Q

introduction

conceptualizing music

arly in the opening volume of A la recherche du temps perdu, Marcel Proust pre-
E sents the first of a number of memorable accounts of listening to music. He
describes Charles Swann’s initial encounter with the andante of Vinteuil’s sonata for
violin and piano:
Doubtless the notes which we hear at such moments tend, according to their pitch
and volume, to spread out before our eyes over surfaces of varying dimensions, to trace
arabesques, to give us the sensation of breadth or tenuity, stability or capr ice. But the
notes themselves have vanished before these sensations have developed sufficiently to
escape submersion under those which the succeeding or even simultaneous notes have
already begun to awaken in us. And this impression would continue to envelop in its
liquidity, its ceaseless overlapping, the motifs which from time to time emerge, barely
discernible, to plunge again and disappear and drown, recognized only by the partic-
ular kind of pleasure which they instill, impossible to descr ibe, to recollect, to name,
ineffable— did not our memor y, like a laborer who toils at the laying down of firm
foundations beneath the tumult of the waves, by fashioning for us facsimiles of those
fugitive phrases, enable us to compare and to contrast them with those that follow.
And so, scarcely had the exquisite sensation which Swann had exper ienced died away,
before his memory had furnished him with an immediate transcr ipt, sketchy, it is true,
and provisional, which he had been able to glance at while the piece continued, so
that, when the same impression suddenly returned, it was no longer impossible to
grasp.1

What Proust summons in this lyr ical, enchanted vignette is the awakening and ini-
tial consolidation of musical understanding. Swann’s first impressions of Vinteuil’s
sonata are vague and unfor med, his mind simultaneously struggling with and savor-
ing the ineffability of the music. But then, with the aid of memory, patterns emerge.
Although these are incomplete and subject to revision, they offer him a way to
make sense of the music, even as it continues to play.
Conceptualizing Music provides an exploration of the process of musical under-

1. Marcel Proust, Swann’s Way (vol. 1 of Remembrance of Things Past), trans. C. K. Scott Moncr ieff
and Terence Kilmartin (New York: Vintage Books, 1981), 227.

4 i nt roduc t i on

standing — that is, the process through which those liquid impressions spoken of by
Proust are transfor med into structures that make it possible to grasp music. In what
follows, I argue that Swann’s— or anyone’s— understanding of music draws on the
same cognitive processes that humans use to organize their understanding of the
world as a whole. Confronted with musical sound, these processes create musical
concepts, the things that enabled Swann to gain a g rasp of the music. The act of
conceptualizing music is the beg inning of a whole chain of cognitive events that
allow us to theor ize about music and to analyze the things that populate our aural
past, present, and future.
The notion that Swann’s musings might give rise to musical concepts demands
some further consideration. Concepts are often thought of as highly stable cognitive
structures of considerable complexity, a view hardly commensurate with the ephemera
attended to or produced by Swann. Recent work in the brain sciences and the mind
sciences, however, has changed how we view concepts. There are now persuasive
arguments that concepts are quite fluid, that they are not ir revocably wedded to
words or to concrete representations, and that they are not even unique to our
species.2 In consequence, the provisional replicas of musical phrases that make it
possible for Swann to secure a foothold in the unfamiliar ter rain of Vinteuil’s sonata
need not automatically be excluded from the conceptual domain. In fact, they are
very much like the concepts we use to structure our understanding of the every-
day world.
This same work in the brain sciences and the mind sciences suggests a way to
account for the apparent simplicity and immediacy of musical understanding, which
seems incommensurate with the complexity of musical structure. For instance, the
sonata by the fictional Vinteuil is intended to be a relatively complex contemporary
work that has captured the fancy of the musical elite of the Paris salons. Nonethe-
less, Swann, whose connoisseurship does not extend to music, is able to gain a grasp
of the work almost immediately. That he is able to do so is not simply novelistic
license but is, in fact, thoroughly plausible: almost everyone has had the exper ience
of listening to an unusual composition or exotic repertoire and being able to make
something of it. This possibility has suggested to some a latent musicality in humans
comparable to the sort of competence for language proposed by Noam Chomsky.3
Competencies of this sort raise as many questions as they answer, however, partic-
ularly where cultural entities such as music are concer ned. It seems more promising
to follow the path of researchers who have rejected linguistic competence as a given

2. See, for instance, Douglas R. Hofstadter and the Fluid Analogies Research Group, Fluid Concepts
and Creative Analogies: Computer Models of the Fundamental Mechanisms of Thought (New York: Basic Books,
1995), chaps. 5– 6, 8– 10; Gerald M. Edelman, The Remembered Present: A Biological Theory of Consciousness
(New York: Basic Books, 1989), chap. 8; and Donald R. Griffin, Animal Minds (Chicago: University of
Chicago Press, 1992), chap. 6.
3. The notion of musical competence has generated a range of commentar y and scholarship. See
John Blacking, “Music, Culture, and Experience,” in Music, Culture, and Experience: Selected Papers of John
Blacking, ed. Reginald Byron, with a foreword by Bruno Nettl (Chicago: University of Chicago Press,
1995), 228 – 31, on musical competence and culture; Steven Pinker, How the Mind Works (New York:
W.W. Norton, 1997), 528 – 38, on musical competence and its relation to other competencies; and Allan
Keiler, “The Origins of Schenker’s Thought: How Man Is Musical,” Journal of Music Theory 33 (1989):
273 – 98, on a nineteenth-century conception of musical faculties akin to musical competence.
 conc e p tual i z i ng mu s i c
5

