SPECIAL ISSUE

SOUND-COLOUR SYNAESTHESIA:
TO WHAT EXTENT DOES IT USE CROSS-MODAL MECHANISMS
COMMON TO US ALL?
Jamie Ward, Brett Huckstep and Elias Tsakanikos
(Department of Psychology, University College London, London, UK)

ABSTRACT

This study examines a group of synaesthetes who report colour sensations in response to music and other sounds.
Experiment 1 shows that synaesthetes choose more precise colours and are more internally consistent in their choice of
colours given a set of sounds of varying pitch, timbre and composition (single notes or dyads) relative to a group of
controls. In spite of this difference, both controls and synaesthetes appear to use the same heuristics for matching between
auditory and visual domains (e.g., pitch to lightness). We take this as evidence that synaesthesia may recruit some of the
mechanisms used in normal cross-modal perception. Experiment 2 establishes that synaesthetic colours are automatically
elicited insofar as they give rise to cross-modal Stroop interference. Experiment 3 uses a variant of the cross-modal Posner
paradigm in which detection of a lateralised target is enhanced when combined with a non-informative but synaesthetically
congruent sound-colour pairing. This suggests that synaesthesia uses the same (or an analogous) mechanism of exogenous
cross-modal orienting as normal perception. Overall, the results support the conclusion that this form of synaesthesia
recruits some of the same mechanisms used in normal cross-modal perception rather than using direct, privileged pathways
between unimodal auditory and unimodal visual areas that are absent in most other adults.

Key words: synaesthesia, music, colour, cross-modal, audiovisual, ventriloquism

INTRODUCTION Whilst there is mileage in any study showing

that a given type of synaesthesia is genuine,
For some individuals, auditory sensations research must also focus upon understanding in
automatically elicit a definite and reliable visual more detail the nature of the underlying cognitive
experience. For example, middle ‘C’ on a piano and neural processes. In particular, research should
might be red but the note three octaves higher might consider how synaesthesia can be used to inform
be green (Ginsberg, 1923) and a “sharp rap on a table theories of normal cognition. Sound-colour
[may arouse] a faint but distinct narrow flash of synaesthesia is an attractive candidate to address
brightness” (Whipple, 1900, p. 393). These this question because there is some evidence to
individuals probably have sound-colour synaesthesia suggest that this variety of synaesthesia might be
(or music-colour synaesthesia, also-known-as present in us all from a young age (Lewkowicz and
chromaesthesia; Cytowic, 2002). This type of Turkewitz, 1980; Marks et al., 1987). Moreover,
synaesthesia is the focus of the present investigation. cross-modal audiovisual mechanisms exist in the
Three experiments are reported which establish normal population more generally (e.g., Calvert,
that this is a genuine form of synaesthesia, insofar 2001).
as these individuals can be distinguished from non- Neuroanatomical studies of young kittens show
synaesthetes given an identical task. Tests of pathways between auditory and visual brain centres
consistency (e.g., Baron-Cohen et al., 1993) and (Dehay et al., 1984), and human neonates show
automaticity (e.g., Mills et al., 1999) have yet to be visual evoked potentials given auditory stimuli
applied to sound-colour synaesthesia, despite the (e.g., Maurer, 1997). These observations have lead
phenomenon being anecdotally reported many Baron-Cohen and colleagues to propose a neonatal
times before. Whilst some studies have reported pruning hypothesis of synaesthesia (Baron-Cohen,
their sound-colour synaesthetes to be consistent 1996; Baron-Cohen et al., 1987, 1993, 1996). They
over long periods of time (e.g., Ginsberg, 1923; assume that everyone is born with synaesthesia-like
Ostwald, 1964), these studies have not compared neural hardware but that most of us lose this
performance to a control group of participants. This during the natural course of development.
could be particularly important for sound-colour Synaesthetes, perhaps because of a genetic
mappings because these are not necessarily modification, do not lose this neural hardware and
arbitrary. For example, most people tend to retain synaesthesia into adulthood. As such, this
associate high pitch sounds with light colours account proposes that there are specialised
(Hubbard, 1996; Marks, 1974, 1982, 1987; Melara, pathways in adult synaesthetes that are not present
1989) and most people are able to generate, on in most other individuals. Thus, in this account,
demand, visual imagery to music (e.g., Karwoski synaesthesia is not an extension of the cross-modal
and Odbert, 1938). mechanisms common to all adults.

Cortex, paper in press published online

2 Jamie Ward and Others

Fig. 1 – Sound-colour synaesthesia could plausibly arise from (I) direct connections between unimodal auditory and visual centers
that are not normally found in other people, or from (II) feedback from multi-modal to unimodal regions that resembles normal cross-
modal perception.

An alternative viewpoint would be that sound- sensory deprivation. Given that pharamcological
colour synaesthesia utilises pathways that are used synaesthesia can be induced within an hour it is
to integrate visual and auditory information as part reasonable to assume that it uses pre-existing
of the normal mechanisms of cross-modal pathways in the brain (Hollister, 1968), although it
perception. There are cross-modal audiovisual areas remains to be seen whether the characteristics are
of the brain that appear to respond more to the similar to those exhibited in the developmental
combined presence of vision and sound than to form.
either, or the sum of, vision or sound alone (e.g., Further evidence to suggest that synaesthesia
Calvert, 2001). Behaviourally, this may be could be an extension of normal cross-modal
important for tasks such as lip-reading (e.g., perceptual mechanisms comes from the observation
Calvert, 2001) or attentional orienting mechanisms that both synaesthetes and non-synaesthetes match
(e.g., Spence and Driver, 1997). If there is colours with sounds in a non-arbitrary way. The
‘disinhibited feedback’ from cross-modal most commonly studied domain is pitch1. In the
audiovisual areas of the brain to unimodal visual synaesthesia literature, it appears that almost all
areas of the brain then this could produce overt cases reported to date show the trend that high
visual responses to auditory stimuli (e.g., pitched sounds are lighter/brighter in colour and
Grossenbacher, 1997; Grossenbacher and Lovelace, low pitched sounds are darker in colour (e.g.,
2001). Given that these pathways are common to Cutsforth, 1925; Riggs and Karwoski, 1934;
us all, synaesthesia would be seen as an extension Whitchurch, 1922; Zigler, 1930). However, it is to
or exaggeration of these particular pathways rather be noted that the authenticity of these historical
than requiring an extra pathway. Figure 1 cases cannot be verified. The same trend is found
simplistically compares the feedback account and in non-synaesthetes. This is found if people are
the direct pathway account. Disinhibited feedback asked to select colours for a pitch (Marks, 1974),
could also account for the fact that sound-colour rate how well colours and pitches go together
synaesthesia can be acquired by damage to the (Hubbard, 1996), or even in Stroop-like tasks in
retino-cortical pathway (e.g., Armel and which participants make a light/dark judgement
Ramachandran, 1999; Vike et al., 1984) or from whilst listening to an irrelevant high or low pitched
ingestion of hallucinogenic drugs (e.g., Hartman tone (Marks, 1987; Melara, 1989). It is tempting to
and Hollister, 1963). The latter are thought to conclude that the synaesthetes and non-
operate by impeding the transmission of synaesthetes are using the same cognitive and
information from the lateral geniculate nucleus to neural mechanisms to derive these associations, but
the cortex (Hollister, 1968). Thus, both forms of
acquired sound-colour synaesthesia can be viewed
as a compensatory ‘switching on’ of visual cortex 1 By which we mean pitch height, rather than pitch class (e.g., C, D#, Bb).
from non-visual information as a result of visual We return to this distinction in the Discussion.
 Sound-colour synaesthesia 3

