Candace Paradoxes Pitch

DOI: 10.1111/j.1468-2249.2008.00268.
x
cOxford,
Music
MUSA
©
1468-2249
0262-5245
Original
XXX
p 2008Analysis
Blackwell
ANDACE
ARADOXESBlackwell
Article
bPublishing
UK OF PITCH
ROWER Publishing
sLtd
PACE Ltd.
CANDACE BROWER
PARADOXES OF PITCH SPACE
Music, like visual art, evokes images of space. The succession of pitches,
harmonies and keys which make up a tonal composition may evoke movement
through a space within which these musical ‘objects’ relate to one another as
near or far, inside or outside, above or below. Given that we can measure
distances between pitches according to their position on a frequency spectrum,
it is tempting to imagine that we can map this space in geometric terms,
representing pitches as points and keys as regions. Yet when we attempt to
locate these musical objects precisely in relation to one another within this
imaginary space, we quickly confront its contradictions. C1 and C8 are at
opposite ends of the frequency spectrum, yet the ear, and the principle of
octave equivalence, also tell us that they are the ‘same’.1 C major and A natural
minor contain the same pitches, yet we perceive them as distinct tonal regions
with different emotional qualities, C major sounding ‘bright’ or ‘happy’ and A
minor ‘dark’ or ‘sad’.
Composers at one time attributed different qualities to enharmonically
equivalent keys as well: Beethoven, for example, described the key of CC
major as ‘hard’ and Dw major as ‘soft’.2 Evidence that the distinction between
enharmonically equivalent keys continued to be taken seriously long after it
was supposed to have been perceptually erased by equal temperament can be
found in Max Reger’s Beiträge zur Modulationslehre (1903; translated as On the
Theory of Modulation (1948)). At the end of a series of exercises showing how
to modulate from C major to all other keys, Reger presents his version of the
shortest route for modulation from C major to B C major (Ex. 1). It is clear
from the succession of keys which Reger advances that he intends this progres-
sion to be heard as ending a full twelve semitones higher along the circle of
fifths than the point where it began. Yet given that the progression is only two
bars long, we cannot help but notice that the sonority at the beginning is
identical to the one at the end. In this passage, we are confronted with one of
the most striking paradoxes of the twenty-four-key system: that keys identical
in sound can nevertheless appear to be very far apart.3
Douglas R. Hofstadter (1999) notes that contradictions such as these are to
be found not only in music, but also in visual art and mathematics. Illustrating
his point with examples drawn from the music of J. S. Bach, the visual art of
M. C. Escher and the mathematics of Kurt Gödel, Hofstadter proposes that
these three modes of thought – musical, visual and mathematical – exhibit
Music Analysis, 27/i (2008) 51

© 2008 The Author.
Journal compilation © 2008 Blackwell Publishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK
and 350 Main Street, Malden, MA 02148, USA
52 CANDACE BROWER
Ex. 1 From Max Reger, On the Theory of Modulation (1948), p. 11
similar paradoxes, which he refers to as ‘strange loops’. Hofstadter makes

mathematics the central strand of his ‘eternal braid’; the other two, music and
visual art, provide colourful counterpoint to the subject of his ‘metaphorical
fugue’. This article will also draw together mathematics, music and visual art,
but with music placed in the centre of the frame. It will show how mathemat-
ical models of musical and visual space on the one hand, and embodied models
of musical and visual perception on the other, provide a framework which
enables us to explain the paradoxes of pitch space by portraying them in visual
form, thereby allowing us to ‘see’ how they work.4
Mathematical Models of Musical and Visual Space

Given that spatial relations in music, like those in visual art, can be described
in geometric terms, we should not be surprised to discover that some of the
same mathematical tools have been applied to these two art forms. The parallels
between mathematical models of musical space and those of visual space have
been brought to the fore by the recent emergence of neo-Riemannian and
transformational theory, as can be seen by comparing Figs. 1 and 2.
Fig. 1 shows that now-familiar model of triadic pitch space, the Tonnetz, as
it appears in an introductory article on neo-Riemannian theory by Richard
Cohn (1998).5 Strikingly similar, Fig. 2 is an example of what mathematicians
refer to as tiling of the plane. To tile a plane means to cover it with polygons
aligned edge to edge, with no overlap and no spaces left unfilled.6 Tilings, part
of the decorative arts since ancient times, have been the subject of mathemat-
ical study since the early seventeenth century, when Johannes Kepler (1619)
discovered all eleven of the so-called Archimedean tilings of the plane.
Comparison of Figs. 1 and 2 reveals that the two are geometrically similar,
each consisting of a gridlike arrangement of right-side-up and upside-down
triangles. Yet further parallels exist between them, reflecting shared principles
of organisation. First, each was created by replicating a single object – its basic
shape. In the case of the Tonnetz, this basic shape is the major triad; in the case
© 2008 The Author. Music Analysis, 27/i (2008)

Journal compilation © 2008 Blackwell Publishing Ltd.
PARADOXES OF PITCH SPACE 53
Fig. 1 Neo-Riemannian Tonnetz
Fig. 2 Tiling of the plane
of the tiling, it is the equilateral triangle. Mathematicians refer to such an

object as a prototile.
Second, the objects embedded within each grid can be manipulated in
similar ways. In tiling the plane, replication of the prototile is carried out through
what mathematicians call congruence transformations. As shown in Fig. 3, these
consist of four basic types: translation, rotation, reflection and glide reflection.
Translation shifts a tile (or group of tiles) a certain distance in one direction
while preserving its overall orientation; rotation changes the orientation of a
tile by rotating it around a point within the plane; reflection flips a tile from
front to back across an axis lying within the plane; and glide reflection com-
bines reflection with translation. As we can see by comparing Figs. 1 and 3,
the neo-Riemannian operators D, P, R and L form a subset of this family of
transformations. The D operator, by which a triad is transformed into its
dominant or subdominant, corresponds to translation, while P, R and L oper-
ators, by which a triad is transformed into its parallel, relative or Leittonwechsel,
correspond to reflection across the axis of its fifth, major third or minor third,
respectively.
Music Analysis, 27/i (2008) © 2008 The Author.

54 CANDACE BROWER
Fig. 3 Congruence transformations
Fig. 4 Period parallelograms
Third, each of the patterns shown can be understood to repeat indefinitely

in all directions. Mathematicians refer to such a pattern as a periodic tiling; it is
constructed by replicating a period parallelogram (a parallelogram defined by the
two basic translation vectors of the tiling). As Fig. 4 shows, the tiling of Fig. 2
can be constructed from any of three different period parallelograms, each
formed by joining a right-side-up triangle to one of its three reflections.
Repeated translation of any one of these objects in nonparallel directions will
eventually cover the Euclidean plane, in the process forming a set of infinitely
extending parallel axes which criss-cross in three directions. In the case of the
Tonnetz, the pattern which is replicated is that of a major triad joined to a
neighbouring minor triad, producing infinitely extending axes of perfect fifths,
major thirds and minor thirds.
Finally, each grid can be made to take the shape of a torus. The tiling of
Fig. 4 can be shaped into a torus by rolling the represented parallelogram up
into a cylinder and then bending the cylinder around to join the ends.7 The
same can be done with the Tonnetz after first modifying it as shown in Fig. 5.
In this modified Tonnetz, because equal temperament and enharmonic
equivalence are presumed, pitch numbers take the place of pitch names, reducing

Fig. 5 Toroidal Tonnetz
the number of distinct points within the grid from a potentially infinite number
to twelve. These are arranged in the shape of a three-by-four-unit period
parallelogram, which may be rolled up on itself in two directions, as shown by
the arrows, to form a doughnut-shaped torus.8
Paradoxes of Visual Space in the Artwork of M. C. Escher

Just as music theorists have used the mathematics of neo-Riemannian theory
to explore the geometry of musical space, visual artists have employed the
mathematics of tiling to explore the geometry of visual space. Most notable
among them is M. C. Escher.9 Escher became fascinated with tilings of the
plane after seeing the elaborate tile mosaics of the Alhambra Palace in Granada,
Spain. Intrigued by these figures, he set out to explore systematically the
possibilities for periodic tiling of the plane. The resulting notebook of sketches
served as the basic material for much of his later artwork.10
However, Escher was not content simply to fill his artworks with abstract
geometric figures. Instead, he carefully transformed each figure, bringing it to
life by carving away portions of one side of a polygon to add to another until
the original figure – triangle, square, rhombus or hexagon – took the shape of
a bird, fish, lizard, horse or frog. We see this process of transformation in
Mouch (Fig. 6), whose succession of frames, numbered 1 to 12, reveal the
artist’s working method. By transforming geometric figures into living
creatures, Escher is able to highlight certain paradoxes of visual perception.

56 CANDACE BROWER
Fig. 6 M. C. Escher, Mouch
Their existence reveals how the human eye elaborates upon what is physically
seen, suggesting that while visual space has properties in common with physical
space, it also possesses a logic of its own.
In Mouch our attention is focused on the paradox of figure-ground reversal.
When the eye is presented with a boundary dividing the visual field into two
regions, it attempts to interpret each as lying either inside or outside – that is,
as either figure or ground. In Mouch, figure-ground ambiguity results from the
periodic division of the plane into black and white figures, which we perceive
alternately as black objects against a white background or white objects against
a black background. The paradoxical effect is heightened by Escher’s use of
another compositional device, that of metamorphosis. The black and white
parallelograms which emerge in frames 1–4 are first transformed into black and
white birds (frames 5–10), then into black and white fish (frame 11) and finally
into white fish swimming to the left interspersed with black birds flying to the
right (frame 12).

Escher was likewise fascinated by the paradox of impossible pathways for

movement. In one of his best-known works, Ascending and Descending, the
paradox of a staircase which doubles back on itself in a circle is endowed with
human significance by the two lines of human figures, one apparently moving
continuously upwards and the other continuously downwards.11 In Ascending
and Descending, Escher takes advantage of the eye’s propensity to interpret
two-dimensional images as three-dimensional while at the same time he
provides contradictory cues for distance, height and depth to create a pathway
which continuously rises while circling back upon itself to remain in place.
Escher’s Möbius Band II confronts us with another ‘impossible’ pathway
for movement: a one-sided yet two-sided surface, along which one can trace a
continuous pathway from front to back. In this case the paradox is not just
visual but physical: we can easily construct a Möbius strip by twisting a
rectangular strip by 180° and joining its ends. Creating such an object allows
us to discover firsthand its paradoxical properties. By placing a finger on one
side of the strip and a thumb on the other, we can discover its local property
of two-sidedness; by tracing our finger smoothly along its surface from front
to back, we can discover its global property of one-sidedness.12 In Möbius
Band II the paradox of a one-sided yet two-sided surface is brought to our
attention by the progression of a single, continuous line of ants travelling from
‘front’ to ‘back’.
Paradoxes of Musical Space in Wagner’s Parsifal

It was in part the presence of similar paradoxes in the chromatic music of
Wagner and other nineteenth-century composers which inspired the develop-
ment of neo-Riemannian theory. David Lewin (1984) suggests that Wagner
uses a kind of sleight of hand to lead us from the everyday diatonicism of
‘Stufen space’ into the otherworldly enharmonicism of ‘Riemann space’. To
identify the differences between Stufen space and Riemann space, Lewin
invokes the metaphor of a Möbius strip intersecting with a cylindrical loop,
proposing that we pass from one to the other via a ‘fault in the terrain’, a
‘splice’ (1984, p. 347), created by an enharmonic Cw/B.
Richard Cohn (1996 and 2004) has likewise pointed out the paradoxical nature
of hexatonic polar progressions, found not only in Parsifal but also in many
other chromatic works of the nineteenth century. Noting that this progression
has struck many commentators as ‘uncanny’, ‘supernatural, magical, [and]
weird’ (Cohn 2004, p. 286),13 Cohn describes the relationship between the
triadic poles of a hexatonic system as ‘a musical analogue of a Lewinian trope,
the recursive gaze of Sieglinde and Siegmund’ (Cohn 1996, p. 21). He notes
that Wagner makes particularly effective use of this progression in the trans-
formed version of the ‘Grail’ motive which appears at the very end of Parsifal,
shown in Ex. 2. As Cohn observes, this progression conveys the mystery of the
final event of the opera: Kundry’s soul taking leave of her body.

58 CANDACE BROWER
Ex. 2 Wagner, Parsifal, Act III, bars 1123–1127 (Kundry’s Entseelung)
Ex. 3 Hexatonic polar progressions
According to Cohn, the uncanniness of the hexatonic polar progression is

due to two notable features: first, its maximally smooth voice leading, illus-
trated in Ex. 3, which gives rise to a powerful sense of resolution at the local
level; and second, the extreme symmetry of the hexatonic collection, which
makes it impossible to locate a triadic centre. As a result, each triadic pole can
be heard to resolve the other, sounding alternately consonant or dissonant,
stable or unstable.
The means by which the symmetry of hexatonic space contributes to the
paradox of the hexatonic polar progression is revealed more clearly in Fig. 7.
It shows how Northern and Southern hexatonic systems divide the neo-
Riemannian Tonnetz symmetrically into halves, each of which is itself divided
symmetrically into its polar complements.14 The opposing perceptions of the
consonance-dissonance relationship between hexatonic poles can be compared
to what we experience when we look at a Necker cube (Fig. 8). The cube
presents us with two equally possible but contradictory views of the same
object, one consisting of the cube as viewed from above, the other from below.
In hexatonic space such paradoxes appear not to be the exception, but the rule.
It is a space within which two similar percepts can easily vie for our attention,
neither asserting dominance over the other.