and who have set about explor ing the cognitive foundations of language.4 Their
task has been to discover what processes are basic to human cognition and then to
determine how they are specified for language. For my part, I would like to explore
how some of these same general cognitive processes are specified for music. By this
means I hope to account for the apparent ease and real rapidity with which we can
conceptualize a highly complex, completely unfamiliar music on our first encounter,
without having to postulate the faculty of musical competence.
At a bit more of a remove, but no less important for a complete account of cog-
nitive processing, is the way concepts come to be organized into the more extended
cognitive structures with which our thought is usually occupied. This is where the-
ories come into play, for theor ies are the cognitive tools that guide the way we rea-
son about the things we exper ience. At first, this might seem to be a rather special-
ized use of the notion of “theory,” for the theory with which much current literature
is occupied — within and without music scholarship— is hardly the stuff of every-
day exper ience. Recent research has countered this view of theory by demonstrat-
ing that the elegant and abstract theor ies of science have much in common with the
tools for reasoning used by very young children.5 Theories are the basic means by
which we make our exper ience coherent and guide further action. The rough-and-
ready transcr ipt that guides Swann’s listening thus has something in common with
the more fully articulated and systematic structures we usually associate with the
idea of “music theory.”
That music theory might have alliances with everyday thought processes is a
provocative claim. On the one hand, music theory often manifests itself as a relent-
lessly practical discipline: a codification of the scales, chords, and grammatical rules
proper to a highly circumscr ibed portion of musical discourse, assembled with the
intent of rendering music comprehensible to those who would become musically
literate. On the other hand, music theory can reach into the far cor ners of abstrac-
tion to embrace complicated mathematical concepts or the arcane symbologies of
voice-leading graphs, as any reader of the Journal of Music Theory, Music Theory Spec-
trum, or Music Analysis will quickly discover. Nonetheless, I want to argue that music
theory, in all its diverse forms, reflects the same basic processes that guide our under-
standing of the everyday world. Theorizing about music is an activity specialized
only in its domain, not in the cognitive processes it involves.
What might these cognitive processes be, and how would they manifest them-
selves? To answer these questions, let us begin at a beginning, with two theor ies of

4. For general introductions to some of the working assumptions of cognitive linguists, see George
Lakoff, “The Invariance Hypothesis: Is Abstract Reason Based on Image-Schemas?” Cognitive Linguistics
1 (1990): 39– 74; and Michael Tomasello, “Introduction: A Cognitive-Functional Perspective on Lan-
guage Structure,” in The New Psychology of Language: Cognitive and Functional Approaches to Language Struc-
ture, ed. Michael Tomasello (Mahwah, N.J.: Erlbaum Associates, 1998), vii–xxiii. In the field of cognitive
linguistics, Ron Langacker’s work is particularly notable for its thoroughness and its systematic approach.
See Langacker, Theoretical Prerequisites, and Descriptive Application, vols. 1 and 2, respectively, of Foundations
of Cognitive Grammar (Stanford, Calif.: Stanford University Press, 1987, 1992); idem, Grammar and Con-
ceptualization (Berlin: Mouton de Gruyter, 2000).
5. See Alison Gopnik and Andrew N. Meltzoff, Words, Thoughts, and Theories (Cambridge, Mass.:
MIT Press, 1997); and Alison Gopnik, Andrew N. Meltzoff, and Patricia K. Kuhl, The Scientist in the Crib:
Minds, Brains, and How Children Learn (New York:William Morrow, 1999).
6 i nt roduc t i on

music from Greek antiquity. These theor ies, and the music to which they refer, are so
unfamiliar that even many who make music theory the focus of their research have
only passing knowledge of what they involve. But there is an advantage in this unfa-
miliar ity, for the disor ienting effect it can have also serves to loosen our notions about
what a theory of music, or a theoretical construct, should be. Despite connections
with music theory as it was practiced in Europe and (to a lesser extent) in the Arabic
world, Greek music theory of antiquity is the theory of an alien society. Nonetheless,
this theory was a beginning — one of the starting points for accounts of musical orga-
nization in the Western tradition. We can discern within it, therefore, the theor ists of
antiquity grappling with basic constructs equivalent to, if still different from, the sort
of constructs taught in beginning music theory classes today. Both aspects will allow
us to see the role basic cognitive structures play in our understanding of music.

ancient music theory and

modern cognitive science
Those who wrote on Greek musical practice in antiquity concer ned themselves
with a wide variety of topics, including the place of music in society, musical aes-
thetics, the construction and nature of musical instruments, and the organization of
pitch mater ials. Specific discussions of pitch mater ials— the usual topic of disquisi-
tions more directly or iented to music theory— centered around the set of pitch
relationships that has come to be called the Greater Perfect System.6 As shown in
figure I.1, this consisted of a set of four tetrachords (hypaton, meson, diezeugmenon,
and hyperbolaion), which, together with one additional note (called Proslambanom-
enos), provided the framework for a two-octave system of pitches basic to Greek
music. The end points of the tetrachords (the notes Hypate hypaton, Hypate meson,
Mese, and so on), together with Proslambanomenos, were regarded as stable, unmov-
able pitches. There was, however, no fixed standard for tuning — the note-names
given in figure I.1 are simply for the pur poses of illustration.Within the boundar ies
marked by each tetrachord were two other pitches, whose placement varied accord-
ing to which of three different genera was understood to be in play. (In the tetra-
chord hypaton, for instance, the variable notes above Hypate hypaton were Parhypate
hypaton and Lichanos hypaton.) The diatonic genus located the movable pitches in a
manner analogous to moder n diatonic scales (for instance, given the reference
pitches on fig. I.1, Parhypate hypaton in the diatonic genus would be equivalent to
F3, and Lichanos hypaton would be equivalent to G3), but the chromatic and enhar-
monic genera situated the pitches in ways that have no comfortable analogue in
modern scale construction. The result was a system in which the placement of
movable pitches could vary widely and in which intervals between successive
pitches could be smaller than a half step and larger than a whole step.