with only the former eliciting an overt, definite internal consistency than control participants, and
perceptual experience. However, to date no study also to investigate how different aspects of an
has made a direct comparison between sound- auditory stimulus are mapped on to the visual
colour synaesthtes and controls using the same domain in both synaesthetes and controls.
task.
If several different notes are played Method
simultaneously, as in a chord, most synaesthetes
report experiencing several colours rather than a Participants
fusion of colours (Cutsforth, 1925; Ginsberg, 1923;
Myers, 1911; Riggs and Karwoski, 1934; Rogers, Ten sound-colour synaesthetes and 10 control
1987; Whipple, 1900). This suggests that the participants took part. The 10 control participants
critical auditory unit is at the level of the single consisted of 8 females and 2 males and none of
note (or pitch) rather than a unit of perception them reported any synaesthetic experiences. The 10
operating across several notes (e.g., pitch synaesthetes had a mean age of 42.5 years,
groupings, interval values or melody). and consisted of 6 females (DSL, LHM, MH,
Synaesthesia involving aspects of music other SAW, SAWE, KZ) and 4 males (JW, JH, SJT, RJ).
than pitch have been documented in far less detail. All ten synaesthetes reported grapheme-to-colour
Other potential influences would include timbre synaesthesia in addition to sound-to-colour
(related to the instrument), loudness, tempo, synaesthesia. Over a 2-3 month interval their
emotional response, consonance/dissonance, the colour association to written letters (N = 26) and
name or pitch class of the note (e.g., C, F#) and numerals (N = 10) was 86% consistent. This
the key or tonality of a piece (e.g., C major) (e.g., compares to a 33% level of consistency from 43
see Marks, 1975). Timbre is the aspect of sound control subjects given a retest after only 2 weeks [t
that enables a perceiver to judge that two sounds (51) = 9.31, p < .001]. Three of our sample (JW,
are different when they are otherwise matched for LHM, KZ) also report vivid complex shapes in
pitch and loudness (e.g., to distinguish between the response to tones (e.g., Zigler, 1930) upon which
same notes on a trumpet and cello). Zigler (1930) colours are superimposed. Other synaesthetes do
reported two cases of synaesthesia that experienced not report shapes as such, although the colour
both colours and shapes in response to tones. responses are typically accompanied by a sense of
Whereas pitch influenced the shade, the timbre had ‘space’ (e.g., in terms of location and/or size). For
more effect on the shape: “Every instrument example, on hearing a siren outside of the window,
excites a specific form, which maintains roughly one subject remarked that it was like a thin bright
the same features at all pitches, intensities and line moving across his visual field, similar to the
durations” (p. 285). In other cases it has been trail left when someone waves a sparkler. These
reported that both pitch and timbre affects the aspects of their sound-induced synaesthesia are not
colour (Cutsforth, 1925; Masson, 1952; examined in this study and we concentrate solely
Whitchurch, 1922) although in some cases the on colour. None of the synaesthetes reported
same colour is elicited by a given pitch irrespective having absolute pitch (i.e., the ability to identify or
of timbre (Myers, 1911). Mudge (1920) notes that produce a particular note without an external
non-synaesthetes also tend to ascribe different reference note) but we turn to this again in the
colours to different timbres (e.g., flute and clarinet discussion.
are light, whereas trombone and bass drum are
dark). However, this study conflated timbre with Sound Stimuli
the different pitch registers of the instruments.
In this study, Experiment 1 compares the Seventy different sound stimuli were used. The
influence of pitch and timbre on synaesthetic sounds were all of 3 seconds duration recorded as
colour experiences. We also consider, in less detail, 16-bit stereo files, at a sampling frequency of
differences between pairs of notes (dyads) and 44,100 Hz (CD quality) and 65 dB. The sounds
single notes. By comparing the performance of were selected to represent a range of pitches and
synaesthetes directly to controls, we hope to shed timbres, and also to contrast single notes (N = 40)
light on the question of whether this form of with pairs of simultaneous notes or dyads (N =
synaesthesia may be using some of the same 30). The 40 single notes consisted of 10 piano
mechanisms that are available to us all. notes, 10 pure tones (sine wave), 10 string notes,
Experiments 2 and 3 are concerned with the and 10 further notes taken from 10 different
automaticity of cross-modally induced synaesthesia. instruments (e.g., flute, bagpipe). The piano, string
and pure tone notes were matched for frequency of
the fundamental and consisted of the notes from
EXPERIMENT 1: SOUND-COLOUR ASSOCIATIONS C1 (33 Hz) up to Eb6 (1245 Hz) separated by
intervals of musical fifths (i.e., 700 cents). Thus,
The aim of this experiment is to demonstrate the following notes were used: C1, G1, D2, A2,
that sound-colour synaesthetes show a higher E3, B3, F#4, Db5, Ab5 and Eb6.
4 Jamie Ward and Others