Fig. 7 Hexatonic polar progressions within the neo-Riemannian Tonnetz
}
}
Fig. 8 Necker cube
Neo-Riemannian theory has shed new light on the paradoxical nature of

musical space by providing us with the means of visualising the geometry of
triadic pitch space as a whole. Not only does it reveal symmetries which might
otherwise go unnoticed, allowing us to see how spatial symmetry contributes
to the paradox of the hexatonic polar progression, but it also neutralises the
diatonic bias of everyday musical experience, thus allowing us to explore
alternative modes of perception.
However, as this article will show, the mathematics of neo-Riemannian the-
ory tells only part of the story. For just as we grasp the paradoxical nature of
Escher’s Ascending and Descending only by projecting ourselves into its world to
experience vicariously the steps of the figures ever trudging both upwards and
downwards, so we grasp the paradoxical nature of the polar progressions of
Wagner’s Parsifal only by projecting ourselves into its musical space, thereby
experiencing from within that space the feelings of decentredness through
which Wagner conveys the disembodiedness of Kundry’s ‘de-souling’.15 To
explain such paradoxes fully we must examine the perceptual processes which
underlie our embodied experiences of both music and visual art.
What do the bodily sensations evoked by the music of Wagner’s operas have
in common with those evoked by Escher’s ‘impossible’ representations? Recent
theories of embodied cognition suggest that our most instinctive responses to
both music and visual art originate in the body. It appears that musical sounds,
like visual images, evoke bodily sensations due to our metaphorical projection of
them into the domain of embodied experience. Furthermore, certain recurrent

60 CANDACE BROWER
Fig. 9 Embodied image schemas
patterns of embodied experience, known as image schemas, appear to play a

central role in such metaphorical projections.16
The Embodied Nature of Musical and Visual Perception

Many cognitive scientists now believe that image schemas play a central role in
human thought and perception, serving as patterns upon which we draw in
order to construct meaning in the world around us. These schemas – sensorimotor
images constituting the basic ‘shapes’ of bodily experience – are abstracted
from such everyday experiences as balancing the body while walking, reaching
out and grasping an object, moving along a pathway towards a particular goal,
rising to a standing position and entering or exiting a closed space. It appears
that we use these image schemas to make sense of patterns perceived in more
abstract domains through metaphorical projections, or cross-domain mappings,
which imbue those patterns with bodily meaning.
Among the image schemas that appear to organise our bodily experience are
those shown in Fig. 9. Figs. 9a–c capture important features of our experience
of the body. The VERTICALITY schema, represented as a vertical axis extending
upwards from a horizontal plane, illustrates the way in which we experience
the body in relation to the ground under the force of gravity. At a somatosensory
level, we experience the ground as a maximally stable location, while associating
ascending motion with overcoming the force of gravity (thus effortful or tensing)
and descending motion as giving in to the force of gravity (thus relaxing). The
CENTRE⁄PERIPHERY schema reflects our experience of the body as the centre and
of the rest of the physical world as extending outwards from it in all directions.
At a somatosensory level, we associate centredness with feelings of stability and

rest, and departure from a centre with feelings of instability and an impulse to
return. The BALANCE schema reflects the ways in which we unconsciously
balance the forces acting upon the body so as to remain stable, upright and at
rest. Fig. 9c specifically reflects our somatosensory experience of the symmetry
of the human body in terms of the weight of the left side balancing that of
the right.
The schemas presented in Figs. 9d, e and f reflect ways in which we expe-
rience the body in motion. The PATH schema shows that we tend to move along
pre-established pathways leading to specific destinations. Paths are either
constructed or worn over time through repeated movement along the same
trajectory; thus they become associated with such properties as continuity,
regularity and predictability. The CONTAINER schema reflects our experience of
space as bounded – that is, as divided into insides and outsides. We experience
the body itself as a container and its boundaries as creating the division between
self and non-self. We also experience ourselves as moving within and between
containers, such as rooms, buildings, cities and regions. We may associate
containment either with stability and protection from external forces or,
conversely, with a lack of freedom, with constraint and inhibition of movement.
Finally, the CYCLE schema reflects the way in which we experience repeating
patterns in space and time. We learn about cycles most directly through such
repetitive actions of the body as breathing and walking, as well as from such
higher-level cycles as waking and sleeping. The CYCLE schema regulates our
experience of time, allowing us to anticipate future events, and is thus associated,
like the PATH schema, with continuity, regularity and predictability.
Certain general features of the image schemas shown in Fig. 9 should be
kept in mind when considering the roles they play in cross-domain mappings.
First, although they are typically presented as visual images, image schemas
combine input from all of the senses; each binds together visual, auditory,
motor and somatosensory information into a single experiential gestalt. For
example, the VERTICALITY schema includes not only the visual image of vertical
and horizontal axes, but also the somatosensory experience of our feet resting
on the ground and the kinaesthetic sensation of our muscles tensing to hold
the body upright.
Second, many image schemas are topological; that is, they can be fitted to
patterns of different shapes and sizes. Thus the CONTAINER schema, although
drawn as a circle in Fig. 9e, can be applied to any bounded region of space,
regardless of its shape or the number of dimensions in which it exists. The
CONTAINER schema, when mapped onto a bounded region of visual space,
causes us to perceive whatever falls on one side of the boundary as figure and
whatever falls on the other as ground. Thus the CONTAINER schema plays an
important role in determining the way in which we divide visual space into
objects, parts of objects and the empty space that surrounds them.
Finally, many of these image schemas have features or properties in common
that allow them to be combined, as illustrated in Figs. 10a–c. For example,

62 CANDACE BROWER
Fig. 10 Combined image schemas
Fig. 11 From Rudolf Arnheim, Art and Visual Perception: A Psychology of the Creative
Eye, published by the University of California Press, p. 12. © 1997 by the Regents
of the University of California
CONTAINER, CENTRE-PERIPHERY and BALANCE schemas may be combined because

they share the property of a centre, as shown in Fig. 10a. Likewise, VERTICALITY
and PATH schemas may be combined because they share the property of linearity,
yielding the schema for a VERTICAL PATH shown in Fig. 10b. Finally, CYCLE and
CONTAINER schemas may be combined because they share the property of
closure, yielding the schema for a CIRCULAR PATH shown in Fig. 10c.
The role these image schemas play in visual perception has been described
by Mark Johnson (1987). Johnson shows how the BALANCE schema affects our
experience of balance in visual art, illustrating his points with reference to
experiments carried out by Rudolf Arnheim on the dynamic properties of
visual ‘objects’ (Arnheim 1997). In one such investigation Arnheim presented
viewers with the image of a disk within a square, such as that shown in Fig. 11,
then asked them to describe their experience of the disk as either at rest or
subject to forces of attraction or repulsion at different points within the square.
Fig. 12 illustrates the ways in which viewers experienced the disk as acted upon
by forces at different locations within the square, constituting what Arnheim
refers to as its ‘structural skeleton’. This figure suggests that the viewer’s sense

Fig. 12 Arnheim’s structural skeleton of the square. From Rudolf Arnheim, Art
and Visual Perception: A Psychology of the Creative Eye, published by the University
of California Press, p. 13. © 1997 by the Regents of the University of California
Fig. 13 Effect of shape and spatial orientation on visual stability. Adapted in part
from Rudolf Arnheim, Art and Visual Perception: A Psychology of the Creative Eye,
published by the University of California Press, p. 102. © 1997 by the Regents of
the University of California
of balance is governed by the geometry of the square – its centre, edges,

vertices and axes of symmetry. Further, it indicates that the viewer experiences
the disk as most stable when it appears at the centre (its four axes of symmetry
defining secondary areas of repose). The edges of the square also contribute to
the viewer’s perception of balance by exerting apparent forces, first of attraction
as the disk approaches an edge, then of repulsion as it moves ‘too close’. We
can see in Arnheim’s structural skeleton the combined influence of CENTRE-
PERIPHERY, BALANCE AND CONTAINER schemas.
Arnheim further observes that our experience of visual ‘objects’ as stable or
unstable depends upon their spatial orientation. He notes that a right-side-up
triangle, such as that shown in Fig. 13a, appears to rest solidly on its base,

64 CANDACE BROWER
while the upside-down version in 13b seems precarious, ready to topple to one
side. Figs. 13c and d reveal even more clearly the combined effects of symmetry
and spatial orientation. Although the equilateral triangle of Fig. 13c seems
equally stable on each of its sides, the unequal triangle of Fig. 13d produces
three different percepts, each more unstable than the last, reflecting the
apparent rise in its centre of gravity and less evenly balanced distribution of
weight. Figs. 13a–d clearly illustrate the contribution of VERTICALITY and
BALANCE schemas to our embodied experience of visual ‘objects’, showing that
we expect them to exhibit properties analogous to those exhibited by physical
objects under the influence of the physical force of gravity.
That our embodied experience of visual ‘objects’ is determined in part by
their geometry and spatial orientation suggests that the same may be true of
the musical ‘objects’ represented in Fig. 1. At first glance, Fig. 1 might appear
to capture our intuition about the relative stability of major and minor triads
by representing the major triad as a right-side-up triangle resting on its stable
base and the minor triad as an upside-down triangle perched on its apex. Yet,
as Fig. 14 shows, this image presents a rather skewed picture of the triad’s
symmetry and spatial orientation. First, by representing the consonant triad as
an equilateral triangle, the neo-Riemannian model suggests that the triad’s
tones are equidistant and thus imposes upon this type of chord a symmetry
which it does not possess. Second, as the arrows in Fig. 14 indicate, the model
portrays ascending intervals within the space of the triad as moving in three
different directions: while the major third moves ‘upwards’, the minor third
moves ‘downwards’, and the perfect fifth moves ‘horizontally’ across the page.17
Fig. 15 presents a modified version of Fig. 14 which preserves the original
topology while changing its geometry and spatial orientation in order to reveal
Fig. 14 The neo-Riemannian ‘basic shape’
Fig. 15 Modified version of the neo-Riemannian ‘basic shape’

more clearly its image-schematic organisation. It first rotates the figure by 90°
to place C at the bottom and G at the top, then flips it across the vertical axis
to position Ew on the left and EÖ on the right, locating these vertically so as to
divide the seven semitones of the perfect fifth into 3 + 4 on the left and 4 + 3
on the right. The positioning of C at the bottom and G at the top allows every
rising interval to ascend within the visual space of the page, thereby capturing
our intuition that the root of a triad corresponds to its lowest, most stable note.
By focusing attention on the vertical arrangement of notes within the triad,
Fig. 15 also highlights an important difference between the geometry of visual
‘objects’ and that of musical ‘objects’; that is, visual ‘objects’ have both height
and width, but musical ‘objects’ have only height. This can be seen clearly in
Fig. 15 in the way that the ‘sides’ of the C major ‘triangle’ fail to add up. To
form a triangle with sides three, four and seven units in length, we would have
to set the internal angles at 180° and 0°, which would cause the shorter sides to
collapse onto the longer one. This lack of a horizontal dimension also applies
to the space within which we imagine musical ‘objects’ to be in motion.
Although we often describe notes and triads as moving ‘up’ or ‘down’, it seems
nonsensical to describe them as moving to the ‘left’ or to the ‘right’.
What, then, is the meaning of the horizontal dimension in Fig. 15? What
determines the placement of pitches along the horizontal axis? These questions
are answered visually in Fig. 16. As the arrows in Fig. 16 show, we can con-
ceive of the triangles used to model major and minor triads as having been
formed by displacing the third of each triad away from the vertical fifth axis,
the third of the minor triad to the left and the third of the major triad to the
right. This allows the geometry of the space as a whole to be seen more clearly,
much as we might open out the gathered folds of a net in order to reveal the
geometric pattern of its construction. Not only does the horizontal displace-
ment of the thirds allow major and minor triads to be more easily differentiated
by eye, but it also allows the interval of a semitone (that is, Ew to EÖ) to be read
as ascending from left to right. Thus, Fig. 16 makes clear that the horizontal
dimension is nothing more than an aid to visualisation. Although the horizontal
Fig. 16 Reinterpretation of the neo-Riemannian ‘basic shape’

66 CANDACE BROWER
displacements of Ew and EÖ make them look as if they are far apart, measurement
along the vertical axis reveals that they are in fact quite close.
Although Fig. 16 allows us to ‘see’ the vertical arrangement of tones within
major and minor triads, it offers no clue as to their relative stability. This
reflects the fact that whereas visual ‘objects’ map more or less directly onto
physical objects through their shared projection on the retina, the mapping of
visual ‘objects’ onto musical ‘objects’ involves not only different domains but
also different sense modalities – sight versus sound. Thus we should not expect
the visual model to portray directly the embodied qualities of musical ‘objects’,
but only those features of their geometry which contribute to their image-
schematic organisation.18
It is to psychoacoustics that we must turn for clues concerning our embod-
ied experience of musical sounds. Research in this field suggests that our
perception of tone combinations as stable or unstable is governed in part by
the way in which their intervallic patterns correlate with that of the overtone
series. It has been found that the overtone series, having been stored in mem-
ory through repeated exposure to the complex sounds of music and speech,
serves as a template for pattern matching carried out at an unconscious level.19
It is here that we find the most likely explanation for our perception of the
major triad as more stable than the minor: its intervallic pattern forms a better
match with the intervallic pattern of the overtone series.
We can gain further insight into the embodied nature of our experience of
musical sounds by examining the image-schematic organisation of the overtone
series itself. As Fig. 17 shows, the overtone series takes on an image-schematic
structure of its own when mapped onto a VERTICAL PATH schema. Fig. 17
portrays the fundamental of a pitch as its ‘ground’ – the lowest, most stable
location within the space of a complex pitch – and the upper partials as the rungs
Fig. 17 Mapping of the overtone series onto a VERTICAL PATH schema

of a vertical pathway or ‘ladder’ which ascends indefinitely towards a vanishing

point on the perceptual horizon. This image suggests that our experience of the
root of a triad as the lowest, most stable pitch can be attributed to its mapping
onto the fundamental – the ‘ground’ element of the overtone series.
Fig. 17 also provides an image-schematic explanation for our experience of
fifths and thirds as pathways for harmonic motion. As Fig. 17 shows, the first
six notes of the series provide two distinct pathways for ascent from C to G,
the first proceeding directly by fifth, and the second indirectly by major and
then minor third. Fig. 16 represents these alternative pathways, the direct route
from C to G vertically and the more indirect route from C to E to G as
zigzagging first to the right and then to the left.
The fact that pitches are replicated within the overtone series helps to
explain not only why we interpret C–G and C–E–G as alternative pathways for
harmonic motion, but also why we reduce the pitches making up the major
triad to three.20 C2 and C3 are perceived to be the ‘same’ as C1; thus they
take on similar feelings of ‘groundedness’, reinforcing our impression that the
root of a triad, regardless of the register in which it sounds, corresponds to its
‘lowest’, ‘most stable’ pitch. Likewise, we perceive G3 to be the ‘same’ as G2,
reinforcing our impression of the fifth of a triad as its ‘highest’ note and of the
third as the ‘middle’. This helps to explain why movement from root to fifth,
or from root to third to fifth, tends to be associated with tensing and its
opposite with relaxing, as shown by the arrows on the right.21 These feelings
of tension and relaxation may be understood as a reflection of the bodily
sensation of moving up and down with respect to the ground.
In its replication of pitches from one octave to another, we find another way
in which pitch space differs from both visual and physical space, yielding a
dimension not easily captured in the visual domain. Psychologists account for
this dimension by describing pitch as having not one but two components,
register and chroma. Change in register is modelled as movement along a vertical
path and change in chroma as movement along a circular path, as shown in
Fig. 18a.22 The two components are sometimes modelled together in the form
of a ‘spiral staircase’, as shown in Fig. 18b, which portrays our experience of
increasing frequency as continuously ascending motion accompanied by a
repeating cycle of changing timbral colouration within each octave.23 This
model of a spiral staircase has been used by the psychologist Roger Shepard to
create so-called Shepard’s tones, a chromatic scale within which each component’s
distribution of partials gradually shifts downwards as the pitch ascends, so that
by the time the pitch has risen a full octave, the distribution of partials is the
same as was the case at the outset.24 The result is a musical version of Escher’s
Ascending and Descending – a scalar series of pitches which can be heard to rise
continuously while also remaining in place.25
This pattern of repeating tone colours lacks any perceptual correlate in
visual space. Although Fig. 16 does not portray this cyclic dimension of
perceived pitch directly, the dotted vertical lines in the background hint at its