6. For a more detailed discussion of the Greater Perfect System and its place in Greek theor y of
antiquity, see Andrew Barker, ed. and trans., introduction to Harmonic and Acoustic Theory (vol. 2 of Greek
Musical Writings), Cambridge Readings in the Literature of Music (Cambr idge: Cambridge University
Press, 1989), 12– 13. For a more general overview of Greek music theor y and a thorough discussion of
the sources, see Thomas J. Mathiesen, Apollo’s Lyre: Greek Music and Music Theory in Antiquity and the
Middle Ages (Lincoln: University of Nebraska Press, 2000), part 4.
 conc e p tual i z i ng mu s i c
7

C5 Nete hyperbolaion
Tetrachord hyperbolaion
G4 Nete diezeugmenon
Tetrachord diezeugmenon

D4 Paramese

C4 Mese
Tetrachord meson
G3 Hypate meson
Tetrachord hypaton

D3 Hypate hypaton
C3 Proslambanomenos

figure I.1 Diagram of the Greater Perfect System of Greek music theory

There were two main theoretical approaches to presenting the elements of the
Greater Perfect System. The Pythagorean approach der ives from a metaphysics asso-
ciated with Pythagoras of Samos, who lived during the sixth centur y b.c. Its hall-
marks are a persistent interest in number and the deployment of numer ical con-
ceptions in cosmological contexts. In contrast to this, the Aristoxenian approach,
associated with the Peripatetic school of the fourth century b.c., places little reliance
on number, trusting instead in observation and reason as the means to knowledge
about music.

Pythagoras and the Blacksmiths

Nicomachus of Gerasa, a mathematician and har monist wr iting around the begin-
ning of the second century a.d., described Pythagoras’s discovery of the basic pr in-
ciples of music theory as follows:
He [Pythagoras] was plunged one day in thought and intense reasoning, to see if he
could devise some instrumental aid for the hear ing which would be consistent and
not prone to er ror, in the way that sight is assisted by the compasses, the measur ing rod
and the dioptra, and touch by the balance and by the devising of measures; and hap-
pening by some heaven-sent chance to walk by a blacksmith’s workshop, he heard the
hammers beating iron on the anvil and giving out sounds fully concordant in combi-
nation with one another, with the exception of one pairing; and he recognized among
them the consonance of the octave and those of the fifth and the fourth. He noticed
that what lay between the fourth and the fifth was itself discordant, but was essential
in filling out the greater of these intervals. Overjoyed at the way his project had come,
with god’s help, to fulfillment, he ran into the smithy, and through a great variety of
experiments he discovered that what stood in direct relation to the difference in the
8 i nt roduc t i on

sound was the weight of the hammers, not the force of the str ikers or the shapes of the
hammer-heads or the alteration of the iron which was being beaten. He weighed
them accurately, and took away for his own use pieces of metal exactly equal in weight
to the hammers.7

Nicomachus continues the story by descr ibing how Pythagoras used the weights to
conduct further exper iments. After suspending the weights from identical str ings,
Pythagoras plucked pairs of strings and discovered the same concords as he had
heard produced by the blacksmiths. He further discovered that the interval of an
octave was produced by weights in a 2:1 ratio, that of the fifth by weights in a 3:2
ratio, and that of the fourth by weights in a 4:3 ratio, as shown in figure I.2a. The
one discordant interval— that of a second sounded by the middle two weights —
was the product of a 9:8 ratio. Additional exper imentation showed that the small-
est weight sounded a fourth, with the next to smallest weight (8:6
4:3) and a fifth
with the next to largest (9:6
3:2; see fig. I.2b). The octave could thus be viewed
as the product of either a fourth plus a fifth (12:9:6; fig. I.2c) or of a fifth plus a
fourth (12:8:6; fig. I.2d). According to Nicomachus, Pythagoras also discovered that
these ratios held constant throughout the musical domain. It made no difference
whether the constituent notes of the intervals were produced through string ten-
sion, string division, beating on pots, or blowing on tubes — the relationships
between these notes always reduced to the self-same ratios.8
The Pythagorean view of music outlined by Nicomachus assumes that music has
its origins in the natural world and that the natural world has a basic (if often
unseen) order that can be expressed through number. It is thus important to Nico-
machus’s story that the r inging of the hammers is accidental and not contr ived: their
harmony has everything to do with the inherent order of the world and almost
nothing to do with the blacksmiths. It is also significant that the concordant inter-
vals are immediately apparent to Pythagoras and that he can discer n them even
amid the discordant clang of the major second. Not only is the basis of musical
order natural, but also it is manifest to all who have ears to hear. The association of
these intervals with the pounding hammers provides the computational tool for

7. Nicomachus, Enchiridion, in Barker, Greek Musical Writings, 2: 256– 57. The dioptra was a rod that
was used for the indirect measurement of the height of tall objects.
8. Nicomachus, Enchiridion, 258. It should be noted that Nicomachus’s story is a complete fiction.
There is no evidence whatsoever that Pythagoras ever conducted any empirical research on the acoustic
origins of harmonic relationships, with or without blacksmiths. Perhaps more important, the ratios
described by Nicomachus simply do not work. To sound the intervals descr ibed in the story, the values
of the weights must be squared— that is, the weights must be in the ratio 4:1 to produce the octave, 9:4
to produce the fifth, and 16:9 to produce the fourth. The ratios given by Nicomachus only work when
used to segment a str ing into different sounding lengths. One-half the length of a str ing will sound an
octave with the entire length of the string; two-thirds the length of the str ing will sound a fifth with the
entire length of the str ing; and three-fourths the length of a str ing will sound a fourth with the entire
length of the str ing. The importance of Nicomachus’s story lies in its influence: in the for m Boethius
gave it in the sixth century, it became the standard account of basic Pythagorean principles for the Mid-
dle Ages and Renaissance; see Ancius Manlius Severinus Boethius, Fundamentals of Music, ed. Claude V.
Palisca, trans. Calvin Bower, Music Theory Translation Ser ies (New Haven, Conn.:Yale University Press,
1989), 17– 19. A somewhat different perspective on Nichomachus’s story can be found in Mathiesen,
Apollo’s Lyre, 399.
 conc e p tual i z i ng mu s i c
9