The remaining 10 single notes from the Procedure

different instruments were all matched for
fundamental frequency corresponding to that of The 70 sound stimuli were presented in a
middle C (i.e., 262 Hz). The 30 dyads also random order and the test was self-paced.
consisted of 10 piano dyads, 10 pure tone dyads Participants were seated in front of a computer in a
and 10 string dyads. The dyads comprised of dimmed room and put on a pair of Sennheiser
simultaneous pairs of notes separated by an interval HD250 headphones. They were given instructions
of a musical fifth ranging from C1 (33 Hz) up to and practice at using the colour picker before the
Bb6 (1865 Hz). Thus, the following dyads were start of the experiment. The screen contained a
used: C1 + G1, G1 + D2, D2 + A2, A2 + E3, E3 ‘Play’ button which played the sound and a ‘Colour’
+ B3, B3 + F#4, F#4 + Db5, Db5 + Ab5, Ab5 + button to call up the colour picker. Participants
Eb6, Eb6 + Bb6. In this way, we are able to could play the sound more than once if they wished
contrast the colours assigned to isolated notes with and go back and modify the colour. However, after
the same notes presented in pairs. participants had opted to move on to the next trial
they were unable to go back and change a previous
Colour Selection colour or listen to the old sound. It took between 20
and 40 minutes to complete all 70 selections.
The experiment used the standard ‘Windows The whole procedure was repeated on two
API Choose Color’ dialog box that was summoned further occasions. The first repetition took place
by the Visual Basic routine used to run the test. within the same session after a 10 minutes interval,
This was chosen because it is likely to be familiar using a different randomised order. The second
to most regular computer uses and has previously repetition took place after a 2-3 month interval.
been used in studies of synaesthesia (e.g., Odgaard
et al., 1999). It offers the participants a wide array Results
of colours to select from. Colours could be chosen
from one of two sets: either a basic palette of 48 Internal Consistency over Time
preset colours or a more fine-grained colour
specification made by moving a cross-hair cursor The consistency of the colours generated for each
over a colour matrix. Participants could customise sound was measured quantitatively by calculating
the colour further by adjusting a vertical slider that the vector difference in RGB choices between test
controls luminance. Given that synaesthetes report and retest. For example, if an RGB value of R = 255,
very precise colour experiences, we predict that G = 200, B = 50 was selected at Time 1 and R = 255,
they would be more likely to use the finer-grained G = 250, B = 75 was selected at Time 2 then the
customised colours than control participants. Each distance between these points in RGB space is given
colour selection can be represented numerically by by √ (02 + 502 + 252) = 55.9. These results were
a single RGB (red, green, blue) vector with values analysed using a 2 × 2 ANOVA on the RGB distance
ranging from 0 to 255 (where 0,0,0 is black and scores. The two factors were group (synaesthete,
255,255,255 is white). It is possible to convert non-synaesthete) and retest interval (same day, 2-3
RGB values to an approximate position in Munsell months). The results are summarised in Figure 2.
space using look-up tables used in software-based There was a main effect of group, indicating
systems for displaying Munsell (Campadelli et al., that the synaesthetes were significantly more
2000)2. The Munsell colour space is considered a consistent over time than controls [F (1, 18) =
more psychophysically plausible representation of 22.31, p < .01]. There was no main effect of time
colour in which colour is described according to although it was borderline significant [F (1, 18) =
three dimensions: value (light to dark on a 0-10 3.49, p = .08] and no interaction between them [F
scale), chroma (grey to coloured) and hue (a (1,1 8) = .01, n.s.]. Thus, the tendency for
circularly varying dimension, red → yellow → synaesthetes to be more consistent in their colour
green → blue → purple → red). Standard RGB associations over time than controls holds for both
colours are a subset of the Munsell colours, but short and long retest intervals. As far as we are
some Munsell colours cannot be represented aware, this is the first time that this result has been
in RGB space (and thus could not be selected). reported for this variety of synaesthesia. On the
This will be a source of noise in our data. same day retest, only one synaesthete (MH) out of
However, given that synaesthetes, but not controls, ten fell within the control range of scores and at
are putatively matching to a perceptual experience the 2 month retest, two synaesthetes (MH and RJ)
we might expect it to disadvantage them more (i.e., fell within the control range of scores.
the method ‘stacks the deck’ against our Figure 3 illustrates the colours selected for the
hypothesis). 10 single piano, 10 single string and 10 single sine
waves on two occasions, for two of our subjects
(the synaesthete LHM and a control participant). It
2 Note, a published Munsell colour chart was also available to ensure that
is to be noted that both show a trend for higher
the monitor was adequately calibrated. pitch notes to elicit lighter colours although LHM
 Sound-colour synaesthesia 5

Fig. 2 – Consistency over time, following a same day retest and a subsequent retest several months later (SD for synaesthetes is 26
and 36 resepctively, and for controls 26 and 23).

is clearly more internally consistent in her that this would be reflected in their use of the
assignment of colours to tones. These are typical customised colour picker.
colour responses to this task (for information, The results show that this hypothesis was borne
LHM’s RGB consistency, as calculated above, was out. Synaesthetes chose significantly more
55.8 whereas CE’s value was 115.0). customised colours (80.7%, SD = 22) than the
The same results are obtained if RGB space is control group [61.6%, SD = 19.2; t (18) = 2.07, p
transformed into separate Munsell dimensions. < .05]. Thus, not only are synaesthetes more
Treating differences in hue, chroma and value as consistent over time in their choice of colours, but
dependent variables in 2 × 2 ANOVAs (group X they also tend to choose very specific colours to
time) shows significant effects of group for hue [F represent their experiences.
(1, 18) = 31.38, p < 001], value [F (1, 18) = 18.37,
p < 001] and chroma [F (1, 18) = 24.63, p < 001]. Effects of Pitch and Timbre on Lightness and
That is, synaesthetes were significantly more Chroma
consistent on all three dimensions than controls.
There were main effects of time for value [F (1, It was noted in the Introduction that previous
18) = 7.04 p < .05] but not hue [F (1, 18) = 1.33, studies have shown that both synaesthetes and non-
n.s.] or chroma [F (1, 18) = 3.76 p = .07], and synaesthetes display a tendency to associate low
there were no interactions between group and time pitch notes with dark colours and high pitch notes
(F < 1). with light colours. However, to the best of our
knowledge, a direct comparison of synaesthetes
Specificity of Colours Selected and non-synaesthetes has never been conducted
using the same task and stimuli. In this section, we
Colours could be selected either from a preset consider single note stimuli. Dyads will be
palette of 48 different colours or the participant considered in a subsequent section and in
could produce a colour of their own by modifying somewhat less detail.
one of the preset colours. It is possible to calculate The Munsell lightness values were analysed in a
how many preset colours each person chose versus 10 × 2 × 3 ANOVA contrasting pitch (× 10), group
how many customised colours they chose. Our (synaesthete, non-synaesthete) and timbre (pure
hypothesis is that synaesthetes will experience very tone, piano, string), averaging across the different
specific colours in response to tones (e.g., a testing sessions. Both groups show a monotonic
particular yellow, rather than a generic yellow) and increase in lightness with pitch, suggesting a
6 Jamie Ward and Others

Fig. 3 – An example of the colours selected (on 2 occasions) for the 10 single piano, sine (or pure tone) and string notes for a
synaesthete (LHM, top) and a control subject (CE, bottom).

common underlying mechanism. This is illustrated a tendency for synaesthetes to select somewhat
in Figure 4. There was no effect of timbre and lighter colours at the highest pitches.
timbre did not interact with any other factors (all F’s The same analysis was conducted on the
< 2), and so this dimension is not graphically Munsell chroma values, although we did not have
displayed. There was a highly significant effect of specific predictions about this dimension3. The
pitch [F (9, 162) = 56.32, p < .001], but no
difference between synaesthetes and controls [F (1,
18) = 2.51, n.s.] although a significant group X 3 Note,
hue is a circularly varying dimension and cannot be analysed in the
pitch interaction [F (9, 162) = 3.62, p < .05] reflects same way.
 Sound-colour synaesthesia 7

Fig. 4 – The relationship between pitch and lightness for synaesthetes and controls for the 30 single notes (collapsed across timbres
and testing sessions): 0 = darkest, 10 = lightest.

chroma values were analysed in a 10 × 2 × 3 However, these regularities appear to be common