68 CANDACE BROWER
Fig. 18 (a) Register and chroma; (b) Shepard’s tones
existence. We can conceive of each vertical line as having been formed by a

metaphorical tracing wheel which was rolled from the bottom to the top of the
page; the serrations along its outer rim represent the twelve pitches of the
chroma circle. Thus, each dotted vertical line is the imprint of this repeating
series of twelve chromatic pitches upon the page.
We can use the image-schematic model of the C major triad shown in
Fig. 16 to construct an embodied model of triadic pitch space which incorpo-
rates these geometric and image-schematic features, thus enabling us to model
our embodied experience of motion within triadic pitch space as a whole.
Using principles of tiling, we can form additional triads from the C major triad
which fit together to form the larger musical space within which we experience
movement among pitches, triads and keys. We may begin with a purely tuned
C major triad, making adjustments of tuning along the way as needed in order
to permit the musical space to close in on itself and thus to form both diatonic
and chromatic collections.
Rather than positing the space all at once, as was done with the neo-
Riemannian Tonnetz, we will construct it in separate stages by applying two
principles in alternation.26 First, during each stage we will add only those triads
needed to achieve some new form of symmetry and/or closure. Second, we will
continue to add triads until musical space closes in on itself in all directions to
form an embodied version of the neo-Riemannian Tonnetz. In the course of
constructing this space, we will witness the formation of a set of nested spaces,

each exhibiting properties of closure and/or symmetry which allows the space
to be mapped onto a CIRCULAR PATH and/or a CONTAINER⁄CENTRE-PERIPHERY⁄
BALANCE schema, thus giving rise to the embodied experience of containment,
centredness and balance. We will also witness the emergence of spatial para-
doxes analogous to those found in the artwork of M. C. Escher: figure-ground
reversal, circular staircases and the Möbius strip.
Because the present model is constructed using neo-Riemannian operators
(D, P, L and R) and its completed structure resembles that of the neo-Riemannian
Tonnetz, it is important to note the ways in which it differs from the neo-Riemannian
model. First, although neo-Riemannians disregard the acoustic origins of the
consonant triad, according equal weight to root, third and fifth, the present
model privileges the root, reflecting the contribution of these origins to our
sense of bodily orientation within triadic pitch space. Moreover, it views the root
as standing in some sense for the triad as a whole, much as the fundamental
does for a complex pitch. Second, although neo-Riemannian theory defines
harmonic distance in terms of voice leading, thus privileging L, P and R
transformations as maximally smooth, the present theory defines harmonic
distance in terms of root movement. This is because voice-leading motion is
presumed to take place within another dimension of musical space, one which
I have referred to elsewhere as melodic space (Brower 2000). In consequence,
whereas neo-Riemannians treat D as equivalent to L + R because these
operations yield identical voice leadings, the present model views the same
operations as distinct: D produces the harmonic motion of a single step along
a pathway of fifths, while L + R produces the motion of two steps along a
pathway of alternating major and minor thirds. Finally, although neo-Riemannians
treat P, L and R as a family of operators because each produces stepwise motion
within a single voice, the present model treats P as distinct from L and R
because it produces not root movement, but a change in the position of the
third of the triad with respect to its root and fifth.
The difference between these two interpretations of harmonic motion is
illustrated in Ex. 4. Ex. 4a depicts the progression C major–A minor as root
movement; thus, it maps root onto root while preserving internal relations
among pitches. By contrast, Ex. 4b represents the same progression as voice
leading; thus it preserves the common tones while mapping the fifth of the first
Ex. 4 Root movement and voice leading

70 CANDACE BROWER
triad onto the root of the second so as to create movement along the shortest
melodic path. These two representations are combined in Ex. 4c, which shows
both the harmonic pathway followed by root movement in the bass and the
melodic pathways followed by voice leading in the upper voices.
The present model’s treatment of root movement and voice leading as
distinct components of harmonic motion helps to highlight the paradoxical
nature of the relationship between them. Harmonic and melodic distances tend
to be inversely related, so that what sounds melodically close often sounds
harmonically distant and vice versa.27 Furthermore, root movement and voice
leading tend to proceed in opposite directions, with root movement down by
fifth or third producing voice leading upwards by step and vice versa. As a
result, whether the perceived direction of harmonic motion will be governed
by root movement or voice leading cannot be determined independently of the
musical context. In passages predominantly made up of root-position triads,
our sense of direction is likely to be governed by root movement; in passages
formed primarily from inverted triads, it is likely to be governed by voice leading.
The model also shares certain features with the cognitively derived model of
pitch space advanced by Fred Lerdahl (2001). Using empirical data provided
by Diana Deutsch and John Feroe (1981), Carol Krumhansl (1990 and 1998)
and others, Lerdahl constructs three different models of tonal pitch space, which
he refers to as basic pitch space, chordal space and regional space. His basic pitch
space corresponds to what I am calling melodic space, while his chordal and
regional spaces combine in the present model to form what I am calling triadic
space. Lerdahl criticises Riemann and his followers for attempting to represent
distances within musical space using a single map, namely that of the Tonnetz,
in which, owing to a conflation of descriptive levels, points are sometimes taken
to represent pitch and on other occasions triad or key. The present model
avoids this problem by depicting pitches as points and triads and keys as
groups of points, each of which forms a closed region of musical space.
Because Lerdahl’s model represents triads and keys, like pitches, as points
rather than as regions, it cannot accommodate overlaps among these regions.
This is particularly problematic in the case of relative major and minor keys. I
will show here that the latter do not constitute different regions of musical
space, but rather alternative ways of perceiving the same region.28
In emphasising the embodied character of our experience of movement
through triadic pitch space, I do not claim that all listeners will hear changes
of harmony or key as rising or falling, or tensing or relaxing, movement. Nor
do I suggest that we can extrapolate from the experiences of Western subjects
to those of non-Western listeners. Because such metaphorical mappings require
active engagement of the bodily imagination, as well as familiarity with the
conventions of the major-minor system, individual listeners may differ
markedly in the degree to which they experience such sensations.29 Neverthe-
less, we find evidence of significant intersubjective agreement among trained
listeners concerning such experiences in the form of a strong consensus among

eighteenth- and nineteenth-century composers and theorists as to how these

properties correlate with specific changes of harmony and key (Steblin 2002).
Construction of the triadic pitch-space model in stages also provides the
basis for a more ambitious claim, namely, that the logic governing this
construction is similar to that which governed the evolution of the tonal
system, from the emergence of triadic harmony in the fifteenth century, with
its emphasis on diatonicism and the major mode, to the extreme chromaticism
of the late nineteenth century and the emergence of hexatonic and octatonic
collections. One could view this evolutionary pathway as reflecting the combined
workings of two opposing compositional impulses: the impulse continually to
push the boundaries of triadic pitch space outwards to expand the possibilities
for harmonic motion, and the contrary inclination to form spaces which are
symmetrical, centred and/or contained. In the final section of this article,
support for this claim will be offered in the form of analyses of works from the
fifteenth to the nineteenth centuries which hint at this pathway. These analyses
also provide evidence that the paradoxes of triadic pitch space are more than
purely mathematical abstractions, which suggests that at least some composers
have intentionally exploited these paradoxes in order to evoke movement of the
body through a space topologically very different from our own.
Constructing an Embodied Model of Triadic Pitch Space

In the first stage of construction we replicate our C major ‘prototile’ above and
below via translation to form the symmetrical space shown in Fig. 19a. This
expands the total number of pitches from three to seven – from those of the C
major triad to those of the key of C major – while extending the vertical
pathways for movement from the single fifth C–G to a chain of fifths, F–C–G–
D, and from the succession of thirds C–E–G to a chain of alternating major
and minor thirds, F–A–C–E–G–B–D. Within this space we may conceive of a
succession of pitches as the movement of a single note along any of the pathways
illustrated and a succession of triads as the movement of a single triad along
the pathway of fifths shown on the left.30 From a mathematical perspective, the
Fig. 19 Construction of triadic pitch space, stage 1

72 CANDACE BROWER
latter involves using the D operator to transform a triad into its dominant and
the S operator to transform a triad into its subdominant.31 From an embodied
perspective, these correspond to motion upwards or downwards, tensing or
relaxing, respectively, as shown by the arrows on the right.
Fig. 19b presents an alternative mapping of this constructed space onto a
CENTRE-PERIPHERY⁄BALANCE schema. It portrays the key of C major as a musical
‘object’ within which are nested three smaller objects, IV, I and V, with I at the
centre balanced by V above and IV below. This mapping reveals another dimen-
sion of our embodied experience of motion within diatonic space, that of the
tonic as centre and thus a maximally stable location. This in turn supports our
experience of the motion from tonic to either dominant or subdominant as
tensing and the motion of return as relaxing. On an embodied level, we can
understand this as reflecting the tensing and relaxing of our muscles as we
move with respect to our own centre of gravity.
The two mappings of Figs. 19a and b yield somewhat conflicting image-
schematic interpretations, with motion from tonic to subdominant connoting
relaxation in the mapping of the VERTICALITY schema and tension in the map-
ping of the CENTRE-PERIPHERY⁄BALANCE schema. This suggests that alternative
image-schematic mappings may at times yield contradictory meanings, and
that our embodied experience of musical sound may at times reflect those
contradictions.
In the second stage of expansion, shown in Fig. 20a, we begin by filling in
the spaces left open in Fig. 19 with E minor and A minor triads formed
through the operation of reflection, thus creating closure while preserving
symmetry. It is at this stage of construction that the minor triad first appears
as a musical ‘object’. The relationship between major and minor triads can be
viewed as analogous to that which emerges between the black and white areas
in Escher’s Mouch: areas initially perceived as ground take on the characteristics
of figures because of their similarity to the objects that they surround.

Transformation of a major triad into a minor triad produces a lowering of

the third which we experience on an embodied level as a decrease in its stability,
reflecting its imperfect correspondence to the overtone series. The bodily
sensation of the lowering of the third can be understood in relation to the
VERTICALITY schema as akin to the weakening of the muscles which hold the
body upright. This contrast in the affective quality of major versus minor has
been expressed in such terms as ‘strong’ versus ‘weak’, ‘hard’ versus ‘soft’,
‘happy’ versus ‘sad’ or ‘bright’ versus ‘dark’.32 Because these properties are not
directly represented by the visual model, Fig. 20 reminds us of them by shading
major triads light and minor triads dark.
The addition of two minor triads expands the possibilities for harmonic
motion by accommodating root movement by fifth along the newly formed
pathway of fifths on the right, as well as root movement by third along the
zigzag chain of thirds. We can trace a descent along this pathway by applying
L and R operators in alternation, transforming major triads into minor and
vice versa.33
Fig. 20b extends the chain of fifths on the right by reflecting F major
across its major-third axis to form D minor. The addition of one more triad
produces a new degree of symmetry which balances the three major triads on
the left with three minor triads on the right. The underlining of the pitches D,
A, E and B on the right reminds us that the triads of this space were
constructed from purely tuned fifths and thirds; thus, D differs from D by a
syntonic comma. As a result, if we were to continue the process of adding pure
triads above and below, each of the interlocking chains of fifths and thirds
would extend into infinity without reaching closure. However, we can easily
impose closure on this musical space through a process known as shearing. By
holding the pitches on the left in place while shifting those on the right
upwards by a syntonic comma, we can align D at the bottom with D at the
top, as shown in Fig. 20c.34 By transforming purely tuned thirds into Pythago-
rean ones in this manner, we can form a ‘circular staircase’ from the vertical
chain of thirds, along which motion can ascend or descend to return to its
starting position.
This modification yields another spatial paradox comparable to one found
in Escher’s drawings, that of figure-ground reversal. As Fig. 20d shows, nestled
within this closed diatonic space are two smaller spaces, each filling in the
interstices of the other. One consists of the primary triads of C major and the
other the primary triads of A minor. Yet because each of these nested spaces
has its own centre-periphery organisation and maps independently onto a
CONTAINER⁄CENTRE-PERIPHERY⁄BALANCE schema, we may flip back and forth
between the two opposing perceptions, just as we do when we look at a Necker
cube or the interlocking pattern of black birds and white birds in Mouch. If we
imagine D to be at the ‘top’, C major will appear at the centre, straddled by
its dominant and subdominant; if we imagine D to be at the ‘bottom’, A minor
will appear at the centre, straddled by its dominant and subdominant.35