a. 12 9 8 6

4:3
3:2
2:1

b. 12 9 8 6
4:3
3:2
2:1

c. 12 9 6
4:3 3:2

d. 12 8 6
3:2 4:3

figure I.2 Ratios of the octave, fifth, and fourth from Pythagorean legend

which Pythagoras had searched, for it allowed him to translate the constituent notes
of the har monic intervals into magnitudes — that is, into number.9 These numbers
then provide further proof of the order of nature, for the ratios of the concordant
intervals (the octave, fifth, and fourth) are all simple (1:2, 2:3, and 3:4), while the
ratio of the discordant second is relatively complex (8:9). Finally, the numbers
involved in the ratios of the concords—1, 2, 3, and 4— were also those of the tetrak-
tys of the decad, which Pythagoreans regarded as the “fount and root of ever-
flowing nature.”10
The account of musical organization presented by the story of Pythagoras and
the blacksmiths is a model of concision. At its core are but three intervals: the
octave, fourth, and fifth. The notes that make up these intervals are assigned mag-
nitudes, the cor respondence of which yields numerical ratios. These ratios come
to stand for the intervals — the ratio 1:2 is the octave— and can also be used to
describe relationships between them, leading to a precise account of the composi-

9. The notion of a computational tool I employ here derives from the work of Edwin Hutchins,
especially that presented in the second chapter of his Cognition in the Wild (Cambridge, Mass.: MIT Press,
1995). Hutchins proposes that various of the navigational tools used by seafarers are in fact computational
tools in that they facilitate computation by transfor ming analog information into digital infor mation.
This is exactly what the hammers did for Pythagoras: they transfor med the analog information of sound
into the digital infor mation of hammer weights (with an inter mediary stage occupied by the computa-
tional tool of a scale for measur ing the weights).
10. Barker, Greek Musical Writings, 2: 30. A tetraktys is any coordinated group of four items; those of
the tetraktys of the decad sum to 10, which is the basis of the base-10 number ser ies used by the Greeks.
10 i nt roduc t i on

tion of intervals. The interval between the Mese and Proslambanomenos of figure I.1
is an octave, a 1:2 ratio. This octave is made up of a fourth (from Mese to Hypate
meson, a 3:4 ratio) and a fifth (from Hypate meson to Proslambanomenos, a 2:3 ratio).
The fifth is in tur n made up of a fourth (from Hypate meson to Hypate hypaton,
another 3:4 ratio) and a tone (from Hypate hypaton to Proslambanomenos, an 8:9
ratio). With these components in place, a Pythagorean theor ist could descr ibe,
through number, relationships between any of the fixed notes of the Greater Perfect
System and eventually character ize the relationships that obtained among the
movable notes of the various genera. The order of the cosmos, which was for
Pythagoreans the order of number, thus found sounding expression in the domain
of music.

Aristoxenus and Aristotelianism

Although the Pythagorean perspective on musical order was an influential one —
for instance, it infor med Plato’s and Aristotle’s writings on music — it was not the
only one available to antiquity. The alter native offered by Aristoxenus in his Ele-
menta harmonica (most likely written toward the end of the fourth century b.c.) starts
with a definition of the science of harmonics:
It is to be understood as the science which deals with all melody, and inquires how the
voice naturally places intervals as it is tensed and relaxed. For we assert that the voice
has a natural way of moving, and does not place intervals haphazardly.We try to give
these matters demonstrations which confor m to the appearances, not in the manner
of our predecessors, some of whom used arguments quite extraneous to the subject,
dismissing perception as inaccurate and inventing theoretical explanations, and saying
that it is in ratios of numbers and relative speeds that the high and the low come
about. Their accounts are altogether extraneous, and totally in conflict with the ap-
pearances. Others delivered oracular utterances on individual topics, without giving
explanations of demonstrations, and without even properly enumerating the percep-
tual data. We, on the other hand, try to adopt initial pr inciples which are all evident
to anyone exper ienced in music, and to demonstrate what follows from them.11

This account of harmonics reveals Aristoxenus to be in conflict not only with the
Pythagoreans (the unnamed antagonists who dismiss perception and explain pitch
relations through ratios) but also with earlier har monic theor ists whose empir ical
work he found deficient because they did not explain their methods of proof or
properly descr ibe their observations.
Aristoxenus’s alter native was to apply Aristotle’s intellectual method to music
more rigorously than did Aristotle himself .12 This entailed restricting the account
of music to ter ms and concepts that could properly be said to belong to the domain
of music. It excluded descr iptions that made recourse to ratios (which are in the
domain of number) or to theor ies about the propagation of physical sound (which

11. Aristoxenus, Elementa harmonica, in Harmonic and Acoustic Theory (vol. 2 of Greek Musical Writings),
ed. and trans. Andrew Barker, Cambridge Readings in the Literature of Music (Cambr idge: Cambridge
University Press, 1989), 149 – 50.
12. For commentar y on Aristoxenus’s approach, see Barker, Greek Musical Writings, 2: 66– 69, 119.
 conc e p tual i z i ng mu s i c 11

are in the domain of physics). Once the definition of these basic musical concepts
was accomplished, an explanation of the entire domain of music could then follow.
Aristoxenus’s demonstration proceeds in three steps. First, he identifies two
forms of vocal motion, the continuous and the intervallic. In the continuous for m,
which is associated with speech, the voice appears to traverse space without stop-
ping, until the point of silence. In the intervallic for m, which is associated with
singing, the voice appears to stand still at specific points, and then pass over some
interval of space before coming to rest at another point.13 The various pitches upon
which the voice pauses when singing constitute musical notes; the spaces between
these notes are musical intervals.
Aristoxenus’s second step is to draw distinctions among the various musical
intervals. The first distinction is made with regard to magnitude, which reflects the
amount of space between the two notes of the interval. That such a space exists is
inferred from the difference between the two pitches that adjoin the interval; the
size of the space can be reckoned in ter ms of how many other notes could be put
inside it.14 The second distinction is made with regard to concord and discord. Aris-
toxenus identifies the concordant intervals as the fourth, fifth, and octave (and their
octave duplications). These are the only concordant intervals he accepts as de-
termined by the intr insic nature of melody — all other intervals are by definition
discordant.15
The final step toward assembling the basic definitions and pr inciples of har-
monics is the der ivation and division of the tone. Aristoxenus defines the tone as the
difference between the first two concords (the fourth and the fifth) and explains
that it can be divided in half (yielding the semitone), in thirds (yielding the least
chromatic diesis), or in fourths (yielding the least enhar monic diesis, which is the
smallest interval recognized as melodic).16 These distinctions allow him to locate
various notes within the tetrachords of the different genera and thereby to specify
the scalar structure of each.
With these steps, Aristoxenus lays out the basic mater ials for his account of musi-
cal organization. Singing involves a specific way of using the voice that creates
musical notes and musical intervals. Intervals can be distinguished according to size
and whether they are concordant or discordant. Concordant intervals — the fourth,
fifth, and octave— are accepted as axiomatic to melody and thus representative of
the basic mater ials from which the various intervals of the Greater Perfect System
can be developed.