ANOVA contrasting pitch (× 10), group to synaesthetes and control subjects alike.
(synaesthete, non-synaesthete) and timbre (pure Finally, we investigated differences in lightness
tone, piano, string), and averaging across the and chroma across the 10 tones taken from
different testing sessions. Pitch exerts an effect on different instruments (bagpipes, flute, etc.) that
chroma [F (9, 162) = 9.28, p < .001] and not just were matched for pitch of the fundamental
lightness. However, chroma differs from lightness frequency (at middle C). Given the results
in two key respects. Firstly, there is no monotonic presented above, one might expect differences to
relationship. Instead, chroma peaks at mid-range be manifest on the chroma but not lightness
pitches (in our experiment, a semitone below dimension. Two separate 2 × 10 ANOVA’s were
‘middle C’). Secondly, there was a highly conducted with group (synaesthete, control) and
significant main effect of timbre on chroma [F (2, instrument (N = 10) as independent variables and
36) = 16.89, p < .001] that was not observed for chroma and lightness as dependent variables.
lightness. Thus, musical notes from the piano and Contrary to our prediction, there were main effects
strings are, literally, more colourful than pure of timbre on both lightness [F (9, 162) = 11.56, p
tones. This is illustrated in Figure 5. Importantly, < .001] and chroma values [F (9, 162) = 3.29, p <
there was no group difference between synaesthetes .001]. Yet again, however, there was no difference
and controls [F (1, 18) = 1.26, n.s.] and none of between synaesthete and control groups and no
the interactions approached significance. In sum, interactions between group and timbre (all F’s < 2).
different aspects of auditory stimuli (pitch, timbre) This is illustrated in Figure 6. Even though the
appear to map on to different aspects of colour frequency of the fundamental was matched, it is
(lightness, colourfulness) in systematic ways. quite conceivable that differences in perceived
8 Jamie Ward and Others

Fig. 5 – The relationship between pitch, timbre and chroma for synaesthetes (top) and controls (bottom) presented with single notes.
The notes are most colourful around the B3 pitch (a semitone below ‘middle C’) and for musical notes (string, piano) relative to pure
tones.

pitch (and hence lightness) occurred across the synaesthetes appear to map auditory properties to
instruments. This is because the perceived pitch is visual properties in very similar ways.
derived from the wider spectral properties of the
tone and not just the fundamental frequency (e.g., Single Tones versus Dyads:
Plomp, 1967). As such, a more convincing Single Colour or Multi-coloured Experiences?
demonstration of different visual correlates of pitch
and timbre is required. However, the main point On the 2-3 month retest trial, synaesthetes were
that we wish to emphasise is that controls and asked whether each of the 70 tones produced a
 Sound-colour synaesthesia 9

Fig. 6 – Different instruments are associated with different amounts of lightness and chroma, but these differences are found equally
in synaesthetes and controls.

single colour or several different colours. This was The results are summarised in Figure 7. A 3 × 2
motivated both by the initial verbal reports of our ANOVA was conducted on the type of instrument
synaesthetes and also from similar observations (piano, sine, string) and the type of tone (single
made in the historical literature on synaesthesia note, dyad). There was a main effect of type of
(Cutsforth, 1925; Ginsberg, 1923; Myers, 1911; tone [F (1, 9) = 14.09, p < .005] with the dyads
Riggs and Karwoski, 1934; Rogers, 1987; Whipple, tending to elicit two or more colours. There was
1900). Given that none of our controls reported also a main effect of type of instrument [F (2, 8) =
actual colour experiences, we did not consider it 7.28, p <.05], with piano and strings being more
pertinent for them to be asked about this. likely to elicit two or more colours than pure tones.
10 Jamie Ward and Others

Fig. 7 – Dyads and timbres with multiple harmonics often elicit more then one colour.

This could either be attributable to an effect of is noteworthy. JW tends to report single colours to
timbre per se or due, more simply, to the fact that dyads and he appears to use a different mechanism
natural musical tones have more complex spectral to derive the colour of dyads than that used to
properties. There was no significant interaction [F derive the colour of isolated notes. For example, a
(2, 8) = 1.59, n.s.]. 370 Hz single piano note elicits a darkish off-
The dyads also show a monotonic increase in yellow colour and a 554 Hz piano note elicits a
Munsell lightness with increasing pitch (collapsing slightly lighter, purer yellow colour (reliably so
across timbres). A 2 × 10 ANOVA of note type over time). However, when the 370 Hz and 554 Hz
(dyad vs. single note) and pitch (10 levels) on the notes are played simultaneously as a dyad then a
synaesthetes revealed differences in lightness pinkish colour is consistently reported that bears no
across pitch [F (9, 81) = 56.29, p < .001] but no obvious relationship to the colour of the isolated
difference between dyads and single notes [F (1, 9) notes. It suggests that the unit of representation
= .40, n.s.] and no interaction [F (9, 81) = .63, that drives the association of colours in JW
n.s.]. Thus, the lightness assigned to single notes operates over several notes (e.g., the interval)
(e.g., C1) is also found when these notes are the whereas for most other synaesthetes it operates at
bass notes of a dyad (e.g., C1 + G1) – at least for the single note level.
intervals of a fifth. Control subjects show the same
pattern: a significant effect of pitch [F (9, 81) = Summary
20.84, p < .001], no difference between dyads and
single notes [F (1, 9) = 1.78, n.s.] and no In summary, Experiment 1 has demonstrated
interaction [F (9, 81) = 1.74, n.s.]. that sound-colour synaesthesia can objectively be
Given that participants were only allowed to distinguished from sound-colour associations
choose one colour for each tone, one may wonder generated by the general population in terms of
how they made their judgement in the case of internal consistency and specificity of colours
dyads that tended to elicit two colours. An selected. As such, it resembles previous reports on
inspection of the colours reveals that (with the grapheme-colour synaesthesia (e.g., Baron-Cohen
exception of JW) the colour of the dyads typically et al., 1993) and synaesthesia in which taste is the
resembles the colour of the component notes that elicited experience (Ward and Simner, 2003).
make it up. Given that the dyads consisted of notes However, we have been able to take our analysis
of similar pitch height and always of the same one step further by considering the possible
timbre, the colours tended to be quite similar to cognitive mechanisms employed by both
each other (e.g., two slightly different shades of synaesthetes and controls. Both synaesthetes and
brown). There is at least one exception to this that controls appear to employ an identical strategy for
 Sound-colour synaesthesia 11

