia University) predictions about harmonic tension in his analysis of the piece. She also found that his predictions for surface dissonance corresponded with increased use of the damper pedal. Moreover, the performer used dynamics to emphasize notes that, in Eugene Narmour's (music, University of Pennsylvania) theory, deviate from melodic expectations.

These theoretical predictions also correlated with perceptual judgments in experiments that Jamileh Jemison (class of 1997) and I conducted here at Cornell. We used the computer-coded information from the original performance to resynthesize the performance in the laboratory. This permitted us to register the perceptual responses precisely with respect to the sounded musical events. One task asked listeners to indicate on a continuously varying control how tension changed over time as they listened to the piece. The resulting tension profiles tended to increase throughout musical phrases and decrease rapidly at the end of major segments. Local waves of tension were superimposed on these larger variations. These responses corresponded with the theoretical analyses of the piece, and with the dynamics and tempo of the performance. Together, these results show a promising degree of convergence between theoretical descriptions of music and experimental measures of performance and perception.



Since 1980, Carol Krumhansl (psychology) has conducted research on the psychology of music at Cornell. The topics investigated include tonal and harmonic organization in Western music, the experience of rhythm, time, and musical form, cross-cultural studies of melody perception, and memory for twentieth- century music. Her research appears in Cognitive Foundations of Musical Pitch (Oxford University Press, 1990) and numerous journal articles. The interdisciplinary project on Mozart's Piano Sonata in Eb, K. 282, will appear in the spring issue of Music Perception (University of California Press).


Spring96 Newsletter / A&S Newsletters / A&S Home / Welcome to CU / CUInfo

This article is Copyright © 1996 Carol Krumhansl. All Rights Reserved.