13. Aristoxenus, Elementa harmonica, 133.

14. Aristoxenus, Elementa harmonica, 136.
15. Aristoxenus, Elementa harmonica, 139. Aristoxenus further distinguishes between composite and
incomposite intervals (p. 137); however, this distinction is not necessary for a basic understanding of his
theory.
16. Aristoxenus, Elementa harmonica, 140. The different dieses apply to the three genera mentioned
above: the half-tone is used in the diatonic genus; the third-tone is used in the chromatic genus; and the
fourth-tone is used in the enhar monic genus. For further discussion of Aristoxenus’s Elementa harmon-
ica, see Mathiesen, Apollo’s Lyre, 319– 34.
12 i nt roduc t i on

Greek Theory and Cognitive Structure

As mentioned, the music theor ies of Pythagoras and Aristoxenus belong to a world
remote from our own. Not only did these theor ists have to grapple with the most
basic of principles, but also the music they would descr ibe is a microtonal one that
is primarily concer ned with the successive notes of melody rather than the simul-
taneous notes of harmony. Despite this — or perhaps because of it— Pythagorean
and Aristoxenian accounts of musical organization give us a glimpse into how the-
ories are formed and, more important, the cognitive processes that are basic to these
theories. In particular, three cognitive processes can be seen at work: categorization,
cross-domain mapping, and the use of conceptual models.

categorization Our ability to categor ize things is a cognitive process so

basic and so pervasive that it can easily escape our notice.Were you to lift your eyes
from this book and survey your sur roundings, you might well see chairs, lamps,
tables, and other books; were you outside, you might see trees, birds, clouds, cars, and
bicycles. If you considered the other things that populate your day, you might think
of friends and family members, facial expressions and gestures, actions and activities.
Your recognition of these things reflects the categor ies through which we structure
our thought: to recognize a book is to identify it as a member of the category book;
to recognize a tree is to identify it as a member of the category tree. Categorization
occurs in all sensor y modalities and throughout the range of mental activities: we
categor ize smells and sounds, thoughts and emotions, skin sensations and physical
movement.
Categor ies are not just basic to thought; they also give insight into our thought
processes. At one time it was thought that categor ies reflected the structure of the
real world, but recent research has shown that the categor ies humans use are shaped
by their interactions with their environments. Our reasons for developing and
employing a given category are part and parcel of the category itself: categor ies are
not only not given by nature, but also they are subject to change and modification
as our thought unfolds.
Two categor ies basic to Pythagorean and Aristoxenian music theory are those for
consonant or dissonant intervals. Consonant intervals (such as the octave, fifth, and
fourth) are fundamental to the conceptualization of Greek music: they mark the sta-
ble pitches of the Greater Perfect System and are the source of derivation for all fur-
ther intervals, both consonant and dissonant. The process of categor ization is also
exhaustive: any interval that can be conceived belongs to one of these two cate-
gories.
This is not to say, however, that consonant and dissonant intervals are given by
nature in any simple way, Pythagoras and the blacksmiths notwithstanding.17 As one

17. I should note here that psychoacousticians distinguish between musical consonance, which is a
cultural construct framed relative to a particular set of musical practices, and sensory consonance, which
is a consequence of how sound waves are processed by the hear ing mechanism (which involves the
cochlea and the auditory cortex). Sensory consonance is thus a fairly straightforward product of nature.
Although musical consonance has its basis in sensory consonance, there is some freedom in how the sen-
sory data are inter preted.
 conc e p tual i z i ng mu s i c 13

example, consider the way Aristoxenians and Pythagoreans classified the interval of
an octave plus a fourth. Aristoxenians considered the interval a consonance, since it
was simply the combination of two smaller consonances. Pythagoreans, in contrast,
classified intervals according to the numer ical ratio for med by their constituent
pitches. As explained by the anonymous (and thoroughly Pythagorean) author of
the Sectio canonis (fourth century b.c.), consonant intervals are those whose ratios are
either multiple (of the for m [mn]:n) or epimor ic (of the for m [n + 1]:n). Dissonant
intervals are those whose ratios are epimer ic (of the for m [n + m]:n, where m is
greater than 1 and neither equal to nor a multiple of n).18 Because the octave plus
a fourth had the epimer ic ratio 8:3, it was regarded as a dissonance.
Another example of how categor ies shape our understanding of phenomena is
provided by Greek theor ists’ treatment of thirds and sixths. Although thirds and
sixths sound fairly consonant, they were nonetheless categor ized as discords. Two
factors bear on this classification. First, forming thirds and sixths requires using the
movable pitches of the Greater Perfect System — at best, a third or a sixth will
involve only one of the stable pitches bounding the constituent tetrachords of the
system. Thirds and sixths were intervals that necessar ily varied in size, and so they
were placed among the dissonances. Second, in the classification of intervals Greek
theory followed a tradition of dichotomous categor ies: there was concord, discord,
and nothing else. By contrast, neither of these factors played a part in the music the-
ory of early India. Indian music theor ists were consequently free to focus on the
qualitative aspect of intervals rather than on their cor respondence with the fixed
notes of a tuning system and to construe intervallic relationships as concordant, dis-
cordant, or neutral.19
These two examples show that while the categor ies for consonant and dissonant
intervals may be basic to Pythagorean and Aristoxenian theor y, just how they are
defined reflects the context and goals of categorization: consonance and dissonance
are not naturally occur ring properties, but ways of constructing an understanding of
musical organization.
Of course, there are numerous other categor ies important for Pythagorean and
Aristoxenian music theor y, including those for pitches, intervals, and numer ical
ratios. These categor ies and others are basic to the sort of systematic account of
musical phenomena provided by these theor ies — indeed, it is simply not possible to
have a theory of music, or of anything else, without first having categor ies.