mapping sounds to colours that involves pitch- effort was made to select a range of different
lightness mappings, and they show identical colours and, hence, a range of different pitches and
influences of timbre on chroma. We therefore timbres were also used. Incongruent colour-tone
believe that this variety of synaesthesia can best be pairs were created by pairing up the 6 colours and
explained as an exaggeration of cross-modal the 6 tones in a different way. There were 120
mechanisms common to us all rather than a trials in total, with equal numbers of congruent and
privileged, direct pathway between audition and incongruent pairings. For half of the trials the onset
vision that is present in synaesthetes but not in of the sound and the colour patch were
others. In synaesthetes we consider the mapping to simultaneous, and for the remainder of trials the
be precise (as shown by high internal consistency) sound preceded the colour patch by 150 msec.
and results in a vivid perceptual experience, Each control subject was given one of the sound-
whereas in controls the mapping is noisy and is not colour sets of a corresponding synaesthete.
associated with reports of actual visual experience.
One further difference between the two groups may Procedure
lie in the automaticity of the mapping between
sound and colour. To test this, the next experiment Stimuli were presented using E-prime software.
uses a cross-modal variant of the synaesthetic Subjects were seated in front of a computer,
Stroop effect. wearing Sennheiser headphones, and with a
microphone placed close to their mouth. They were
told that they would see some coloured patches on
EXPERIMENT 2: the screen and, at the same time, hear some sounds
CROSS-MODAL STROOP INTERFERENCE presented over the headphones. They were
instructed to name the colour of the patch on the
The aim of this experiment was to establish screen as quickly and accurately as possible and
that, in the synaesthetes but not controls, the colour that they should try to ignore the sound. They were
experiences are automatically elicited upon asked to produce a single colour term (e.g., “blue”
presentation of a tone. This is analogous to the rather than “dark blue” or “royal blue”). Each trial
Stroop-like interference effect found in grapheme- proceeded as follows. There was a central fixation
colour synaesthesia (e.g., Mills et al., 1999). In this cross presented for 1000 msec. Following this, the
instance, synaesthetes and controls were asked to tone was played through the headphones. A
name the colour of a patch on the screen whilst coloured rectangle appeared in the centre of the
simultaneously listening to tones over the screen against a mid-gray background either at the
headphones which they were asked to ignore. For same time as the tone or 150 msec later. When
the synaesthetes, the tones either elicited a subjects made their response, both the colour and
synaesthetic colour that was congruent with the the tone stopped and there was a 1000 msec
colour on the screen or a synaesthetic colour that interval before the beginning of the next trial.
was incongruent with it.
Results and Discussion
Method
The results are summarized in Figure 8. A 2 ×
Design 2 × 2 mixed ANOVA revealed a significant main
effect of SOA [F (1, 16) = 25.72, p < .001], a
The experiment uses a mixed 2 × 2 × 2 design significant effect of stimulus congruency [F (1, 16)
with tone-colour pairing (congruent, incongruent) = 6.28, p < .05] but the effect of group (i.e.,
as one factor, sound-colour onset asynchrony as a synaesthete vs. control) did not reach significance
second factor (sound precedes colour by 0 or 150 [F (1, 16) = 3.97, p = .06]. Critically, there was a
msec) and presence or absence of synaesthesia as significant interaction between group and stimulus
the between-subject factor. congruency [F (1, 16) = 8.56, p < .01] suggesting
that the synaesthetes show a greater effect of
Participants sound-colour congruency/incongruency than
controls. No other interaction approached
Nine synaesthetes from Experiment 1 were significance (p > .20). In order to examine the
retested together with nine controls. One of the group × stimulus congruency interaction further,
synaesthetes who took part in Experiment 1 was two separate repeated measures were run on the
unavailable for further testing. synaesthete group and on the control group. For
the synaesthetes, there was a significant effect of
Materials SOA [F (1, 8) = 13.55, p < .01] and of stimulus
congruency [F (1, 8) = 8.07, p < .05] but no
For each synaesthete, six tones were selected interaction. For the control group, a similar
from the set used in Experiment 1 on the basis that analysis revealed only a significant effect of SOA
they had previously elicited a consistent colour. An [F (1, 8) = 13.55, p < .01]. This suggests that the
12 Jamie Ward and Others

Fig. 8 – Synaesthetes show cross-modal Stroop-like interference when incongruent sounds and colours are presented together (0 msec
SOA) and when sound precedes colour (150 msec SOA). For example, synaesthetes are slowed at naming a red patch when also given a
‘blue sound’ (SD for synaesthetes and controls across conditions is 190 and 107 respectively).

main effect of congruency reported in the initial after the Stroop experiment (showing them the
analysis is entirely attributable to the performance stimuli again if need be) but we found only two
of the synaesthetes. other synaesthetes who reported the phenomenon.
This is the first ever demonstration of a Thus, this illusion is by no means universal.
synaesthetic Stroop effect that is cross-modal in Notwithstanding this, it prompted us to ask
nature. It suggests that the synaesthetic colours are whether non-spatial auditory cues could lead to
automatically elicited even when the inducing ‘covert’ spatial orienting to a congruent visual
stimulus is task irrelevant and even when the stimulus even when the cue was completely
inducing stimulus is in a different sensory uninformative.
modality. It is convincing evidence for the
authenticity of this type of synaesthesia over and
above the intriguing commonalities that exist EXPERIMENT 3:
between synaesthetes and controls in terms of the CROSS-MODAL ORIENTING OF SPATIAL ATTENTION
way that they choose colours given sounds.
In a previous pilot study, one of our In the Stroop experiment, above, participants
synaesthetes (LHM) commented that when the were required to ignore the sound whilst attending
sound and the colour were congruent it felt as if to the colour dimension. In the present experiment,
the sound was coming out of the colour on the both the sound and the colour are completely
screen, rather than through the headphones. That is, irrelevant and uninformative for the task at hand.
she reported an illusory capture of the location of The task is a cross-modal variant of the Posner
sound by a congruent visual stimulus. This cueing paradigm (Posner, 1980; Posner and Cohen,
constitutes a synaesthetic version of the 1984). In the original task, a peripheral lateralised
‘ventriloquist effect’ that has been documented in visual cue preceded the presentation of a lateralised
studies of normal cross-modal perception (e.g., visual target on either the same side or the opposite
Driver, 1996; Spence and Driver, 2000). The side as the cue. The cue was found to facilitate
synaesthete had not studied psychology and was same-side detection times (up to around 200 msec
unlikely to be aware of the literature on this SOA) even though the cue itself was uninformative
subject. In the current study we took the (e.g., on half the trials it appeared on the wrong
opportunity to question all the other synaesthetes side). At longer delays (> 300 msec) inhibitory
 Sound-colour synaesthesia 13

effects were found associated with a cost in squares would contain an asterisk (*) and the task
shifting attention back to a previously empty of the participants was to identify the target as
location. Similar effects have been found for quickly as possible. None of the synaesthetes
lateralised auditory cues and visual targets (e.g., reported an experience of colour to the asterisk.
Spence and Driver, 1997). In the present Participants made their response by pressing the
experiment, the auditory cue is presented through space bar upon target detection. Participants were
both ears of the headphones and, as such, has no informed that they would also hear a sound
lateralised component. The auditory cue is, through the headphones whilst performing the task.
however, presented at the same time as two They were told that the tone is not relevant to the
coloured rectangles on the left and right of the task and that they should ignore it. The fixation
screen, one of which is synaesthetically congruent cross was displayed for 1000 msec on a mid-grey
with the sound. The task is to detect a target (an background. Following this the coloured squares
asterisk) presented in one of the two rectangles a were displayed on the screen against the same mid-
short time after the cue (50, 150 or 300 msec). grey background. The squares were approximately
Will an auditory cue orient attention to the 5 cm in size and their centres were separated by 22
synaesthetically congruent location? cm on the screen (i.e., the targets were 8.9 degrees
from fixation). The auditory tone was always
Method played at the same time as the coloured patches
appeared. The target appeared after a delay of
Design either 50, 150 or 300 msec. Both the target and the
tone remained until the participant made a response
The experiment uses a 2 × 3 × 2 mixed design or until a 1500 msec time-out (whichever came
with location of the target as one factor (either on first). The next trial then began immediately.
the congruent or incongruent colour), the onset
asynchrony between auditory cue and visual target Results and Discussion
as another factor (50 msec, 150 msec, 300 msec)
and the type of subject (synaesthete or control) as The results are summarized in Figure 9. A 2 × 3
the final factor. × 2 mixed ANOVA on the reaction times revealed a
significant congruency X group interaction [F (1, 16)
Participants = 4.30, p < .05] which suggests that synaesthetic
congruency between sound and colour can result in
The same synaesthete and control participants spatial orienting of attention. There was a marginally
as in Experiment 2 took part in this experiment. insignificant 3 way interaction between congruency
× group × SOA [F (2, 32) = 3.11, p = .06] and a
Materials marginally insignificant effect of group [F (1, 16) =
3.82, p = .07] consistent with the observation that
The 6 consistent colour-tone pairings used in synaesthetes tend to be slower particularly when
Experiment 2 were used again in the present cued to the wrong location. Other effects did not
experiment. For each tone, there were 20 possible approach significance (p > .10). There was no
permutations of colour and target location pairings evidence of an interaction between congruency and
– the congruent colour was either on the left or SOA. We would have predicted this interaction based
right (x2), the incongruent colour was one of the on the standard Posner cueing paradigm. If the target
remaining colours (x5) and the target either does not appear in the cued location, attention is
appeared on the congruent or incongruent colour shifted to the other location and there is a cost of
(x2). Thus, for the 6 tones there were 120 unique returning to the previous location (‘inhibition of
pairings of colour, sound and target position. With return’). Presumably it takes some (unknown)
the addition of the onset-asynchrony variable amount of time for the synaesthesia to be elicited and
between the sound and the target each experimental so it is perhaps not too surprising that there was no
block consisted of 360 trials. In addition there were inhibition of return in the time window used, but we
90 catch trials in which the target was absent would predict it at longer SOAs. The proportion of
(some 20% of trials). The 450 trials were presented missed targets and false alarms was low (misses =
in a random order. 1.06%, false alarms = 3.70%) and did not differ
between synaesthetes and controls [misses t (16) =
Procedure .67, n.s.; false alarms t (16) = .23, n.s.].
To summarise again, when the target appears in
The participants were seated 70 cm away from an incongruently coloured location with respect to
a colour computer monitor. They were told that the sound then the synaesthetes, but not the controls,
they would first of all see a fixation cross (+) in tend to be slowed. This presumably occurs because
the centre of the screen that they should attend to, attention is ‘captured’ by the congruent location.
and that they would then see two coloured squares These results add further weight to the conclusion
on either side of the screen. One of the coloured that synaesthesia is automatically elicited. It also
14 Jamie Ward and Others