74 CANDACE BROWER
Fig. 22 Circle of fifths progression along the diatonic Möbius strip
Having formed a circular pathway from the zigzag chain of thirds, we can
form another from the two chains of fifths. As shown in Fig. 21a, this involves
twisting the double chain of fifths and thirds by 180° while bringing the ends
together to align D at the top with D at the bottom, then inserting an ‘imper-
fect tile’, the B diminished triad, into the space remaining in order to link B
to F via a diminished fifth. The result of this operation, as shown in Figs. 21b
and c, is that one-sided yet two-sided geometric object the Möbius strip.36
With the insertion of the B diminished triad, we can trace a circular route
for harmonic motion along the diatonic chain of fifths which passes through
the three major triads on the front of the Möbius strip to the three minor triads
on the back via the diminished triad straddling the seam, as shown in Fig. 22.
Whereas the ‘two sides’ of the Möbius strip portrayed in Escher’s Möbius Band
II are indistinguishable, the two sides of the diatonic Möbius strip are easily
differentiated owing to the appearance of major triads on the ‘front’ and minor
triads on the ‘back’. Furthermore, whereas in Möbius Band II the dividing line
between ‘front’ and ‘back’ has been erased by the artist, yielding an apparently
seamless one-sided surface, the seam in the diatonic Möbius strip remains
marked by the presence of the diminished fifth, its imperfection serving to
orientate us within the circle of fifths.37

Fig. 23 (a) Construction of triadic pitch space, stage 4, part 1; (b) and (c) modulation
using D, S, L and R operators
We may pursue the process of expansion begun in Figs. 19 and 20 by

continuing to add major and minor triads above and below. As shown in Fig. 23a,
the first few steps of this expansion involve adding D major/B minor above and
B w major/G minor below. Within this expanded space, the key space of C
major/A minor appears as a bounded object dividing triadic pitch space into
two regions, diatonic and chromatic.
Just as we were previously able to transform one triad into another, so we
can now apply D, S, L and R operators to this musical object in order to
transform one key into another, yielding once again the metaphor of an object
in motion. As Figs. 23b and c show, D and S operators produce movement
upwards or downwards by fifth – to the sharp side or to the flat side – while L
and R operators produce movement either upwards or downwards by third and
from ‘right’ to ‘left’ or ‘left’ to ‘right’, transforming major into minor and vice
versa. That each of these operations produces upwards or downwards move-
ment helps to explain why modulation to the sharp side has been associated
with tensing or hardening and that to the flat side, conversely, with relaxing or
softening, these being the feelings we associate with upwards and downwards
movements of the body. Likewise, the association of modulation to the sharp
side with brightening and to the flat side with darkening may be accounted for
in terms of our experience of the brightness of the sky versus the darkness of
the ground when turning our eyes upwards or downwards.38 Finally, the expe-
rience of modulation from major to minor as a shift from ‘bright’ to ‘dark’ or
from ‘happy’ to ‘sad’ may be explained as the result of a lowering of the third
of each primary triad.
Fig. 24 portrays the embodied sensations associated with modulation in
these four directions as the four points of an ‘embodied compass’. The dotted

76 CANDACE BROWER
Fig. 24 The four points of the embodied compass
diagonal line draws attention to the fact that modulation to the sharp side or
to a major key tends to be experienced somewhat similarly as ‘brightening’, and
that to the flat side or to a minor key likewise as ‘darkening.’ This effect is
attributable to the raising or lowering of the key container in the former case
and of the thirds of the primary triads in the latter.39
Fig. 23c also reveals the special nature of the transformation of a major key
into its relative minor and vice versa. Whereas application of D, S and L
operators changes the pitch content of the key, application of the R operator
changes only the mapping of the key container onto the CENTRE-PERIPHERY
schema. Using the R operator to transform the key of C major into that of A
minor causes the pitch D, which formerly appeared at the ‘top’ of the key
space, to now appear at the ‘bottom’. When D occupies the upper position, C
major appears in the centre of the key container, straddled by its major
dominant and subdominant; when D is in the lower position, A minor appears
in the centre, straddled by its minor dominant and subdominant.
The process of adding major and minor triads above and below the diatonic
space of C major/A minor is continued in Fig. 25 to produce a series of twelve
major triads on the left balanced by twelve minor triads on the right. At this
stage, closure can once again be achieved by aligning pitches at the bottom
with those at the top. Whereas in diatonic space we employed shearing to bring
D at the bottom into alignment with D at the top, in chromatic space we use
scaling to bring Gw into alignment with FC, and likewise Ew with DC, thus trans-
forming Pythagorean fifths and thirds into equal-tempered ones.40
As Fig. 25b shows, this chromatic chain of fifths and thirds can be doubled
back on itself to form not a Möbius strip, but a cylindrical loop. Within this
cylindrical loop, the imperfection of the diminished fifth and the ‘twist’ of the
diatonic Möbius strip have disappeared. Thus, repeated application of the D
or S operator will produce movement along either one chain of fifths or the
other. If we follow the pathway on the left we get a series of twelve major triads;
if we follow the pathway on the right, a series of twelve minor triads. Or, by
applying L and R operators in alternation, we encounter all twenty-four major

Fig. 25 Construction of triadic pitch space, stage 4, part 2
and minor triads in succession, yielding the equivalent of what Cohn (1997)
has dubbed the <LR> cycle.41
Within this space we once again confront the paradox of a pathway which
is both vertical, extending infinitely upwards and downwards, and circular,
closing in on itself to return to where it began. As a result, we may ascend from
C major to FC major or descend from C major to Gw major to reach the ‘same’
location. Yet we do not experience modulation to these two keys in the same
way. Rather, we experience movement along the sharp side of the circle of fifths
as ‘tensing’, ‘hardening’ or ‘brightening’ and movement along the flat side as
‘relaxing’, ‘softening’ or ‘darkening’. Thus FC major and Gw major take on
different meanings in the embodied imagination – ‘tense’ versus ‘relaxed’,
‘hard’ versus ‘soft’, ‘bright’ versus ‘dark’ – reflecting their positions at opposite
ends of a vertically oriented space centred around C major.

78 CANDACE BROWER
Fig. 26 Construction of triadic pitch space, stages 5–7
Having filled the musical plane completely in the vertical direction, we can
continue to add triads on both sides of our double chain of fifths, thus expanding
musical space in the ‘horizontal’ direction. We begin by reflecting our prototile
to the left, using the P operator to produce its parallel, C minor. From this
inverted prototile we can construct a series of interlocking major and minor
triads on the left which parallels the one on the right, thus producing a space
which is symmetrical around the fifth axis containing our starting pitch, C, as
shown in Fig. 26a.
This expansion increases the possibilities for modulation by adding the P
operator to those already in place. Using D, S, L, R and P operators we can
transform any major or minor key into any one of its five most closely related
keys: dominant, subdominant, mediant, submediant and parallel.42 Within this
expanded key space we may modulate from any major key to its parallel minor
(or vice versa) by following one of two different routes. As shown in Fig. 26a,
we can move from C major to C minor by following a pathway of descent along
the double chain of fifths and thirds, or we can move directly from C major to

C minor by using the P operator to reflect the key container across the fifth
axis.43 As in the case of enharmonically equivalent keys, these two modulations
are not the same, for although they both end in the key of C minor, each takes
us on a different journey through triadic pitch space. The appearance of these
two regions at different stages of mathematical construction is reflected in the
history of compositional practice as well: composers of the late Baroque
favoured descent along the circle of fifths, while Classical and Romantic
composers increasingly favoured the immediate contrast of mode made possible
by direct modulation.44
As a consequence of the previous expansion, two sets of diagonal pathways
have opened up, one made up of successive major thirds and the other of
successive minor thirds. We may once again expand musical space to a point
of closure by simply extending each diagonal chain of major thirds by one step,
as shown in Fig. 26b. We can then align the pitch at the top of each chain of
major thirds with the one at the bottom to form the set of circular pathways
shown by the arrows, marking off that closed region of triadic pitch space
constituting the hexatonic collection. Within hexatonic space we can move
from one major or minor triad to the next via translation, or we can apply P
and L operators in alternation to produce Cohn’s ‘<LP> cycle’. Or we can
move directly from one hexatonic pole to the other via glide reflection, producing
the maximally smooth voice leading shown in Ex. 3.
By adding one more link to the chain of consecutive minor thirds which
unfolds along the other diagonal, we achieve closure by yet another means, as
shown in Fig. 26c. This forms another set of circular pathways marking off that
region of triadic pitch space which constitutes the octatonic collection. Within
octatonic space we can again move from one major or minor triad to the next
via translation, or we can alternate between P and R operations to realise the
<PR> cycle.
Because hexatonic and octatonic collections constitute closed regions
contained within a larger chromatic space, we may conceive of them as we do
the diatonic collection – as containers or objects which may themselves be set
in motion. As Fig. 27 shows, it is within the space of the neo-Riemannian
Tonnetz that hexatonic and octatonic modulation takes place, the hexatonic
container moving through four possible locations within neo-Riemannian space
(H1–H4) and the octatonic container through three (O1–O3). Each of these
containers can be visualised as a cylindrical band which wraps around the
hexatonic or octatonic torus, respectively, sliding around it three or four times
before returning to its starting position.
Over the course of the expansions of Figs. 26a–c, a fourth, ‘hidden’ axis has
emerged consisting of a series of chromatic semitones … Cw–C–CC ... . We can
fill the musical plane completely by extending this chromatic axis by a full
twelve semitones in order to form the embodied Tonnetz of Fig. 28. Like the
three-by-four Tonnetz of Fig. 27, this twelve-by-twelve Tonnetz can be rolled up
in two directions, as shown by the arrows, to form a torus.45 This final

80 CANDACE BROWER
Fig. 27 Hexatonic and octatonic modulation
Fig. 28 Embodied Tonnetz

Fig. 29 Nested and intersecting spaces within the embodied Tonnetz
transformation produces a space which has neither a fixed centre nor fixed
boundaries, our starting pitch, C, having become the geometric equivalent of
every other. Yet the nested regions of musical space formed over the course of
its construction have not vanished, although their boundaries have become
invisible to the eye. Rather, their existence is made apparent through our
embodied experience of movement within and between these regions. Just as
the kinaesthetic and somatosensory sensations of the body serve to orientate
us within physical space, the bodily sensations associated with changes of
harmony and key allow us to intuit our location within the harmonic space
modelled by the embodied Tonnetz.
Fig. 29 shows that the nested and overlapping spaces which emerged in the
construction of the embodied Tonnetz combine to form a coherent geometric
pattern. At the centre of the figure appears that rhomboidal-shaped region of
triadic pitch space which constitutes the neo-Riemannian Tonnetz. Contained
within it are diatonic, hexatonic and octatonic spaces, each forming one of the
Tonnetz’s main constituent axes. Diatonic space unfolds vertically from D to D,
top to bottom, while hexatonic and octatonic spaces unfold diagonally along

82 CANDACE BROWER
each of its sides, along the major-third and minor-third axes, respectively. The
neo-Riemannian Tonnetz – the natural home of hexatonic and octatonic collec-
tions – overlaps with those regions of the embodied Tonnetz which I have
designated as key space and expanded key space. Together they make up the
natural home of the diatonic collection.
By representing diatonic, hexatonic and octatonic collections as intersecting
regions of a single space, Fig. 29 highlights how differently we experience
movement within them. Diatonic space offers a strong sense of vertical orien-
tation, giving rise to feelings of tension and relaxation as we move up and down
within its space by fifth or by alternating major and minor thirds. Its dual
centre-periphery organisation also makes it possible for us to orient ourselves
within its boundaries with respect to a triadic centre, either C major or A
minor, while its diminished fifth likewise provides orientation by locating us at
the ‘boundaries’ of diatonic space – the ‘seam’ in the Möbius strip. Finally,
when the key container itself is set in motion within key space and expanded
key space, we can attune to feelings of tensing and relaxing, brightening and
darkening, to sense whether we are moving from sharps to flats, or from major
to minor, or vice versa.
Hexatonic and octatonic spaces differ in that they provide few cues to spatial
orientation. Each unfolds diagonally along either the major-third or the minor-
third axis, thus prohibiting root movement upwards or downwards by fifth.
Each has an even rather than an odd number of triads and thus lacks the
centre-periphery organisation of the diatonic collection. And each confronts us
with the paradox of enharmonic equivalence more quickly than movement of
the diatonic container through key space, their smaller circles of major and
minor thirds bringing us from sharp to flat over the space of just an octave.
Furthermore, the maximally smooth voice leading and frequent changes of
inversion associated with movement within hexatonic and octatonic spaces
greatly undermines the effect of being grounded in the root of each triad.46 As
a result, movement within these spaces can seem uncanny, mysterious and
unsettling, as if the laws of nature have been transcended, causing us to lose
our sense of tonal centredness, stability, groundedness and balance.
Like the patterns of tilings which filled Escher’s notebooks, Figs. 19–29
constitute no more than an inventory of possible spaces; it is left to composers
to realise these in the context of particular musical works. The remainder of
this article will examine passages from four pieces in which the paradoxical
features of these spaces are brought to light in musically significant ways: the
fifteenth-century motet ‘Absalon fili mi’;47 the finale of Haydn’s String
Quartet in G major, Op. 76 No. 1; Brahms’s Intermezzo in B minor, Op. 119
No. 1; and Wagner’s Parsifal. These analyses suggest that at least some
composers have been aware of the paradoxes of triadic pitch space and have
intentionally exploited them, allowing listeners to experience sensations like
those associated with Escher’s Mouch, Ascending and Descending and Möbius
Band II.