cross-domain mapping Cross-domain mapping is a process through

which we structure our understanding of one domain (which is typically unfamil-
iar or abstract) in ter ms of another (which is most often familiar and concrete). For
example, one way to think about the elusive concepts of electr ical conductance is
in terms of a hydraulic model: flipping the light switch tur ns on the juice, and elec-

18. Sectio canonis, in Harmonic and Acoustic Theory (vol. 2 of Greek Musical Writings), ed. Andrew
Barker, Cambridge Readings in the Literature of Music (Cambr idge: Cambridge University Press, 1989),
193.
19. Lewis Rowell, Music and Musical Thought in Early India, Chicago Studies in Ethnomusicology
(Chicago: University of Chicago Press, 1992), 157 – 60.
14 i nt roduc t i on

trical cur rent flows to the light bulb to light the room. By this means we take what
we know about a fairly concrete and familiar source domain — the flow of water
and other liquids — and map it onto a rather abstract and unfamiliar target domain:
that of electr icity. As a wealth of research on analogy and metaphor has shown, the
process of mapping structure from one domain to another is basic to human under-
standing.
One place cross-domain mapping is evident is in the Pythagorean and Aristox-
enian construal of interval. Because musical pitches are ephemeral and virtually
intangible, relationships between pitches — musical intervals — represent something
of a challenge to understanding. One way to meet this challenge is to map structure
from the physical world onto music, a process evident in Nicomachus’s story of
Pythagoras and the blacksmiths. Pythagoras hears harmonious sounds, traces their
origins to the blacksmiths’ hammers, and then proceeds to conduct various exper-
iments using weights equivalent to those of the hammers. These exper iments lead,
among other things, to a highly pragmatic objectification of musical pitch, as pitches
are translated into physical objects that can be weighed, studied, and preserved. By
performing a mapping from the concrete physical domain proper to the black-
smiths’ hammers onto the domain of musical sound, Nicomachus’s story allows us
to structure the latter domain in ter ms of the for mer. Of course, musical notes are
not physical objects that can be weighed, studied, and preserved— they remain
ephemeral and virtually intangible. Nonetheless, we are so accustomed to the map-
ping between concrete physical objects and musical sound that we sometimes have
to be reminded that notes are not endur ing physical objects.
Aristoxenus’s construal of musical interval involves a slightly different mapping.
As we have seen, according to Aristoxenus, when the voice moves intervallically, it
appears to stand still at a given place (a musical pitch) and then pass over an inter-
val of space (a musical interval) before coming to rest at another place (another
musical pitch). Underlying this account is a mapping from the familiar domain of
two-dimensional space onto that of music. This mapping allows us to apply the
methodology of measur ing space to music. The difference between two linear mea-
sures yields a third measure; similarly, the difference between the intervals of a fifth
and a fourth yields the interval of a tone. Since linear measures can be easily divided
into equal halves or thirds or fourths, the musical tone can be similarly divided,
something impossible from the Pythagorean perspective.
On closer inspection, the Pythagorean and Aristoxenian construals of interval are
indeed incommensurate. From the Pythagorean perspective, pitches are physical
objects, and an interval descr ibes the relationship between these objects. From the
Aristoxenian perspective, pitches are breadthless points that simply mark out an
expanse of two-dimensional space, and an interval is the expanse itself. Each map-
ping gives an account of interval, but each leads to a different conceptualization of
musical structure. This point can be generalized for music theory as a whole: map-
ping structure from a nonmusical domain onto music is a way of creating musical
structure, and different mappings will lead to different accounts of musical structure.

conce ptual mode ls Both categor ization and cross-domain mapping

provide the basis for fundamental ontological assertions about musical mater ials: this
 conc e p tual i z i ng mu s i c 15

interval is a consonance; the pitches of an octave are physical objects. They can also
lead to conditional statements: if the interval is an octave, then it is a consonance; if
a pitch is an object, then its properties are measurable. Propositions like this are basic
to conceptual models, which act as guides to reasoning and inference. In their sim-
plest for m, conceptual models consist of concepts in specified relationships, which
pertain to a specific domain of knowledge.
For an example of a conceptual model, let us return to the classification of con-
sonant and dissonant intervals presented in the Sectio canonis, according to which all
consonant intervals have either multiple or epimor ic ratios, and all dissonant inter-
vals have epimer ic ratios. This classificator y system relies on a conceptual model
that organizes concepts related to interval, concord, discord, and the three classes of
ratios. The simple patter n of inference that follows from this model is that if an
interval has a multiple or an epimor ic ratio, it is a concord; if it has an epimer ic
ratio, it is a discord.
Integral to this model are the products of categor ization and cross-domain map-
ping. Two types of categories are involved in the model: those pertaining to music
(the categor ies of concord and discord) and those pertaining to number (the mul-
tiple, epimoric, and epimer ic ratios). Cross-domain mapping cor relates the two types
of categor ies by construing musical interval as a relationship between two objects
(namely, musical pitches) to which magnitudes (in the for m of numbers) can be
assigned. Specific classes of ratios can then be used to distinguish between the musi-
cal categor ies.
The robustness of this particular conceptual model is reflected in the debate over
the status of the octave plus a fourth that continued into the Middle Ages. In the
second centur y a.d., Ptolemy showed the speciousness of the cor relation of con-
cord with multiple or epimor ic ratios and argued for a classification of intervals
based on empir ical evaluation and the postulate that a concord added to a concord
produces a concord.20 Although Ptolemy still used ratios to descr ibe various inter-
vals, they were no longer part of the conceptual model through which intervals
were classified into concords and discords. In the sixth century, Boethius presented
both the Pythagorean and Ptolemaic models but took no position on which he pre-
ferred.21 After Boethius, when an author wished to invoke the author ity of the
Pythagorean approach, the Pythagorean model of intervallic classification was cited;
when an author wished for a more empirically satisfying classification, the Ptole-
maic model was used.22
Conceptual models provide the first level of organization for concepts. They are
too limited and localized, however, to provide the comprehensiveness we expect
from theor ies of music. Theories achieve this comprehensiveness by integrating

