Categorization of Musical Patterns by Self-Organizing Neuronlike Networks

Gjerdingen, Robert O.

doi:10.2307/40285472

Cited by 75 publications

(33 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Some models simulate more complex aspects of music learning and perception, such as categorization and memory of feature patterns. Gjerdingen (1990) exposed a four-level network based on Grossberg's (1987) adaptive resonance theory to early works of Mozart. The input layer codes music theoretic concepts (i.e., harmonic tritone, contrapuntal dissonance) and low-level music features (i.e., melodic contour, pitch of the major diatonic scale plus a unit for the alterations of flat and sharp).…”

Section: Models Of Distributed Knowledge Representationmentioning

confidence: 99%

Implicit learning of tonality: A self-organizing approach.

Tillmann¹,

Bharucha²,

Bigand³

2000

Psychological Review

355

332

View full text Add to dashboard Cite

Tonal music is a highly structured system that is ubiquitous in our cultural environment. We demonstrate the acquisition of implicit knowledge of tonal structure through neural self-organization resulting from mere exposure to simultaneous and sequential combinations of tones. In the process of learning, a network with fundamental neural constraints comes to internalize the essential correlational structure of tonal music. After learning, the network was run through a range of experiments from the literature. The model provides a parsimonious account of a variety of empirical findings dealing with the processing of tone, chord, and key relationships, including relatedness judgments, memory judgments, and expectancies. It also illustrates the plausibility of activation being a unifying mechanism underlying a range of cognitive tasks. Natural environments contain highly structured systems to which we are exposed in everyday life. The human brain internalizes these regularities by passive exposure, and the acquired implicit knowledge influences perception and performance. Aspects of language and music provide two examples of highly structured systems that may be learned in an incidental manner. In each case, there is a paradox. On the one hand, a thorough formal description of the structure has proven to be extremely challenging. On the other hand, native speakers or nonmusician listeners internalize the regularities underlying linguistic or musical structures with apparent ease. A substantial corpus of research has been devoted to the learning process of language, but little has been devoted to the learning of music. The central purpose of the present article is to investigate how implicit knowledge of some basic features of Western musical grammar may be acquired and mentally represented.Barbara Tillmann, Department of Psychology, Universit6 de Bourgogne, LEAD-CNRS, Dijon, France, and Department of Psychological and Brain Sciences, Dartmouth College; Jamshed J. Bharucha, Department of Psychological and Brain Sciences, Dartmouth College; Emmanuel Bigand, Department of Psychology, Universit6 de Bourgogne, LEAD-CNRS.This research was supported in part by National Science Foundation Grant SBR-9601287 and National Institutes of Health Grant 2P50 NS17778-18 to Jamshed J. Bharucha and by a grant from the International Foundation for Music Research to Emmanuel Bigand. Barbara Tillmann has been supported by the French Ministry of Education and Research and by the Deutsche Akademische Austauschdienst DAAD. We thank Herv6 Abdi, Pierre Perruchet, and Philippe Schyns for insightful comments at different stages of the present work and Carol Krumhansl and Fred Lerdahl for comments on the article.Correspondence concerning this article should be addressed to Barbara Tillmann, Department of Psychological and Brain Sciences, Dartmouth College, 6207 Moore Hall, Hanover, New Hampshire 03755. Electronic mall may be sent to barbara.tillmann@dartmouth.edu. 885We present a connectionist model that (a) simulates the implicit learning of p...

show abstract

Section: Models Of Distributed Knowledge Representationmentioning

confidence: 99%

Implicit learning of tonality: A self-organizing approach.

Tillmann¹,

Bharucha²,

Bigand³

2000

Psychological Review

355

332

View full text Add to dashboard Cite

show abstract

“…Expectancies concerning the upcoming "what" of a melody have been extensively investigated in previous psychological studies and incorporated into computer simulations of musical processing (Bharucha, 1987a(Bharucha, , 1987b(Bharucha, , 1989Gjerdingen, 1989Gjerdingen, , 1990. Within the context of laboratory experimentation, expectancies have often been investigated with production tasks, in which subjects are presented with an initial melodic context and asked to predict a future interval through vocal (Carlsen, 1981;Unyk & Carlsen, 1987) or keyboard responses (Abe & Hoshino, 1990;Schmuckler, 1990).…”