Fig. 9 – A non-lateralised and uninformative sound can orient attention to a synaesthetically congruent colour patch in a Posner-
type experiment (SD for synaesthetes and controls across conditions is 141 and 85 respectively).

suggests that synaesthesia appears to use the same similar. Table I summarises what we consider to be
attentional mechanisms as those found in veridical the main similarities and differences between
perception (or an analogous mechanism that operates synaesthetes and controls. First of all, we believe
on the same lines). that both groups use the same cognitive mechanism
to map between the auditory and visual domain.
This mechanism appears to involve mappings
GENERAL DISCUSSION between pitch and lightness. It is conceivable that
different aspects of complex sounds became
In this study we were able to find clear associated with different aspects of vision. For
differences between sound-colour synaesthetes and example, the presence or absence of timbre
controls given the same task. Differences were influences the number of colours and the richness of
found in terms of internal consistency of colour the colours (in terms of the Munsell chroma)
associations, specificity of the colour selection, and observed. Whilst this remains to be fully explored
in terms of the automaticity of colour association there is no evidence, as yet, to suggest fundamental
as measured by Stroop-like interference with differences between the way that synaesthetes and
incongruent sound-colour pairs and attentional others (presumably engaged in imagery) associated
orienting to a congruent sound-colour pair. This is sound and colour. We speculate that the same
the first time that these objective tests have been cognitive mechanism is used by both groups but
conducted on this variety of synaesthesia, although that the mechanism differs in terms of its precision
our results are entirely consistent with previous and automaticity between the groups. For
anecdotal reports in the literature. synaesthetes the mapping between sounds and
Although we were able to find clear differences colour is very precise as is evidenced by their high
between the synaesthetes and controls on these tests, internal consistency and selection of customised
in other ways the two groups were remarkably colours. In contrast, the mapping between sound

TABLE I
Similarities and differences in sound-colour associations between synaesthetes and non-synaestetes

Synaesthetes General Population

Nature of mapping Pitch → lightness Pitch → lightness

Specificity of mapping Precise Noisy
Mode of induction Automatic Strategic
Perceptual experience Conscious, explicit None or implicit
 Sound-colour synaesthesia 15

and colour in non-synaesthetes is noisy. Although all, they may reflect direct hard-wired connections
controls show the same central tendency to map between unimodal auditory and unimodal visual
certain pitches on to certain lightnesss they show no centres that are found in synaesthetes but that are
strong commitment to a precise colour. Moreover, not present in other people (except perhaps in the
we suggest that whereas the synaesthetes evoke this neonatal period, e.g., Baron-Cohen, 1996; Baron-
mechanism automatically, the controls may evoke Cohen et al., 1993; Harrison, 2001). Secondly, they
the mechanism only strategically. Experiment 1 may reflect feedback pathways from bimodal (or
demonstrated that controls can generate broadly multimodal) audiovisual regions to unimodal visual
similar sound-colour associations to synaesthetes, regions of the brain (e.g., Grossenbacher and
but Experiments 2 and 3 established that there is no Lovelace, 2001). These pathways are assumed to be
evidence that controls do so automatically4. We present in us all to some extent (e.g., Calvert et al.,
speculate that this may reflect the amount of 1999; Driver and Spence, 2000; Macaluso et al.,
feedback connections in a model such as that in 2000). We suggest that the fact that non-
Figure 1 above. synaesthetes appear to use the same heuristic for
It is important to note that whilst we consider matching sounds to colours as non-synaesthetes is
objective differences between synaesthetes and more consistent with the latter view than the former.
controls to be crucial, this does not make It might be possible to develop some hybrid theory
phenomenology redundant. For example, one of our that was intermediate between these positions. For
control subjects had an internal consistency example, under the direct linkage account one could
comparable to synaesthetes but did not claim to assume that the pruning between auditory and visual
have synaesthetic experiences (unfortunately, she centres is only partially achieved in non-
wasn’t tested on Stroop) and we would not wish to synaesthetes. However, in the absence of further
diagnose her as having synaesthesia. In our view, empirical evidence we are inclined towards the
reported experiences together with objective tests more parsimonious explanation that sound-colour
comprise a primary source of data that need to be synaesthesia recruits the normal mechanisms of
explained. As such, an explanation in terms of audiovisual perception and attention.
synaesthesia is more parsimonious than an It should be noted that our results are
explanation in terms of learned verbal associations completely neutral with regards to the nature of the
between colour and sound, because the latter but not underlying neural mechanism. For example, a
the former requires an ad hoc account along the failure of neural pruning (or apoptosis) has been
lines of a highly specific delusion pertaining to their put forward as one candidate mechanism by
own experiences5. We do not wish to be glib about proponents of the direct linkage theory (Baron-
the importance of ascertaining that synaesthetic Cohen, 1996; Maurer, 1997). However, the same
experiences are perceptual. This is far from easy mechanism could apply equally well if one were to
(e.g., imagery may use perceptual resources), assume that the pruning failure was between some
although our study on cross-modal synaesthetic cross-modal nexus and unimodal visual areas. In
orienting offers support for a perceptual account. sound-colour synaesthesia that is pharmacologically
However, we do wish to note the importance of induced it is also conceivable that the same route is
considering phenomenological reports alongside used although the neural mechanism is likely to be
objective tests. This situation is not unique to very different. For example, the drug may stimulate
synaesthesia. For example, the interest in blindsight release of neurotransmitters from the cross-modal
comes from the self-reported lack of vision together route and/or impede transmission to the unimodal
with the residual objective abilities, and not from visual pathway (Hartman and Hollister, 1963;
performance on the objective tests alone (e.g., Hollister, 1968).
Cowey, 2004). Similarly, the search for the neural It is interesting to speculate where in the brain
correlates of consciousness is predicated on the the hypothesised cross-modal nexus giving rise to
assumption that conscious reports can serve as this type of synaesthesia might lie. Functional
primary data that require explanation (e.g., Frith et imaging studies have consistently identified the left
al., 1999). Superior Temporal Sulcus (STS) as being involved
As noted earlier, there are two different accounts in the integration of auditory and visual information
of how synaesthetic experiences might arise. First of (for a review see Calvert, 2001). This region of the
brain responds to the presence of (congruent)
auditory and visual stimuli more than would be
4 One other study (Melara, 1989) noted interference when subjects were expected from either modality alone or from the
asked to make light/dark judgments when ignoring low/high pitches
respectively (relative to high/low pitches respectively), and also sum of effects from each modality (i.e., the effects
interference when asked to make low/high pitch judgments when ignoring are super-additive). For example, it responds
light/dark patches respectively (relative to dark/light respectively). This
differs from our experiment in which subjects were asked to respond with a strongly when presented with congruent lip
colour name rather than a lightness judgment, so the level of interference movements and heard speech (Calvert et al., 2000),
may be reduced or eliminated for controls.
5 It could be the case that cross-modal links do become incorporated into when presented with congruent phonemes and
linguistic structures (Osgood, 1960), although the fact that cross-modal
links emerge early (Lewkowicz and Turkewitz, 1980) suggests that the written graphemes (Raij et al., 2001), and also with
effects are not purely language related. non-speech (white noise) and meaningless vision
16 Jamie Ward and Others