20. Ptolemy, Harmonics, in Harmonic and Acoustic Theory (vol. 2 of Greek Musical Writings), ed. Andrew
Barker, Cambridge Readings in the Literature of Music (Cambr idge: Cambridge University Press, 1989),
286– 90.
21. Boethius, Fundamentals of Music, 81– 82, 169.
22. For a discussion of these modes of reasoning, see C. André Barbera, “The Consonant Eleventh
and the Expansion of the Musical Tetractys: A Study of Ancient Pythagoreanism,” Journal of Music The-
ory 28 (1984): 191– 223. Barbera’s treatment of Aurelian’s Musica disciplina on p. 210 is especially illumi-
nating.
16 i nt roduc t i on

clusters of conceptual models. And as we shall see in the following chapters, con-
ceptual models also play a role in categor ization and cross-domain mapping. There,
as in theor ies, they provide guides for reasoning and inference about specific and
circumscr ibed domains of knowledge.

cognitive processes and music theory Much has been left out
of this discussion of theories of music from Greek antiquity, with respect to both the
theories themselves and the cognitive processes behind them. To be sure, these the-
ories are of a different order than Swann’s musings on Vinteuil’s sonata. Nonethe-
less, the cognitive processes we have seen at work in Pythagorean and Aristoxenian
theory are the same processes through which we organize our understanding of the
world as a whole. Just how this occurs— how categor ization, cross-domain map-
ping, and the use of conceptual models shape our theor ies of music and guide our
analyses of musical works — is the subject of the remainder of this book.

overview
I have divided the chapters that follow into two parts. In the first, I present a
detailed overview of research on the three cognitive processes highlighted in this
introduction. This overview is itself framed around specific musical topics, such as
motivic transfor mation, text painting, and the ways in which we structure our
understanding of a specific musical domain. The research that has been done in
cognitive science over the past three decades has been extensive and far ranging, and
one of the jobs of this portion of the book is to br ing this work to bear on basic
issues of musical understanding. Another objective is to show in some detail how
these processes relate to one another and how they form the bedrock for our
thought about music. The second part of the book moves from this foundation to
analytical studies of specific musical issues. These issues include relationships
between categor ization and musical syntax, the problem of musical ontology, text-
music relations, and conceptions of musical for m and musical hierarchy.
Chapter 1 begins the overview of research in cognitive science with a close look
at processes of categorization. For centur ies, writers in the West regarded categor ies
as fixed and immutable, and any variation in categor ization was taken as evidence of
the failure of the human intellect to deal with the structure of the real world. It took
the pioneer ing work of Eleanor Rosch and others in the 1970s to show that cate-
gory structure was not as simple as first believed. In particular, some levels of cate-
gorization are preferred over others, and some members of a category are regarded
as better representing the categor y than others (a phenomenon known as graded
membership).
The key to how this research can be applied to music is provided by the musi-
cal motive (or, as Proust would have it, the motif ). Motives are generally reckoned to
be one of the basic building blocks of musical works, but they are also a bit slippery:
the “same” motive typically assumes a number of diverse shapes over the course of
a work. Thinking of a motive as a cognitive categor y makes it possible to account
for its identity, as well as its diversity, and reveals how aspects of categorization are
embodied by musical mater ials. These preliminary applications of categorization to
 conc e p tual i z i ng mu s i c 17

music also show ways musical mater ials can be organized over the course of a work
and offer an explanation of how it is possible to have musical concepts that are inde-
pendent of language.
If categor ization can be said to be the source of musical concepts, cross-domain
mapping is the means by which these concepts are placed in cor relation with oth-
ers. Chapter 2 examines the process of cross-domain mapping in some detail, begin-
ning with the work of cognitive linguists who, in the 1980s, proposed that metaphor
was a basic structure of understanding. This proposal gained added weight when it
was shown that metaphor ical projection (which is one way to accomplish cross-
domain mappings) was a general process not restricted to linguistic expressions but
grounded in embodied exper ience.
One example of cross-domain mapping that involves music in a rather immedi-
ate way is the technique of text painting, a compositional device that aims to rep-
resent in music specific images summoned by the text of a vocal work. Text paint-
ing provides a point of departure for the exploration of how cross-domain mapping
is manifested in our understanding of music, as it leads to an extension of cross-
domain mapping called conceptual blending. In a conceptual blend, elements from
two cor related domains are projected into a third, giving rise to a r ich set of possi-
bilities for the imag ination. As I show in the latter part of chapter 2, text painting
can lead to such blends, as can program music.
Chapter 3, which focuses on conceptual models and theor ies, gets to the heart of
the perspective on cognition developed in part I. My point of departure is research
by Jeanne Bamberger on children’s representations of musical structure. In my
analysis of Bamberger’s study of one specific eight-year-old boy, I show the part
played by categor ization and cross-domain mapping in the conceptual models used
by this boy to come to ter ms with a musical environment. I also show how these
models are combined to for m a theor y of music and how this theor y changes in
response to changes in the task at hand. This close-up glimpse of the structure and
role of conceptual models and theor ies leads, in the middle of the chapter, to a more
generalized character ization of these knowledge structures, which I connect with
work on similar structures in artificial intelligence, cognitive anthropology, ethno-
musicology, and developmental psychology. In the latter part of the chapter, I return
to music theory and explore the role of conceptual models and theor ies (that is,
theories framed relative to a cognitive perspective) in analyses by Jean-Philippe
Rameau and Heinr ich Schenker, two of the best-known music theor ists of the last
three hundred years.
Although the features of cognitive structure discussed in part I might seem to
be relatively detailed, in truth all are associated with relatively high-level cognitive
processes. My reason for focusing on this level is quite simple: it allows me to engage
in issues of immediate and occasionally central importance to music scholarship and
to do so in a way that connects with extensive research in cognitive psychology and
cognitive linguistics. Part II explores this possibility in g reater depth by consider-
ing various problems of musical understanding from the perspective on cognitive
structure and music theory developed in part I.
In chapter 4, I turn to the matter of how musical mater ials are organized within
a work — more properly, the problem of musical syntax and, by extension, musical
18 i nt roduc t i on