The Paradox of the Circular Staircase in ‘Absalon fili mi’

We find what may be the earliest example of a circular staircase in the famed
‘Absalon fili mi’ (Ex. 5). Cited by Edward Lowinsky (1946) and others as an
example of what was to become known as musica reservata, this work stands
out as among the most harmonically daring of its time.48 Noted especially for
its remarkable text painting, the closing reference to an infernal descent is
set musically by a displacement into newly chromatic regions of triadic pitch
space, forming a progression which, according to Howard Mayer Brown, ‘not
only demonstrates . . . the outer limits of musical space, but also symbolizes
the idea of a descent into hell in an almost physically palpable way’ (1976,
p. 131). This departure culminates in a shift from major to minor mode,
capturing the meaning of the closing line of text, ‘Let me live no longer, but
descend into hell weeping’. The following analysis realises that descent in visual
form, showing how the journey ends, paradoxically, in a region parallel to the
one in which it began.
Although clearly modal in its overall conception, ‘Absalon fili mi’ exhibits
many features which mark it as tonally progressive, including key signatures of
two to four flats, notated accidentals and numerous V–I cadences.49 From a
compositional perspective, we may view its remarkable harmonic excursion as
a stroke of compositional genius, one which makes the most of the expressive
resources of the modal system. From a historical perspective, we can view its
partial key signature and its modulation far to the flat side of the circle of fifths
in bars 77–83 as a preliminary stage in the formation of the key space shown
in Fig. 25. In this work, triadic harmony appears to have been released from
its diatonic moorings, possibly for the first time in the history of Western
music, carving out through its remarkable trajectory precisely those
regions of triadic pitch space within which movement from one key to another
becomes possible.50
As indicated by its key signature of two to four flats, the motet begins in a
region of triadic space already remarkably low for its time. Yet in its closing
section, heard for the second time in bars 69–85, we are taken on a journey
which extends to new depths of key space. Our final destination is among the
darkest of its regions, the key of Bw minor. The precise nature of the journey is
revealed in Ex. 6 and Fig. 30, which show how three distinct musical ‘objects’
– single pitch, triad and key – follow the same descending pathway.
The journey begins in bar 78, immediately following a V–I cadence in Bw
major which confirms the tonality of the motet as a whole. Over the next six
bars, all four voices join together in a double canon, with the tenor imitated
by the bassus at the fifth and echoed an octave higher by the superius and
altus, thereby outlining a series of descending thirds from F to Gw. Throughout
this canonic passage, melodic motion cascades from one voice to another
along a chain of alternating major and minor thirds. Ex. 6 illustrates the way
in which the melodic descent is carried out by the tenor and bassus in tandem.

84 CANDACE BROWER
Ex. 5 ‘Absalon fili mi’, bars 69–85

Ex. 6 ‘Absalon fili mi’, bars 77–83, tenor and bassus
Fig. 30 ‘Absalon fili mi’, bars 76–85: harmonic and tonal journeys

86 CANDACE BROWER
As the beaming indicates, the tenor voice initiates the descent by falling
through the space of a Bw major triad, a movement continued by the bassus in
bar 79. Movement by falling thirds passes back and forth between the two
voices to outline a succession of major triads, B w–Ew–A w–D w–Gw. Adding to the
overall feeling of descent is a melodic sequence, marked by the higher-level
beams, which forms the pattern F–Ew–D w in the tenor and Bw–A w–Gw in the
bassus.
The succession of major triads ending on Gw in bar 83 might lead us to
expect a cadence in Dw major. But when the voices break out of their sequence,
the Gw, rather than ascending to Aw, descends to F, leading to a V–i cadence in
Bw minor. The lack of a third in the final sonority lends a certain ambiguity of
mode, suggesting that the work may have ended after all in the key of Bw major.
But as Fig. 30 makes clear, the harmonic and tonal motion of the intervening
bars leads to a much lower region of triadic pitch space than the one in which
it began. The feeling of falling, softening and darkening as we descend along
the circle of fifths and shift from major to minor powerfully evokes a grieving
father’s descent into hell. This suggests that the paradoxical nature of triadic
pitch space – that it is open yet closed, vertical yet circular, one-sided yet
two-sided – can be intuited directly via its effect on both body and mind.
The Paradox of Enharmonic Modulation in Haydn’s String Quartet in

G Major, Op. 76 No. 1
It is the uncanny effect of enharmonic modulation that is exploited in the last
movement of Haydn’s String Quartet in G major, Op. 76 No. 1 (Ex. 7). Haydn
is like a master magician showing off his compositional tricks, impressing and
enchanting us with his enharmonic sleight of hand. The modulation in question
appears at the height of the development section, ushered in by a new motive,
marked ‘x’ in the score. This motive serves as a ‘vehicle’ for the modulation,
its identity as an object in motion confirmed by sheer repetition and preservation
of shape.51
Haydn prepares us for his show of enharmonic ingenuity in a number of
ways. Bar 94 brings about an abrupt change of texture from imitative counter-
point to sustained chords, followed over the next two bars by a diminuendo and
a fall in register in all voices. In bar 98 harmonic motion comes almost to a
standstill: Dw minor is sustained for four bars while the subito pianissimo in bar
100 encourages us to listen ever more closely. It is on the downbeat of bar 102
that the composer performs his distinctive magic. Having introduced the
enharmonic pivot chord Dw minor as iv in Aw minor, he causes us to reinterpret
it as CC minor, or iii in the key of A major, by transforming Dw/CC minor into
A major via semitonal ascent from GC (notated as Aw) to AÖ in the first violin.
With this slender melodic thread Haydn leads us out of the gloomy depths of
the key of Aw minor into the brilliant heights of A major, a change of key
confirmed by the ensuing V–I cadence.52 The tonal distances traversed by

Ex. 7 Haydn, String Quartet in G major, Op. 76 No. 1/iv, bars 94–106
harmony and key at this instant – from seven flats to three sharps and from
minor to major – are represented in Fig. 31.53
By contrasting the minuteness of the melodic motion required to carry out
the transformation with the vastness of the key space traversed, Haydn draws
attention to the paradoxical nature of the relationship between melodic and
harmonic motion: that what we experience as near in one dimension may
appear to be quite distant in another. In the nineteenth century composers
increasingly exploited this inverse relation, using chromatic voice leading to
move easily yet mysteriously to seemingly distant tonal regions.
Figure-Ground Reversal and the Möbius Strip in Brahms’s Intermezzo

in B minor, Op. 119 No. 1
That the odd geometry of diatonic pitch space should have escaped theoretical
notice for so long is perhaps not surprising given that one does not normally

88 CANDACE BROWER
Fig. 31 Enharmonic transformation in Haydn, String Quartet in G major, Op. 76

No. 1/iv, bars 101–102
look for strangeness in what seems most familiar. Yet Brahms’s unusual
treatment of the diatonic collection in the opening bars of his Intermezzo in B
minor, Op. 119 No. 1 (Ex. 8), suggests that he was aware of its paradoxical
features and meant to bring them to light. The steady semiquaver rhythm and
continuously falling thirds of the opening present a nearly seamless musical
surface reminiscent of Escher’s Möbius Band II. Its continuous cycling through
the diatonic collection locates us firmly within the key space of D major/B
minor. Yet figure-ground ambiguity causes the ear to flip back and forth
between these two keys, which are counterposed in almost perfect balance.54
In the first three bars, the ambiguity of harmony and metre reinforces that
of key. Bar-length units are marked off by the repeating pattern of the right-
hand melody and the left-hand arpeggiation, as shown by the dotted lines
in Ex. 8. Yet these units resist segmentation into clearly defined beats and

Ex. 8 Brahms, Intermezzo in B minor, Op. 119 No. 1, bars 1–4
harmonies. The division into beats is obscured by the uniform surface rhythm
and the 3/8 metre, which lends itself to alternative groupings of 2 + 2 + 2 or
3 + 3. Similarly, the unbroken succession of diatonic thirds and the sustaining
of every note to the end of the bar serves to blur triadic boundaries such that
each pitch may be heard as the root, third or fifth of the sounding harmony.
Some pitches can be heard as members of more than one harmony, as shown
by the overlapping circles in Ex. 8, a phenomenon which produces a smoothly
shifting, rather than a discretely changing, harmonic landscape. Thus, we are
encouraged by the seamlessness of the musical surface to process bars as
wholes rather than as sums of their constituent parts.
A seamless surface coupled with pattern repetition at the level of the
individual bar creates ideal conditions for the kind of figure-ground reversal
illustrated in Fig. 20d. In bar 1, emphasis on the pitches FC, B and E tilts the
ear towards hearing first i and then iv of B minor. This gives way in bar 2 to
an impression of V7 and then I of D major, an impression reinforced by the
4–3 suspension on the last beat of the bar. This is superseded in bar 3 by
emphasis on the pitches CC, FC and B, suggesting v and then i of B minor. Thus,
as shown in Fig. 32, each of these first three bars presents us with a different
view of the diatonic Möbius strip: first its ‘dark’ side in bar 1, followed by a
turn to its ‘bright’ side in bar 2, then another to its ‘dark’ side in bar 3. Only
in bar 4 do we experience a momentary lifting of the tonal haze as we hear the
clear outlines of a 4–3–2 descent in the melody supported by iiø7–i64 –V 42 in B
minor.
That Brahms intends the tonal ambiguity of the opening bars to come to
the listener’s attention is further suggested by the way it plays out over the
course of the movement as a whole. Its ABA′ form is set in the keys of B minor
and D major, thus reflecting on a formal level the figure-ground reversal of the first
three bars. Yet neither of the first two sections ends with a perfect cadence which
might serve to confirm its tonic.55 In fact, the identity of the tonic key, whether

90 CANDACE BROWER
Fig. 32 Brahms, Intermezzo in B minor, Op. 119 No. 1, bars 1–3
Ex. 9 Brahms, Intermezzo in B minor, Op. 119 No. 1, bars 63–67
D major or B minor, remains open to question until the final cadence. Bars 63
and 64 present the beginnings of a circle-of-fifths progression which seems to
prepare for a V7–I ending in D major, an expectation confirmed by the recall
of material from bar 2 (Ex. 9). Yet at the last moment Brahms undermines our
expectations of this resolution by substituting AC for AÖ (bar 65) while contin-
uing the pattern of falling thirds across the bar line. Above this we hear the G
once again function as a suspension, but in place of the expected 4–3, a 6–5
motion is introduced, all four upper voices sustaining the constituent pitches
of the diminished seventh chord across the bar line before resolving simultane-
ously as part of a V9–i cadence in B minor. In retrospect, the surrounding of
D major by its ‘alter ego’, B minor, in the first three bars foreshadows its

eventual ‘downfall’, expressed in the penultimate bar through the movement of

the falling third, D to B.
An Enharmonic Seam between Diatonic and Hexatonic Space in

Wagner’s Parsifal
The last excerpt to be analysed is taken from Act III of Wagner’s Parsifal, the
same passage discussed at length in Lewin (1984). It was this work perhaps
more than any other which motivated Lewin to investigate the transformational
nature of Riemann’s harmonic theories and to extend them in group-theoretical
ways, thus laying the groundwork for the later development of neo-Riemannian
and transformational theory. In his analysis of Act III, Lewin contrasts the
otherworldly chromaticism and enharmonicism of ‘Riemann space’, which he
associates with the ‘Arabian … world of magic and miracle’, with the everyday
diatonicism of ‘Stufen space’, which he associates with the Grail brotherhood
of Act I (1984, p. 347). According to Lewin, passage between Stufen space and
Riemann space is effected via the ‘hidden seam’ (p. 347) of the enharmonic
equivalence between Cw and B. The transformation of the former into the latter
is marked both dramatically and musically as being among the opera’s most
significant events. Most notably, it accompanies both the kiss which Kundry
gives Parsifal in Act II and Parsifal’s unveiling of the Holy Grail near the end
of the opera, at the point when the chalice begins to shine.
It is in reference to the latter passage (Act III, bars 1097–1099) that we find
Lewin’s intriguing yet elusive reference to a musical Möbius strip. He presents
us with two different versions of the ‘Grail’ motive, first as it appears in its
diatonic version at the end of Act I, where it accompanies the unveiling of the
Holy Grail by Amfortas (Ex. 10a), then as it appears transformed in the
parallel passage near the end of Act III, to accompany the unveiling of the Grail
by Parsifal, armed now with the reclaimed Spear (Ex. 10b). Lewin re-notates
this motive as in shown Ex. 10c to show how its tonal journey might be
represented if carried out in Stufen space, thereby leading to an extraordinarily
distant Ewww at the point of termination.
The analysis shown in Ex. 10c is ultimately deemed ‘untenable’ (p. 346) on
the grounds that it suggests a much greater distance than that which we seem
to have covered over the final three diatonically related harmonies of the pas-
sage. Yet Lewin rejects as equally untenable the two analyses shown in Ex. 10b
since they require us to interpret either the Ew–Cw/B major third or the Cw/B–G
major third as a diminished fourth.56 It is with the intention of resolving this
conflict that he invokes the metaphor of a Möbius strip intersecting with a
cylindrical loop, noting that ‘in some contexts the two spaces [Riemann space
and Stufen space] may coexist locally without apparent conflict; in this way the
surface of a Möbius strip would locally resemble the surface of a cylinder to an
ant who had not fully explored the global logic of the space’ (p. 345; italics as
original). Later he goes on to add that

92 CANDACE BROWER
Ex. 10 Lewin’s analysis of the ‘Grail’ motive, original and transformed
if we imagine pressing together between our fingers a section of a Möbius strip

and a section of a cylindrical loop, then the Ewww chord can be imagined
metaphorically on some part of the loop that diverges from the strip, and the
subdominant function can be imagined on some part of the strip that diverges
from the loop. (p. 346)
We can make sense of Lewin’s foray into the world of Escherian logic by
reinterpreting the passage in light of the model presented here. If we take
‘Stufen space’ to correspond to diatonic space and ‘Riemann space’ to correspond
to hexatonic space, we may proceed to rationalise the passage in question by
mapping it onto the model of triadic pitch space shown in Fig. 29.57 Fig. 33a
shows that the original version of the ‘Grail’ motive fits neatly within the key

Fig. 33 Pathways of harmonic motion in Wagner, Parsifal, Act I, bars 1415–1417,

and Act III, bars 1097–1099
space of Ew major, outlining a purely diatonic progression along the Möbius

strip. It initially descends by third and then by fifth, eventually bringing us full
circle to where we began, and hence prolongs the dominant of Aw major.
Fig. 33b, on the other hand, reveals the more circuitous route taken by the
transformed version of the ‘Grail’ motive. Its descent begins in the same place
– Ew major functioning as V of Aw major – but subsequently diverges, using the
‘hidden seam’ of Cw/B as its point of exit to move along a diagonal axis, first
to Cw minor, its hexatonic polar complement, then back around to the top of
the loop via the enharmonic relation between Cw and B. It then continues by
way of an L transform to G major, which is in turn transformed into its own
polar complement, Ew minor. Having brought us to a point within expanded
key space parallel to that at which we began, the phrase exits the hexatonic
system to complete its journey in diatonic space, following the pattern formed
by the original diatonic version of the motive and descending along the fifth
axis from Ew to Aw to Dw, to end not on Ew, but on Dw, thus accomplishing the
large-scale modulation from I to V to IV.
In his interpretation of the passage, Lewin suggests that this progression
through Riemann space forms an important part of the listener’s experience of