Section: Experiments 1 Expectancy Ratingsmentioning

confidence: 99%

Harmonic and rhythmic influences on musical expectancy

Schmuckler

Boltz

1994

Perception & Psychophysics

View full text Add to dashboard Cite

The effects of harmony and rhythm on expectancy formation were studied in two experiments. In both studies, we generated musical passages consisting of a melodic line accompanied by four harmonic (chord) events. These sequences varied in their harmonic content, the rhythmic periodicity of the three context chords prior to the fmal chord, and the ending time of the final chord itself. In Experiment 1, listeners provided ratings for how well the final chord in a chord sequence fit their expectations for what was to come next; analyses revealed subtle changes in ratings as a function of both harmonic and rhythmic variation. Experiment 2 extended these results; listeners made a speeded reaction time judgment on whether the final chord of a sequence belonged with its set of context chords. Analysis of the reaction time data suggested that harmonic and rhythmic variation also influenced the speed of musical processing. These results are interpreted with reference to current models of music cognition, and they highlight the need for rhythmical weighting factors within the psychological representation of tonal/pitch information. 313One basic concern ofcognitive psychologists has been understanding the structure of cognitive representations. In audition, the two primary areas in which such investigations have proceeded involve examining the nature of listeners' representations oflinguistic and musical structure. Music, in particular, provides an excellent domain for such inquiries; musical behavior requires a diversity of perceptual, cognitive, and motoric skills, making it a paradigmatic case of general psychological functioning. In audition, the availability of detailed theoretical analyses of musical structure provide an independent source of hypotheses and predictions concerning the perceptual and cognitive processing of music; such analyses have proven invaluable in our understanding of musical cognition.As a possible consequence of these advantages, a number of informal and formal theoretical frameworks for understanding musical perception have been proposed by psychologists and musicologists

show abstract

“…Two examples are the combination of octaveequivalent pitch categories into melodies and chords, and the combination of these chords. These patterns are far removed from the harmonic structure of individual tones, and most attempts to explain the recognition of and memory for structure at these levels inescapably resort to the prior existence of representations for octave-equivalent pitch categories (e.g., Bharucha, 1987;Deutsch & Feroe, 1981;Dowling, 1978;Gjerdingen, 1990;Krumhansl, 1979Krumhansl, , 1990Lerdahl & JackendoIT, 1983).…”

Section: Learning Octave Equivalence By Neural Self-organizationmentioning

confidence: 99%

“…Even though such models were developed soon afterward (Grossberg, 1970(Grossberg, , 1972(Grossberg, , 1976von der Malsberg, 1973), they are only beginning to be widely understood, as witnessed by recent variants (Kohonen, 1984; Rumelhart & McClelland, 1986). With some exceptions (e.g., Laden & Keefe, 1989), recent work on neural net learning in audition has been focused on larger musical structure (Bharucha, 1991;Gjerdingen, 1990). …”

mentioning

confidence: 98%

Two Issues in Auditory Cognition: Self-Organization of Octave Categories and Pitch-Invariant Pattern Recognition

Bharucha

Mencl

1996

Psychol Sci

View full text Add to dashboard Cite

The study of auditory and music cognition prorides opportunities to explore general cognitil'e mechanisms in a specific, highly structured domain. We discuss tll'O problems lI'ith implications for other domains of perception: the selforganization ofperceptual categories and inrariant pattern recognition. The perceptual category lI'e consider is the oc/are. We sholl' holl' general principles of self-organization operating on a cochlear spectral represelltation can yield octare categories. The example of illl'ariant pattei'll recognition lI'e consider is the recognition of inmriant frequency pattel'lls trallSformed to different absolute frequencies. We suggest a system that uses pitch or musical key to map tones into a pitch-inmriallt format;The goal of this essay is to introduce to a general psychological audience two (of many) issues in auditory and music cognition that'involve constraints and solutions that are common to other domains of perceptual cognition. How do tones an octave apa~t come to be perceived as similar in pitch? How can the characteristic spectrum of a sound source be recognized even though it may be heard at different absolute pitch levels? These have traditionally been construed narrowly as auditory problems. Research on the computational capabilities of neural nets enables us to examine them afresh and suggests directions that may generalize to other areas of cognition.We suggest that the first question can be addressed in terms of the perceptual learning of categories through neural selforganization. Octave equivalence is widely believed to be universal, but it is erroneously thought to therefore be innate. A general-purpose perceptual learning mechanism coupled with the acoustic regularities of the environment would not only enable octave equivalence to be learned but would compel such learning.The first person to suggest how octave equivalence might be achieved by a neural net was Deutsch (1969). This question could not have been framed as a learning problem then because precise models of how such a network might organize itself through perceptual learning were not yet available. Even though such models were developed soon afterward (Grossberg, 1970(Grossberg, , 1972(Grossberg, , 1976von der Malsberg, 1973), they are only beginning to be widely understood, as witnessed by recent variants (Kohonen, 1984; Rumelhart & McClelland, 1986). With some exceptions (e.g., Laden & Keefe, 1989), recent work on neural net learning in audition has been focused on larger musical structure (Bharucha, 1991;Gjerdingen, 1990).

show abstract

Categorization of Musical Patterns by Self-Organizing Neuronlike Networks

Cited by 75 publications

References 16 publications

Implicit learning of tonality: A self-organizing approach.

Implicit learning of tonality: A self-organizing approach.

Harmonic and rhythmic influences on musical expectancy

Two Issues in Auditory Cognition: Self-Organization of Octave Categories and Pitch-Invariant Pattern Recognition

Contact Info

Product

Resources

About