(checkerboards) when presented in temporal 1994). It is an empirical question whether the

synchrony (Calvert et al., 2001). One obvious sample of synaesthetes reported here could use
experiment would be to determine how this region their elicited colours in order to deduce the identity
responds, in non-synaesthetes, when presented with of a note so as to acquire a kind of absolute pitch
high/low pitch together with dark/light shades. In (none of our sample report the ability, despite a
this instance, the congruency between vision and few of them being musically competent). A number
audition is not determined spatially (as in lip of conditions would be necessary for this to occur,
reading), temporally (as in Calvert et al., 2001) or including (a) that notes a semi-tone apart have
meaningfully (as in phoneme-letters) but is perceivable differences in colour (b) these
determined by the stimulus dimensions themselves. differences are not modified by the context in
It would then be interesting to determine whether which the note is played (c) the perceptual
this region responds strongly, in sound-colour differences can be committed to long-term memory
synaesthetes, when given unimodal auditory stimuli. and (d) these stored perceptual differences can
In the synaesthetes reported here, the mapping acquire linguistic labels. These are open, but
from sound to colour appears to be determined by testable, questions. At present, all that we wish to
pitch height rather than pitch class. Pitch height note is that pitch is a multi-dimensional attribute
refers to a monotonically varying dimension and this may produce multiple forms of
(related to the ascending order of octaves), whereas synaesthesia – only one of which we have been
pitch class is a circularly varying dimension that able to document in the present study.
describes position within an octave (Shepard, In sum, our conclusion is that the type of
1964). For example, piano notes of 220 Hz and sound-colour synaesthesia reported here is a
440 Hz have different pitch heights but the same genuine phenomenon in which pitch heights map
pitch class because both notes are ‘A’ (or “la”) on on to colours as a function of the lightness. We
the Western musical scale. This octave equivalence suggest that this variety of synaesthesia recruits
is not simply a result of musical training or a normal mechanisms of cross-modal perception and
cultural artefact of this system of music. It is found attention and can therefore be used to speak to
in many different cultures (Nettl, 1956) and may theories of normal cognition. For example, it raises
relate to a 2:1 ratio of pitch heights. Thus, the fact important questions about the relationship between
that 440 Hz (as opposed to, say, 450 Hz) is strategically-generated imagery versus perception.
denoted as ‘A’ is purely cultural whereas the fact The study of synaesthesia may also provide a way
that notes of 220 Hz, 880 Hz and 110 Hz are of disentangling different aspects of musical
considered perceptually similar to 440 Hz is cognition (e.g., the different dimensions underlying
assumed to reflect an underlying property of pitch and timbre perception). Finally, it might also
auditory processing. For some synaesthetes, pitch shed light on the necessary and sufficient neural
class appears to be the strongest determinant of substrates of conscious perceptual experience.
elicited colour (Bernard, 1986; Carroll and These are the challenges for the future.
Greenberg, 1961; Haack and Radocy, 1981;
Langfeld, 1914; Myers, 1915; Peacock, 1985; Acknowledgments. The Leverhulme Trust (Ref:
F/07134/AG).
Riggs and Karwoski, 1934; Rogers, 1987; Vernon,
1930). Thus, all notes corresponding to the pitch
REFERENCES
class of ‘A’ would have the same colour in spite of
differences in pitch height. In some cases, both ARMEL KC and RAMACHANDRAN VS. Acquired synaesthesia in
pitch class and pitch height are important. The case retinitis pigmentosa. Neurocase, 5: 293-296, 1999.
BAHARLOO S, SERVICE SK, RISCH N, GITSCHIER J and FREIMER NB.
reported by Riggs and Karwoski (1934) assigned Familial aggregation absolute pitch. American Journal of
colours to notes such that the same note one octave Human Genetics, 67: 755-758, 2000.
apart had the same colour, but with the higher note BARON-COHEN S. Is there a normal phase of synaesthesia in
development? Psyche, 2: 1996.
being lighter. All the reported cases in which pitch BARON-COHEN S, BURT L, SMITH-LAITTAN F, HARRISON and
class determines the colour of a note also have BOLTON P. Synaesthesia: Prevalence and familiality.
Perception, 25: 1073-1079, 1996.
absolute pitch. Absolute pitch is the ability to BARON-COHN S, HARRISON J, GOLDSTEIN LH and WYKE M.
identify or produce a note without an external Coloured speech perception: Is synaesthesia what happens
reference (Ward, 1999). It is related to early when modularity breaks down? Perception, 22: 419-426, 1993.
BARON-COHEN S, WYKE MA and BINNIE C. Hearing words and
musical training and possibly a genetic seeing colours: An experimental investigation of a case of
predisposition (Baharloo et al., 2000). The parallels synaesthesia. Perception, 16: 761-767, 1987.
with synaesthesia are intriguing. It has been BERNARD JW. Messiaen's synaesthesia: The correspondence
between color and sound structure in his music. Music
suggested that absolute pitch could be an outcome Perception, 4: 41-68, 1986.
of neural pruning (Keenan et al., 2001) and that all CALVERT GA. Crossmodal processing in the human brain: Insights
from functional neuroimaging studies. Cerebral Cortex, 11:
infants are born with the ability (Ward, 1999). It 1110-1123, 2001.
has also been claimed that many people possess a CALVERT GA, BRAMMER MJ, BULLMORE ET, CAMBELL R, IVERSEN
more implicit absolute pitch in which songs can be SD and DAVID AS. Response amplification in sensory-specific
cortices during cross-modal binding. NeuroReport, 10: 2619-
sung in their normal key despite an inability to 2623, 1999.
label the note or key even after training (Levitin, CALVERT GA, CAMBELL R and BRAMMER MJ. Evidence from
 Sound-colour synaesthesia 17