94 CANDACE BROWER
the opera’s magical effects. As we pass through hexatonic space, our sense of
gravitational pull, vertical orientation and centredness is disrupted, giving rise
to a momentary sense of uncanniness, only to be restored, along with our sense
of tonal groundedness, at the final cadence. At the same time, chromatic trans-
formation of the ‘Grail’ motive, accomplished via the enharmonic relationship
between Cw and B, serves to change what was once heard as a dominant into
a subdominant. This in turn prepares us for the large-scale plagal cadence
which closes the entire opera. Lewin’s analysis reveals that the enharmony of
Cw/B, which might otherwise appear to be nothing more than a notational
inconvenience – a flaw in the logic of the diatonic system – takes on powerful
symbolic and affective meaning with respect to the drama as a whole.
Hofstadter points out that it was the presence of paradoxes such as these in
the field of mathematics which inspired Bertrand Russell and Alfred North
Whitehead to attempt to banish them from their great monument to the tenets
of logical empiricism, Principia Mathematica. This treatise, inspired by Russell’s
‘desperate quest for a way to circumvent paradoxes of self-reference’ (Hof-
stadter 1999, p. iv), was an attempt to construct a mathematical system which
would be both complete and free of internal contradiction. The futility of their
quest was revealed only thirty years later by Kurt Gödel’s incompleteness
theorem, which revealed the inevitability of the contradictions that appear in
even the most rigorously derived mathematical systems.
New light has been shed on the paradoxes of mathematics in a recent book
by George Lakoff and Rafael E. Núñez on the bodily basis of mathematical
thought (Lakoff and Núñez 2000). The authors show that mathematical
reasoning is largely spatial in nature, and that many of its basic concepts –
those which form its axiomatic foundations – have image-schematic origins.
Concepts such as equality, line, circle and set appear to originate in cross-
domain mappings of image schemas such as BALANCE, PATH, CYCLE and
CONTAINMENT. Lakoff and Núñez propose that mathematical systems such as
algebra, analytical geometry, number theory and set theory begin with axioms
which are image-schematic in nature, then build through successive layerings
of image-schematic mappings towards increasingly higher levels of complexity
and abstraction. As one metaphorical mapping is overlaid upon another,
contradictions arise within and between these systems, reflecting their different
image-schematic origins. For example, the axiom of Euclidean geometry which
asserts that nonparallel lines must intersect at a single point is contradicted by
axioms governing the geometry of elliptical and hyperbolic spaces, in which
lines are both ‘straight’ and ‘curved’. Thus mathematics, like music, reflects the
capacity of the embodied imagination to produce internally coherent systems
which are logical yet also paradoxical.
In the case of music, these paradoxes appear to be woven into the very fabric
of musical space itself – a residue of its cyclic nature. We may remain oblivious
to the fact that we are moving in circles; we may fail to notice when traversing
the chromatic circle of fifths that we have returned to the point where we

began; and we may be unaware of the twist in the diatonic Möbius strip as we
move from front to back. Because these paradoxes are not only pervasive but
inescapable, a composer must make special efforts to bring them to our
attention. The analyses presented here suggest that at least some composers
have shared with Escher the urge to create works of art which force us to
confront these contradictions.
Music transports us to a world in which the logic of everyday space, time
and movement gives way to impossible possibilities, leading us into contradic-
tion no matter which way we turn – pathways which are vertical yet circular,
locations which are the same yet different, near yet far, above yet below.
Far from constituting flaws in the logic of an otherwise perfect system, such
paradoxes help to make music meaningful for us, a source of never-ending
mystery and delight.
NOTES
M. C. Escher, Mouch, © 2008 The M. C. Escher Company – Holland. All
rights reserved. www.mcescher.com.
1. Octave equivalence appears to be one of the few truly universal principles of pitch
perception, applying not only to humans, but to other primates as well. See
Wright, Rivera, Hulse, Shyan and Neiworth (2000).
2. As reported by Schindler (1966, p. 368).
3. In his late essay ‘Ideas for a “Study on the Imagination of Tone”’, Hugo Riemann
(1914–15) identifies enharmonic equivalence as among the most important
theoretical puzzles remaining to be solved, predicting that ‘the study of enharmonic
identification … ultimately will solve and explain the contradictions between the results
of tone-psychological investigations and the practical experiences of musicians’ (Wason
and Marvin 1992, p. 110; italics as original).
4. As noted by Lakoff and Johnson (1980), the verb ‘to see’ is often used metaphoric-
ally to mean ‘to understand’, reflecting the workings of the conceptual metaphor
SEEING IS UNDERSTANDING. I use quotation marks here and elsewhere to draw
attention to the metaphorical connotations of a term wherever they might
otherwise be overlooked.
5. This article describes the origins of the Tonnetz, or ‘table of tonal relations’, in
the work of Hugo Riemann and other nineteenth-century German theorists and
explains how the Tonnetz came to be resurrected and recast in mathematical terms
largely through the efforts of Lewin (1982 and 1987) and Hyer (1989 and 1995).
6. For a useful introduction to the mathematics of tiling, see Grünbaum and
Shephard (1989).
7. Thurston (1997) notes that this property is shared by all periodic tilings.
8. It is precisely this property which distinguishes the neo-Riemannian Tonnetz from
its nineteenth-century predecessors. The numerical version of the Tonnetz was
first proposed by Hyer (1995).

96 CANDACE BROWER
9. Other artists who have incorporated tilings into their artwork include Koloman
Moser (1868–1918) and Victor Vasarely (1908–97).
10. The contents of Escher’s notebooks, unpublished during his lifetime, appear in
Schattschneider (1990).
11. This and some of Escher’s other artworks can be viewed online at http://
www.mcescher.net/ (accessed 6 June 2008).
12. For a simple explanation of local versus global properties, see Weeks (2002).
13. In this article Cohn offers examples of hexatonic polar progressions spanning
more than 300 years, from Carlo Gesualdo’s ‘Moro lasso’ to Schoenberg’s String
Trio, Op. 45.
14. The alternative layout of fifths along diagonal axes and major and minor thirds along
vertical and horizontal axes shown in Fig. 7 was introduced by Balzano (1980) and
is favoured by Cohn (1997). The geographic labelling of the four hexatonic systems
as Northern, Southern, Eastern and Western was introduced by Cohn (1996).
15. The term ‘de-souling’ is Cohn’s (2004, p. 297) rendering of the term Entseelung,
which translates literally as ‘removal of the soul’.
16. The concept of an image schema, first introduced by the linguist George Lakoff
and the philosopher Mark Johnson (Lakoff 1987; Johnson 1987; and Lakoff and
Johnson 1999), has since given birth to a thriving sub-discipline of cognitive
science which has engaged scholars in such fields as linguistics (Langacker 1987
and Talmy 2000), psychology (Gibbs and Colston 1995), literature (Turner 1991
and 1996), anthropology (Strauss and Quinn 1997), mathematics (Lakoff and
Núñez 2000), visual art (Esrock 2002) and music (Saslaw 1996 and 1997–8;
Mead 1997–8; Brower 1997–8 and 2000; Zbikowski 1997 and 2002; Larson
1997–8 and 2004; Cox 1999; Snyder 2000; Kemler 2001; Butterfield 2002; Iyer
2002; Eitan and Granot 2004; and Straus 2006). The following discussion is
particularly indebted to Johnson (1987).
17. Hyer (1995) notes the contradiction implied by the visual layout of Riemann’s original
Tonnetz: that whereas we descend in register when moving downwards along the major-
third axis, we ascend in register when moving downwards along the minor-third axis.
18. As Butterfield (2002) observes, the concept of a musical object is itself metaphorical.
He suggests that we identify as musical ‘objects’ those features of a musical work
– including the work itself – which share properties with objects in the physical world,
such as replicability, durability, containment and preservation of shape and size.
19. See especially Ernst Terhardt’s work on the perception of virtual pitch, a term
used by Terhardt to refer to our perception of the fundamental frequency of a
complex tone as its sounding pitch even when it is not actually sounded (Terhardt
1974 and 1979). The implications of this work have been extended to the percep-
tion of combinations of tones (Terhardt 1984; and Parncutt 1988 and 1989).
According to Terhardt (1984, p. 293), ‘the nature of the fundamental note (root) of
musical chords is identical with that of the virtual pitch of individual complex tones’.
20. The pitches shown in Figs. 15 and 16 constitute a type of pitch class because they
subsume those pitches that share the same letter name, regardless of register.
However, they differ from the pitch classes of atonal set theory in that they

exclude their enharmonic equivalents. Throughout this article, the use of a letter
name alone should be taken to imply this more limited notion of what we might
call letter-name pitch class.
21. The metaphor of root as ground also helps to explain why major and minor triads
sound more stable in root position than in first or second inversion. When the
root appears in the bass, it functions as the fundamental – as a stable foundation
for the sonority as a whole.
22. Although Western musicians tend to take for granted the correlation between
increasing frequency and ascending movement, cross-cultural studies suggest that
this mapping may not be universal (Zbikowski 2002 and Ashley 2004). For exam-
ple, Balinese and Javanese musicians refer to ‘high’ and ‘low’ pitches as ‘small’
and ‘large’, while the Suya of the Amazon basin refer to them as ‘young’ and ‘old’
(Zbikowski 2002). That changes of pitch tend to evoke feelings of rising and
falling movement in Western listeners may reflect the pervasively triadic nature of
Western tonal music, which supports the mapping of root (or tonic) as ground
and of melodic motion as rising and falling in relation to this ground.
23. Révész (1954) and Shepard (1964).
24. These are described in Shepard (1964).
25. For an auditory demonstration of Shepard’s tones, see http://www.cs.ubc.ca/nest/
imager/contributions/flinn/Illusions/ST/st.html (accessed 6 June 2008). A similar
effect can be experienced when one listens to certain organ works played on
mixture stops. For example, the bass line of Bach’s Fantasia in G minor, BWV
542, includes a sequential series of stepwise descents (bars 31–34) which seem to
create the illusion of a continuous descent in this respect.
26. Construction of the model in stages makes it possible to address certain
problematic features of the neo-Riemannian model noted by Lerdahl (2001):
first, that it appears to privilege root relations by third over those by fifth despite
the predominance of the latter even in the highly chromatic music which inspired
development of the model; and second, that it appears to posit the existence of
the chromatic collection before the diatonic even though the latter is both
syntactically central and historically prior to the former.
27. A particularly clear example of this inverse relation is to be found in Ex. 3.
28. Although the embodied features of the present model have yet to be empirically
tested, its representation of relative distances among chords and keys, similar to those
offered by Lerdahl (2001), fits well with the results of probe-tone studies (Krumhansl
1990 and 1998). They also correlate effectively with the results of a recent study
showing that key distances are represented in the form of a dynamic topography
in the brain ( Janata, Birk, Van Horn, Leman, Tillmann and Bharucha 2002).
29. Riemann (1914–15) also stresses the role played by the imagination in transform-
ing musical sounds. He notes that ‘the “Alpha and Omega” of musical artistry …
exists in the mental image of musical relationships that occurs in the creative
artist’s imagination – a mental image that lives before it is transformed into nota-
tion and re-emerges in the imagination of the hearer’ (Wason and Marvin 1992,
p. 82; italics as original). Later he observes that such images are at least partly
somatosensory in nature: ‘As a result of the valuation of pitch-level motion as

98 CANDACE BROWER
alternating rising and falling … a psychic experience emerges comparable to an

ascent into the lighter regions and a descent into the darker, like the flight of the
bird in the air or the swimming of the fish in water – not as something viewed,
but as something actually experienced’ (p. 96).
30. That a succession of triads can be understood metaphorically as a single object
in motion reflects a key feature of our experience of the movement of physical
objects, namely, preservation of shape. Psychologists have shown that shape
preservation plays an important role in our perception of the apparent motion of
visual ‘objects’. Ternus (1938) found that observers who were shown a group of dots
in which one dot moved to a different location while the other two stayed in place saw
all of the dots as moving together so long as the shape of the whole was preserved.
31. Neo-Riemannian theorists generally treat S as simply the inverse of D (D – 1). Here,
I treat S as its own operator to reflect the difference in our embodied experience
of upwards and downwards movement. I also restore Riemann’s original con-
notations for D and S; thus, the D operator transforms any triad into its
dominant, while the S operator transforms the same triad into its subdominant.
32. As Steblin (2002) notes, such descriptors, applied to keys as well as triads, have
a history going back at least to Zarlino. It appears that the association of major
keys with happiness and minor keys with sadness may be a response not only to
the lowering of the third of the primary triads, but also to our tendency to map
melodic motion onto bodily movements carried out under the influence of tonal
and gravitational forces, which appear to pull more strongly downwards in minor
than in major keys (Brower 2000 and Arnheim 1984). That the association of
major with happy and minor with sad has a bodily explanation and is not merely
the product of enculturation is suggested by empirical studies showing that these
associations are made even by very young children (Kastner and Crowder 1990)
and brain-damaged adults (Peretz, Gagnon and Bouchard 1998).
33. In applying L and R operators to the triangular shapes of Fig. 20, we are once
again reminded that musical objects differ geometrically from the visual ‘objects’
used to portray them. When the operation of reflection is used to transform the
triangles of Fig. 20 into their adjacent neighbours, they require not one but two
reflections, the second perpendicular to the first, owing to the extra horizontal
dimension of the visual model. Because pitch space lacks this dimension, the axis
of reflection actually constitutes a point, not a line, in pitch space – namely, that
point which lies halfway between the pitches to be preserved as common notes.
34. Shearing is a term used by physicists and graphic artists alike to refer to the
distortion which results from sliding two parts of an object in a direction parallel
to their plane of contact.
35. These opposing perceptions are not equally likely, however, because the greater
stability of the major triad biases us towards perception of the major mode. The
Necker cube induces a similar bias: it is more frequently seen as if viewed from
the top than from the bottom. The former appears more stable than the latter
because it conforms more closely to our everyday view of cube-shaped objects. It
seems likely that the raising of the leading note in the minor mode came about
in part to compensate for this perceptual bias.
36. That diatonic space takes the shape of a Möbius strip when arranged as a chain
of interlocking fifths and thirds was first noted by Mazzola (1990) and Mazzola,