semiotics. Although semioticians are usually quick to grant that music has a syntax,
they are more doubtful about whether its semantic level has any depth. By taking
a close look at how composers make use of categor ies of musical events — in this
case, the way Mozart and Beethoven use motives in the opening movements of
three str ing quartets — I am able to provide insight into how musical mater ials are
organized in the service of musical discourse, as well as how features of this orga-
nization contr ibute to meaning construction as a whole.
In chapter 5, I confront a somewhat larger problem — one that may seem ir re-
deemably abstract: the problem of what counts as a work of music. I view this prob-
lem, usually called the problem of musical ontology, as one of cultural knowledge
and try to show that, as opposed to being hopelessly recondite, the problem is of
immediate importance for understanding music. By approaching the entire work of
music as a categor y— a category that includes all the scores, performances, repre-
sentations, and such that are said to be “of ” the piece — I develop a model for the
cultural knowledge upon which judgments about musical ontology are made.
Determinations of what counts as an instance of a particular musical work are thus
one of the ways members of a musical community construct and negotiate their
identity. My examples for this chapter are two songs taken from the traditions of
popular music and jazz: “I Got Rhythm” and “Bye Bye Blackbird.” The latter offers
an intr iguing case of how the cultural knowledge relative to which deter minations
of musical ontology are made can become complicated when implicated in the lay-
ered discourse structures Mikhail Bakhtin called “double-voiced discourse,” and
which were extended to African American culture through Henry Louis Gates Jr.’s
notion of “Signifyin(g).”
Chapter 6 returns to the analysis of individual musical works by pursuing one
of the entailments of text painting noted in chapter 2: under certain circumstances,
combinations of words and music, through the process of conceptual blending, cre-
ate worlds for the imagination well beyond those that spr ing from words or music
alone.Where only a few fairly circumscr ibed instances of text-music relations were
considered in chapter 2, here research on conceptual blending is applied to the
whole of five Lieder from the nineteenth centur y. These analyses offer a way to
flesh out the theory of conceptual blending as it applies to music and provide a fur-
ther perspective on musical syntax. In these songs, we see words and music com-
bining to create r ich domains in which the imagination can play, as well as discover
how musical syntax shapes our understandings of the words themselves.
The final analytical chapter (chap. 7) turns to music theory itself, specifically to
the theor ies of musical for m and hierarchy that go back to the eighteenth and early
nineteenth centur ies. Accounts of the for m a musical work takes, or of how its ele-
ments relate to one another, are basic to theor izing about music— indeed, we can
see these emerging in the course of M. Swann’s ruminations on Vinteuil’s sonata —
but at times it seems that theorists are talking about quite different things. For instance,
there are two common ways to talk about musical for m: the first approaches for m
as der iving from the assembly of relatively static building blocks that are combined
to create the finished work; the second approaches for m as an emergent property of
the work, which becomes manifest only as the music unfolds over time. The first
approach yields a view of musical for m that is quite static, the second a view rather
 conc e p tual i z i ng mu s i c 19

more dynamic. Using the analytical framework provided by cross-domain mapping,

I discuss the source of these two models for musical for m, as well as two models for
musical hierarchy, and explore some of the ways these models interacted over the
course of the history of music theory.
In the conclusion I return to M. Swann and to his final encounter with Vinteuil’s
sonata after a year in which it became thoroughly intertwined with his love affair
with Odette, the courtesan with whom he had become acquainted around the same
time he first encountered the andante. This will provide a frame for a review of the
points made in the preceding chapters and an instrumentality for drawing conclu-
sions from the whole.

cognitive structure,
theory, and analysis
A central claim of this book is that through developing an appreciation of how
aspects of cognitive structure shape our understanding of music we can better appre-
ciate the active role of theories of music in that understanding. A further claim is
that our analyses of musical phenomena — from the most mundane and localized of
accounts to the most abstract and comprehensive— similarly reflect cognitive struc-
ture, in that every analysis is based on some sort of theory of music. Musical analy-
ses are in truth dialogues, and not just dialogues between the analyst and an imag-
ined audience: musical analyses are also dialogues between the analyst and some
body of theoretical knowledge. Analysis rarely, if ever, simply cor roborates a theory:
analysis pulls theor y and pushes it, extending and chang ing theory just as it also
extends and changes our understanding of musical phenomena.
The analyses I present throughout this book are no different, except that they
engage cognitive theory as well as music theor y. The intent of the analyses is to
show how our understanding of particular musical phenomena can be character ized
in terms of specific cognitive processes and structures and thereby connect that
understanding to research in cognitive science as a whole. The analyses are not intended
as definitive statements about how we can account for such understanding; they are
intended to be the initiation of a dialogue with cognitive theory, a dialogue whose
purpose is to expand our knowledge of both music and cognition. Analysis is thus
a central concer n of what follows, but, unlike cognitive structure or theor y, I have
not treated it as a central topic for investigation. Instead, analysis will be a funda-
mental tool to explore both and to provide new insight into the conceptual worlds
wrought by musical sound.