Göller and Müller (2002). Interestingly, the Möbius strip was discovered by a
procedure similar to that shown in Fig. 21. In experimenting with different ways
of forming closed loops from rectangular strips created by gluing triangles
together, A. F. Möbius discovered that any strip containing an even number of
triangles forms a cylindrical loop, while any strip containing an odd number of
triangles must be twisted by 180° to make the ends meet, forming what is now
known as the Möbius strip (Biggs 1993).
37. A precursor of the model shown in Figs. 19–21 appears in Hauptmann ([1893]
1991; first published in 1853 as Die Natur der Harmonik und der Metrik). Haupt-
mann portrays the diatonic pitches of C major as a chain of thirds, F–a–C–e–G–
b–D (where the capital and lowercase letters refer to the roots of major and of
minor triads respectively), made up of a ‘triad of triads’ (p. 9), I, IV and V, which
he describes as closing in on itself to form a circle. Because Hauptmann’s model
is one-dimensional, however, it does not show the ‘twist’ in the chain of interlock-
ing fifths and thirds. Furthermore, because Hauptmann, like many of his contem-
poraries, favoured just intonation, the third D–F is a syntonic comma smaller
than pure, thus producing not one but two ‘diminished’ triads. It appears that,
like many other German theorists of the nineteenth century, Hauptmann viewed
triadic pitch space as extending indefinitely towards increasingly chromatic
regions from the centrally located key of C major (Engebretsen 2002).
38. Although one finds varying descriptions of the qualities of individual keys in the
eighteenth and nineteenth centuries, there appears to have been a strong
consensus concerning the association of modulation to the sharp side of the circle
of fifths with tensing, hardening and brightening, and to the flat side with relax-
ing, softening and darkening (Steblin 2002). For example, according to Rameau
(1754), ‘the side of the dominant, that of the rising fifth, is rightly the side of
strength, so that the more fifths there are in going up, the more this strength
increases; the same reasoning holds conversely for softness, on the side of the
subdominant’ (quoted in Steblin 2002, p. 97). And according to Vogler (1778),
‘if we go up by fifths through G, D, A, and E, there is always an increase of
strength, effect, cutting quality and penetration. If we go down by fifths through
F, Bw, Ew, and Aw, all strength is reduced and the impression becomes duller and
darker’ (quoted in Steblin 2002, p. 121).
39. Steblin (2002) notes that some theoretical confusion arose in the eighteenth
century concerning the distinction between major mode and sharp keys on the
one hand and minor mode and flat keys on the other because movement from
sharp to flat keys and modulation from major to minor mode both require the
addition of flats and, likewise, movement from flat to sharp keys and modulation
from minor to major mode both require the addition of sharps. It seems likely
that this confusion was caused in part by the similar experiential qualities of these
changes in key and mode.
40. Scaling, a term used by graphic artists to refer to a uniform change in the size of
an object, here refers to a reduction in the size of each perfect fifth by one-twelfth
of a Pythagorean comma.
41. An early representation of the <LR> cycle appears in Vogler (1778). Vogler
portrays the cycle in circular score notation, the falling thirds in the bass being
filled in with passing notes in order to allow each triad to be approached by its
dominant. See Wason (1985), p. 17.

100 CANDACE BROWER
42. The results of a probe-tone study suggest that the modulations produced by
applying these five operators do in fact correspond to those keys heard as most
closely related (Krumhansl 1998).
43. Within this space, we can also chart a circular route around our starting pitch of
C by applying P, L and R operators in succession to produce what Cohn has
termed the <PLR> cycle. The resultant hexagon-shaped musical object, which binds
together all of the triads containing the pitch C, appears to be what Riemann
meant by Klangvertretung, translated by Wason and Marvin as ‘representation
of a tonal complex’ (1992, p. 86). According to Wason and Marvin, Riemann
used this term to refer to the psychological processing of a note in terms of its
six possible harmonic functions as root, third or fifth of a major or minor triad.
44. We find evidence of such preferences in the writings of composers and theorists
of the time. Georg Andreas Sorge, for example, notes the possibility of direct
modulation in his Vorgemach der musicalischen Composition (1745–7), only to reject
it as too abrupt, comparing the sounding of C major and C minor together to ‘a
knock on the head’ (p. 30).
45. Technically speaking, every horizontal row of pitches should consist of successive
sharpings and flattings of the same pitch, CÖ becoming Cwwwwww on the left and
CCCCCCC on the right. This permits every pitch in the grid to have its own letter
designation. For practical reasons, I have substituted the pitches of the chromatic
scale, thus yielding enharmonic spellings of triads which become more obvious as
we move away from the centre of the grid.
46. Cohn (1996) proposes treating the pitches of a triad within a hexatonic system as
equally weighted, regardless of their status as root, third or fifth. His proposal
underlines the degree to which the maximally smooth voice leading associated with
hexatonic progressions undermines the privileged position of the root within the triad.
47. We find evidence of such intentions in the score of Bach’s ‘Canon per tonos’ from
Das musikalische Opfer, a work which sequences upwards by major second six times
in order to conclude in the key in which it began. That Bach meant the paradox
of a circular staircase to come to the attention of its royal dedicatee, Frederick
the Great, can be inferred from his inscription: ‘As the modulation rises, so may
the King’s glory’. But although Hofstadter (1999, p. 10) describes the work as an
‘Endlessly Rising Canon’ and indicates by this that Bach intended the canon to
portray the king’s glory as ascending infinitely, Chafe (1984) argues that Bach
most likely intended just the opposite: not infinite ascent but merely the illusion of
ascent. Thus the dedication in fact reveals the earthly limitations of the king’s glory.
48. See Lowinsky (1946). Although this motet was long attributed to Josquin Desprez,
its authorship is now questioned, with many scholars arguing, largely on stylistic
grounds, that it was composed by Pierre de la Rue (Benthem 1989; Rifkin 1991; and
Meconi 1998). However, Davison (1996) maintains that the work cannot be
definitively assigned to either composer and that its authorship must remain uncertain.
49. Randel (1971) notes the ubiquity of the V–I cadence in fifteenth-century music
and refers to it as an example of ‘emergent triadic tonality’.
50. It is perhaps not surprising that the expansion of triadic space should have
proceeded far to the flat side before moving in the opposite direction. The
medieval gamut offered the ‘soft’ (flat) B as an alternative to the ‘hard’ (natural)

B, providing an extra link in the chain of perfect fifths. It seems only natural that
composers would have viewed the Bw at the bottom of the chain as the more
logical point for expansion. The closing bars of ‘Absalon fili mi’ reveal an inter-
mediate stage in the formation of key space, one which extends the chain of major
and minor thirds by seven steps, thus permitting a return to the initial tonic but
not to the initial mode.
51. That Haydn conceived of this motive as a ‘vehicle’ for the modulation is further
suggested by its only other appearance in the movement: in bar 143 of the
recapitulation, where it forms a prominent feature of the new theme which ushers
in an equally striking change of key from G minor to G major.
52. The remarkable effect of enharmonic modulation is noted by Anton Reicha in his
Traité de mélodie (1814); he observes that ‘when the key of FC is suddenly changed
into Gw, and CC into Dw (and vice versa), we fall from a very brilliant key into a
very sombre key, or from a very sombre into a very piercing key’ (quoted in
Steblin 2002, p. 126).
53. One could hear this modulation as simply a continuation of the descent along the
circle of fifths to the key of the Neapolitan, Bww major. However, assuming that
one has kept track of one’s position within the key space, this only delays the
arrival of the enharmonic modulation needed to return to the home key of G
minor. Given Haydn’s unusual treatment of this passage, his notation of the
modulation at precisely this point, and the fact that the key of Bww major falls far
outside the range of keys that in the late eighteenth century were judged accept-
able, it seems likely that the composer intended the modulation to be heard as it
is notated. Further support for this hearing is provided by the upwards resolution
of the Aw, strongly suggesting its reinterpretation as GC.
54. Many other ambiguous features of this piece have been pointed out by Dunsby
(1981).
55. Jordan and Kafalenos (1989) explore the tonal ambiguity of this composition in
terms of a ‘double trajectory’ of B minor and D major, showing how Brahms
maintains the ambiguity throughout.
56. The BÖ shown in parentheses in Ex. 10b reflects Wagner’s own notation, while the
Cw reflects our initial hearing of this triad as a flattened submediant in the key of
Ew major.
57. From the passages cited above, it would appear that Lewin conceived of Riemann
space as the Möbius strip and Stufen space as the cylindrical loop, precisely the
opposite of the interpretation offered here. This suggests that he had not fully
worked out the implications of his topological metaphor but was attempting
rather to capture the idea of two musical spaces that intersect, yet which remain
topologically distinct.
REFERENCES
Arnheim, Rudolf, 1984: ‘Perceptual Dynamics in Musical Expression’, Musical
Quarterly, 70/iii, pp. 295–309.
____, 1997: Art and Visual Perception: a Psychology of the Creative Eye, rev. edn
(Berkeley and Los Angeles, CA: University of California Press).

102 CANDACE BROWER
Ashley, Richard, 2004: ‘Musical Pitch Space across Modalities: Spatial and
Other Mappings through Language and Culture’, in S. D. Lipscomb,
R. Ashley, R. O. Gjerdingen and P. Webster (eds.), Proceedings of the 8th
International Conference on Music Perception & Cognition, Evanston, IL, 2004
(Adelaide: Causal Productions), pp. 64–71.
Balzano, Gerald, 1980: ‘The Group-Theoretic Description of 12-Fold and
Microtonal Pitch Systems’, Computer Music Journal, 4/iv, pp. 66–84.
Benthem, Jaap van, 1989: ‘Lazarus versus Absalom. About Fiction and Fact in
the Netherlands Motet. For Willem Elders’, Tijdschrift van de Koninklijke
Vereniging voor Nederlandse Muziekgeschiedenis, 39, pp. 54–82.
Biggs, Norman, 1993: ‘The Development of Topology’, in John Fauvel, Raymond
Flood and Robin Wilson (eds.), Möbius and His Band: Mathematics and
Astronomy in Nineteenth-Century Germany (Oxford and New York: Oxford
University Press), pp. 104–19.
Brower, Candace, 1997–8: ‘Pathway, Blockage, and Containment in Density
21.5’, Theory and Practice, 22–3, pp. 35–54.
____, 2000: ‘A Cognitive Theory of Musical Meaning’, Journal of Music Theory,
44/ii, pp. 323–79.
Brown, Howard Mayer, 1976: Music in the Renaissance (Englewood Cliffs, NJ:
Prentice Hall).
Butterfield, Matthew, 2002: ‘The Musical Object Revisited’, Music Analysis,
21/iii, pp. 327–80.
Chafe, Eric, 1984: ‘Allegorical Music: the “Symbolism” of Tonal Language in
the Bach Canons’, Journal of Musicology, 3/iv, pp. 340–62.
Cohn, Richard, 1996: ‘Maximally Smooth Cycles, Hexatonic Systems, and the
Analysis of Late-Romantic Triadic Progressions’, Music Analysis, 15/i, pp.
9–40.
____, 1997: ‘Neo-Riemannian Operations, Parsimonious Trichords, and Their
Tonnetz Representations’, Journal of Music Theory, 41/i, pp. 1–66.
____, 1998: ‘Introduction to Neo-Riemannian Theory: a Survey and a Historical
Perspective’, Journal of Music Theory, 42/ii, pp. 167–80.
____, 2004: ‘Uncanny Resemblances: Tonal Signification in the Freudian Age’,
Journal of the American Musicological Society, 57/ii, pp. 285–323.
Cox, Arnie W., 1999: ‘The Metaphoric Logic of Musical Motion and Space’
(PhD diss., University of Oregon).
Davison, Nigel, 1996: ‘Absalom fili mi Reconsidered’, Tijdschrift van de Koninkl-
ijke Vereniging voor Nederlandse Muziekgeschiedenis, 46/i, pp. 42–56.
Deutsch, Diana and Feroe, John, 1981: ‘The Internal Representation of Pitch
Sequences in Tonal Music’, Psychological Review, 88, pp. 503–22.
Dunsby, Jonathan, 1981: Structural Ambiguity in Brahms: Analytical Approaches
to Four Works (Ann Arbor, MI: UMI Research Press).
Eitan, Zohar and Granot, Roni, 2004: ‘Musical Parameters and Images of
Motion’, in Richard Parncutt, Annekatrin Kessler and Fränk Zimmer (eds.),
Proceedings of the Conference on Interdisciplinary Musicology, Graz/Austria,

15–18 April 2004, http://www-gewi.uni-graz.at/staff/parncutt/cim04/

CIM04_paper_pdf/Eitan_Granot_CIM04_proceedings.pdf (accessed 9
June 2008).
Engebretsen, Nora, 2002: ‘The Chaos of Possibilities: Combinatorial Group
Theory in Nineteenth-Century German Harmony Treatises’ (PhD diss.,
State University of New York at Buffalo).
Esrock, Ellen, 2002: ‘Touching Art: Intimacy, Embodiment, and the Somato-
sensory System’, Consciousness and Emotion, 2/ii, pp. 233–53.
Gibbs, Raymond W. and Colston, Herbert L., 1995: ‘The Cognitive Psycho-
logical Reality of Image Schemas and Their Transformations’, Cognitive
Linguistics, 6/iv, pp. 347–78.
Grünbaum, Branko and Shephard, Geoffrey C., 1989: Tilings and Patterns: an
Introduction (New York: W. H. Freeman).
Hauptmann, Moritz, [1893] 1991: The Nature of Harmony and Metre, trans.
and ed. W. E. Heathcote, 2nd edn (New York: Da Capo Press).
Hofstadter, Douglas R., 1999: Gödel, Escher, Bach: an Eternal Golden Braid,
2nd edn (New York: Basic Books).
Hyer, Brian, 1989: ‘Tonal Intuitions in Tristan und Isolde’ (PhD diss., Yale
University).
____ , 1995: ‘Reimag(in)ing Riemann’, Journal of Music Theory, 39/i, pp. 101–38.
Iyer, Vijay, 2002: ‘Embodied Mind, Situated Cognition, and Expressive
Microtiming in African-American Music’, Music Perception, 19/iii, pp. 387–414.
Janata, Petr; Birk, Jeffrey L.; Van Horn, John D.; Leman, Marc; Tillmann,
Barbara; and Bharucha, Jamshed J., 2002: ‘The Cortical Topography of
Tonal Structures Underlying Western Music’, Science, 298, pp. 2167–70.
Johnson, Mark, 1987: The Body in the Mind: the Bodily Basis of Meaning,
Imagination, and Reason (Chicago: University of Chicago Press).
Jordan, Roland and Kafalenos, Emma, 1989: ‘The Double Trajectory: Ambiguity
in Brahms and Henry James’, 19th-Century Music, 13/ii, pp. 129–44.
Kastner, Marianna and Crowder, Robert, 1990: ‘Perception of the Major/
Minor Distinction: IV. Emotional Connotations in Young Children’, Music
Perception, 8/ii, pp. 189–202.
Kemler, Deanna, 2001: ‘Music and Embodied Imagining: Metaphor and
Metonymy in Western Art Music’ (PhD diss., University of Pennsylvania).
Kepler, Johannes, 1619: Harmonice Mundi (Linz).
Krumhansl, Carol, 1990: Cognitive Foundations of Musical Pitch (New York:
Oxford University Press).
Krumhansl, Carol L., 1998: ‘Perceived Triad Distance: Evidence Supporting
the Psychological Reality of Neo-Riemannian Transformations’, Journal of
Music Theory, 42/ii, pp. 265–81.
Lakoff, George, 1987: Women, Fire, and Dangerous Things: What Categories
Reveal about the Mind (Chicago: University of Chicago Press).
Lakoff, George and Johnson, Mark, 1980: Metaphors We Live By (Chicago:
University of Chicago Press).

104 CANDACE BROWER
____, 1999: Philosophy in the Flesh: the Embodied Mind and Its Challenge to
Western Thought (New York: Basic Books).
Lakoff, George and Núñez, Rafael E., 2000: Where Mathematics Comes From:
How the Embodied Mind Brings Mathematics into Being (New York: Basic
Books).
Langacker, Ronald, 1987: Foundations of Cognitive Grammar, vol. 1: Theoretical
Prerequisites (Stanford, CA: Stanford University Press).
Larson, Steve, 1997–8: ‘Musical Forces and Melodic Patterns’, Theory and
Practice, 22–3, pp. 55–71.
____, 2004: ‘Musical Forces and Melodic Expectations: Comparing Com-
puter Models and Experimental Results’, Music Perception, 21/iv, pp.
457–98.
Lerdahl, Fred, 2001: Tonal Pitch Space (New York: Oxford University Press).
Lewin, David, 1982: ‘A Formal Theory of Generalized Tonal Functions’,
Journal of Music Theory, 26/i, pp. 23–60.
____, 1984: ‘Amfortas’s Prayer to Titurel and the Role of D in Parsifal: the
Tonal Spaces of the Drama and the Enharmonic Cw/B’, 19th-Century Music,
7/iii, pp. 336–49.
____, 1987: Generalized Musical Intervals and Transformations (New Haven, CT
and London: Yale University Press).
Lowinsky, Edward E., 1946: Secret Chromatic Art in the Netherlands Motet,
trans. Carl Buchman (New York: Columbia University Press).
Mazzola, Guerino, 1990: Geometrie der Töne, Elemente der mathematischen
Musiktheorie (Basel: Birkhaüser).
Mazzola, Guerino, in collaboration with Göller, Stefan and Müller, Stefan, 2002:
The Topos of Music: Geometric Logic of Concepts, Theory, and Performance
(Basel: Birkhäuser Verlag).
Mead, Andrew, 1997–8: ‘Shedding Scales: Understanding Intervals in Different
Musical Contexts’, Theory and Practice, 22–3, pp. 73–94.
Meconi, Honey, 1998: ‘Another Look at Absalon’, Tijdschrift van de Koninklijke
Vereniging voor Nederlandse Muziekgeschiedenis, 48/i, pp. 3–29.
Parncutt, Richard, 1988: ‘Revision of Terhardt’s Psychoacoustical Model of the
Root(s) of a Musical Chord’, Music Perception, 6/i, pp. 65–94.
____, 1989: Harmony: a Psychoacoustical Approach (Berlin: Springer-Verlag).
Peretz, Isabelle; Gagnon, Lise; and Bouchard, Bernard, 1998: ‘Music and
Emotion: Perceptual Determinants, Immediacy, and Isolation after Brain
Damage’, Cognition, 68/ii, pp. 111–41.
Rameau, Jean-Philippe, 1754: Observations sur notre instinct pour la musique
(Paris: Prault fils).
Randel, Don M., 1971: ‘Emerging Triadic Tonality in the Fifteenth Century’,
Musical Quarterly, 57/i, pp. 73–86.
Reger, Max, 1948: On the Theory of Modulation, trans. John Bernhoff (New
York: Edwin F. Kalmus).
Reicha, Anton, 1814: Traité de mélodie (Paris).

Révész, Géza, 1954: Introduction to the Psychology of Music, trans. G. I. C. de

Courcy (Norman, OK: University of Oklahoma Press).
Riemann, Hugo, 1914–15: ‘Ideen zu einer “Lehre von den Tonvorstellungen”’,
Jahrbuch der Musikbibliothek Peters, 21–2, pp. 1–26.
Rifkin, Joshua, 1991: ‘Problems of Authorship in Josquin: Some Impolitic
Observations, with a Postscript on Absalon, fili mi’, in Proceedings of the
International Josquin Symposium (Utrecht: Vereniging voor Nederlandse
Muziekgeschiedenis), pp. 45–52.
Saslaw, Janna, 1996: ‘Forces, Containers, and Paths: the Role of Body-Derived
Image Schemas in the Conceptualization of Music’, Journal of Music Theory,
40/ii, pp. 217–43.
____, 1997–8: ‘Life Forces: Conceptual Structures in Schenker’s Free Composition
and Schoenberg’s The Musical Idea’, Theory and Practice 22–3, pp. 17–33.
Schattschneider, Doris, 1990: Visions of Symmetry: Notebooks, Periodic Drawings,
and Related Works of M. C. Escher (New York: W. H. Freeman).
Schindler, Anton Felix, 1966: Beethoven as I Knew Him: a Biography, ed.
Donald W. MacArdle, trans. Constance S. Jolly (Chapel Hill, NC: University
of North Carolina Press).
Shepard, Roger N., 1964: ‘Circularity in Judgments of Relative Pitch’, Journal
of the Acoustical Society of America, 36/xii, pp. 2346–53.
Snyder, Bob, 2000: Music and Memory: an Introduction (Cambridge, MA: MIT
Press).
Sorge, Georg Andreas, 1745–7: Vorgemach der musicalischen Composition
(Lobenstein).
Steblin, Rita, 2002: A History of Key Characteristics in the Eighteenth and Early
Nineteenth Centuries, 2nd edn (Rochester, NY: University of Rochester
Press).
Straus, Joseph N., 2006: ‘Normalizing the Abnormal: Disability in Music and
Music Theory’, Journal of the American Musicological Society, 59/i, pp. 113–
84.
Strauss, Claudia and Quinn, Naomi, 1997: A Cognitive Theory of Cultural
Meaning (Cambridge: Cambridge University Press).
Talmy, Leonard, 2000: Toward a Cognitive Semantics, 2 vols. (Cambridge, MA:
MIT Press).
Terhardt, Ernst, 1974: ‘Pitch, Consonance, and Harmony’, Journal of the
Acoustical Society of America, 55/v, pp. 1061–9.
____, 1979: ‘Calculating Virtual Pitch’, Hearing Research, 1, pp. 155–82.
____, 1984: ‘The Concept of Musical Consonance: a Link between Music and
Psychoacoustics’, Music Perception, 1/iii, pp. 276–95.
Ternus, Joseph, 1938: ‘The Problem of Phenomenal Identity’, in Willis D. Ellis
(ed.), A Source Book of Gestalt Psychology (New York: Harcourt and Brace),
pp. 149–60.
Thurston, William P., 1997: Three-Dimensional Geometry and Topology, vol. 1,
ed. Silvio Levy (Princeton, NJ: Princeton University Press).

106 CANDACE BROWER
Turner, Mark, 1991: Reading Minds: the Study of English in the Age of Cognitive
Science (Princeton, NJ: Princeton University Press).
____, 1996: The Literary Mind: the Origins of Thought and Language (Oxford:
Oxford University Press).
Vogler, Georg Joseph, 1778: Kuhrpfälzische Tonschule (Mannheim: C. F. Schwan
and M. Götz).
Wason, Robert W., 1985: Viennese Harmonic Theory from Albrechtsberger to
Schenker and Schoenberg (Rochester, NY: University of Rochester
Press).
Wason, Robert and Marvin, Elizabeth West, 1992: ‘Riemann’s “Ideen zu einer
Lehre von den Tonvorstellungen”: an Annotated Translation’, Journal of
Music Theory, 36/i, pp. 69–117.
Weeks, Jeffrey, 2002: The Shape of Space (New York: Marcel Dekker).
Wright, Anthony A.; Rivera, Jacquelyne J.; Hulse, Stewart H.; Shyan, Melissa;
and Neiworth, Julie J., 2000: ‘Music Perception and Octave Generalization
in Rhesus Monkeys’, Journal of Experimental Psychology: General, 129/iii,
pp. 291–307.
Zbikowski, Lawrence M., 1997: ‘Conceptual Models and Cross-Domain
Mapping: New Perspectives on Theories of Music and Hierarchy’, Journal
of Music Theory, 41/ii, pp. 193–225.
____, 2002: Conceptualizing Music: Cognitive Structure, Theory, and Analysis
(New York and Oxford: Oxford University Press).
ABSTRACT
Parallels between the mathematics of tiling, which describes geometries of visual
space, and neo-Riemannian theory, which describes geometries of musical space,
make it possible to show that certain paradoxes featured in the visual artworks
of M. C. Escher also appear in the pitch space modelled by the neo-Riemannian
Tonnetz. This article makes these paradoxes visually apparent by constructing
an embodied model of triadic pitch space in accordance with principles drawn
from the mathematics of tiling, on the one hand, and from cognitive science,
on the other – specifically, the notion that our experience of pitch relationships
is governed in part by the metaphorical projection of patterns abstracted from
embodied experience known as image schemas. These paradoxes are illustrated
with reference to passages drawn from four compositions to whose expressive
character such paradoxes contribute: the fifteenth-century motet ‘Absalon fili
mi’; the finale of Haydn’s String Quartet in G major, Op. 76 No. 1; Brahms’s
Intermezzo in B minor, Op. 119 No. 1; and Wagner’s Parsifal.


Candace Paradoxes Pitch

Uploaded by

Copyright:

Available Formats

Candace Paradoxes Pitch

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Candace Paradoxes Pitch

Uploaded by

Copyright:

Available Formats

DOI: 10.1111/j.1468-2249.2008.00268.

PARADOXES OF PITCH SPACE

Music Analysis, 27/i (2008) 51

Ex. 1 From Max Reger, On the Theory of Modulation (1948), p. 11

similar paradoxes, which he refers to as ‘strange loops’. Hofstadter makes

Mathematical Models of Musical and Visual Space

© 2008 The Author. Music Analysis, 27/i (2008)

Fig. 1 Neo-Riemannian Tonnetz

Fig. 2 Tiling of the plane

of the tiling, it is the equilateral triangle. Mathematicians refer to such an

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 3 Congruence transformations

Fig. 4 Period parallelograms

Third, each of the patterns shown can be understood to repeat indefinitely

© 2008 The Author. Music Analysis, 27/i (2008)

Fig. 5 Toroidal Tonnetz

Paradoxes of Visual Space in the Artwork of M. C. Escher

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 6 M. C. Escher, Mouch

© 2008 The Author. Music Analysis, 27/i (2008)

Escher was likewise fascinated by the paradox of impossible pathways for

Paradoxes of Musical Space in Wagner’s Parsifal

Music Analysis, 27/i (2008) © 2008 The Author.

Ex. 2 Wagner, Parsifal, Act III, bars 1123–1127 (Kundry’s Entseelung)

Ex. 3 Hexatonic polar progressions

According to Cohn, the uncanniness of the hexatonic polar progression is

© 2008 The Author. Music Analysis, 27/i (2008)

Fig. 7 Hexatonic polar progressions within the neo-Riemannian Tonnetz

Neo-Riemannian theory has shed new light on the paradoxical nature of

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 9 Embodied image schemas

patterns of embodied experience, known as image schemas, appear to play a

The Embodied Nature of Musical and Visual Perception

© 2008 The Author. Music Analysis, 27/i (2008)

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 10 Combined image schemas

CONTAINER, CENTRE-PERIPHERY and BALANCE schemas may be combined because

© 2008 The Author. Music Analysis, 27/i (2008)

of balance is governed by the geometry of the square – its centre, edges,

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 14 The neo-Riemannian ‘basic shape’

Fig. 15 Modified version of the neo-Riemannian ‘basic shape’

© 2008 The Author. Music Analysis, 27/i (2008)

Fig. 16 Reinterpretation of the neo-Riemannian ‘basic shape’

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 17 Mapping of the overtone series onto a VERTICAL PATH schema

© 2008 The Author. Music Analysis, 27/i (2008)

of a vertical pathway or ‘ladder’ which ascends indefinitely towards a vanishing

Music Analysis, 27/i (2008) © 2008 The Author.

Fig. 18 (a) Register and chroma; (b) Shepard’s tones

existence. We can conceive of each vertical line as having been formed by a

© 2008 The Author. Music Analysis, 27/i (2008)

Ex. 4 Root movement and voice leading

Music Analysis, 27/i (2008) © 2008 The Author.

© 2008 The Author. Music Analysis, 27/i (2008)

eighteenth- and nineteenth-century composers and theorists as to how these

Constructing an Embodied Model of Triadic Pitch Space

Fig. 19 Construction of triadic pitch space, stage 1

Music Analysis, 27/i (2008) © 2008 The Author.