functional magnetic resonance imaging of crossmodal binding translations of sensory dimensions. Psychological Bulletin,
in the human heteromodal cortex. Current Biology, 10: 649- 82: 303-331, 1975.
657, 2000. MARKS LE. Bright sneezes and dark coughs, loud sunlight and soft
CALVERT GA, HANSEN PC, IVERSEN SD and BRAMMER MJ. moonlight. Journal of Experimental Psychology: Human
Detection of audio-visual integration sites in humans by Perception and Performance, 8: 177-193, 1982.
application of electrophysiological criteria to the BOLD MARKS LE, HAMMERAL RJ and BORNSTEAIN MH. Perceiving
effect. NeuroImage, 14: 427-438, 2001. similarity and comprehending metaphor. Monographs of the
CAMPADELLI P, SCHETTINI R and ZUFFI S. A system for the Society for Research in Child Development, 52: 1-93, 1987.
automatic selection of conspicuous color sets for qualitative MASSON DI. Synesthesia and sound spectra. Word, 8: 39-41, 1952.
data display and visual interface design. Journal of Image and MAURER D. Neonatal synaesthesia: Implications for the processing
Graphics, 5: 500-503, 2000. of speech and faces. In S Baron-Cohen and JE Harrison (Eds),
CARROLL JB and GREENBERG JH. Two cases of synaesthesia for Synaesthesia: Classic and Contemporary Readings. Oxford:
color and musical tonality associated with absolute pitch. Blackwell, 1997.
Perceptual and Motor Skills, 13: 48, 1961. MELARA RD. Dimensional interaction between color and pitch.
COWEY A. Fact, artefact and myth about blindsight. Quarterly Journal of Experimental Psychology: Human Perception and
Journal of Experimental Psychology, 57A: 577-609, 2004. Performance, 15: 69-79, 1989.
CUTSFORTH TD. The role of emotion in a synaesthetic subject. MILLS CB, BOTELER EH and OLIVER GK. Digit synaesthesia: A
American Journal of Psychology, 36: 527-543, 1925. case study using a stroop-type test. Cognitive
CYTOWIC RE. Synaesthesia: A Union of the Senses (2nd ed.). Neuropsychology, 16: 181-191, 1999.
Cambridge, MA: MIT Press, 2002. MUDGE EL. The common synaesthesia of music. Journal of
DEHAY C, BULLIER J and KENNEDY H. Transient projections from Applied Psychology, 4: 453-456, 1920.
the fronto-parietal and temporal cortex to areas 17, 18, 19 in MYERS C. Two cases of synesthesia. British Journal of Psychology,
the kitten. Experimental Brain Research, 57: 208-212, 1984. 7: 112-117, 1915.
DRIVER J. Enhancement of selective listening by illusory MYERS CS. A case of synaesthesia. British Journal of Psychology,
mislocation of speech sounds due to lip-reading. Nature, 381: 4: 228-238, 1911.
66-68, 1996. NETTL B. Music in primitive culture. Cambridge: Harvard
DRIVER J and SPENCE C. Multisensory perception: Beyond University Press, 1956.
modularity and convergence. Current Biology, 10: R731- ODGAARD EC, FLOWERS JH and BRADMAN HL. An investigation of
R735, 2000. the cognitive and perceptual dynamics of a colour-digit
FRITH C, PERRY R and LUMER E. The neural correlates of synaesthete. Perception, 28: 651-664, 1999.
conscious experience: An experimental framework. Trends in OSGOOD CE. The cross-cultural generality of visual-verbal
Cognitive Sciences, 3: 105-114, 1999. synesthetic tendencies. Behavioral Science, 5: 146-169, 1960.
GINSBERG L. A case of synaesthesia. American Journal of OSTWALD PF. Color hearing. Archives of General Psychiatry, 11:
Psychology, 34: 582-589, 1923. 40-47, 1964.
GROSSENBACHER PG. Perception and sensory information in PEACOCK K. Synaesthetic perception: Alexander Scriabin’s color
synaesthetic experience. In S Baron-Cohen and JE Harrison hearing. Music Perception, 2: 483-505, 1985.
(Eds), Synaesthesia: Classic and Contemporary Readings. PLOMP R. Pitch of complex tones. Journal of the Acoustical
Oxford: Blackwell, 1997. Society of America, 41: 1526-1533, 1967.
GROSSENBACHER PG and LOVELACE CT. Mechanisms of POSNER MI. Orienting of attention: The VII Sir Frederic Bartlett
synaesthesia: Cognitive and physiological constraints. Trends Lecture. Quarterly Journal of Experimental Psychology, 32:
in Cognitive Sciences, 5: 36-41, 2001. 3-25, 1980.
HAACK PA and RADOCY RE. A case study of a chromesthetic. POSNER MI and COHEN Y. Components of visual orienting. In H
Journal of Research in Music Education, 29: 85-90, 1981. Boumer and DG Bouwhuis (Eds), Attention and Performance
HARRISON J. Synaesthesia: The Strangest Thing. Oxford: Oxford X. Hove: Lawrence Earlbaum Associates, 1984.
University Press, 2001. RAIJ T, UUTELA K, KRAIJ T and HARI R. Audiovisual integration of
HARTMAN AM and HOLLISTER LE. Effect of mescaline, lysergic letters in the human brain. Neuron, 28: 617-628, 2001.
acid diethylamide and psilocybin on color perception. RIGGS LA and KARWOSKI T. Synaesthesia. British Journal of
Psychopharmacolgia, 4: 441-451, 1963. Psychology, 25: 29-41, 1934.
HOLLISTER LE. Chemical psychoses: LSD and related drugs. ROGERS GL. Four cases of pitch-specific chromesthesia in trained
Springfield: Charles C. Thomas, 1968. musicians with absolute pitch. Psychology of Music, 15: 198-
HUBBARD TL. Synesthesia-like mappings of lightness, pitch and 207, 1987.
melodic interval. American Journal of Psychology, 109: 219- SHEPARD RN. Circularity in judgements of relative pitch. Journal
238, 1996. of the Acoustical Society of America, 36: 2345-2353, 1964.
KARWOSKI TF and ODBERT HS. Color-music. Psychological SPENCE C and DRIVER J. Audiovisual links in exogenous covert
Monographs, 50: 1-60, 1938. spatial orienting. Perception and Psychophysics, 59: 1-22,
KEENAN JP, THANGARAJ V, HALPERN AR and SCHLAUG G. Absolute 1997.
pitch and planum temporale. NeuroImage, 14: 1402-1408, SPENCE C and DRIVER J. Attracting attention to the illusory
2001. location of a sound: Reflexive crossmodal orienting and
LANGFELD HS. Note on a case of chromesthesia. Psychological ventriloquism. NeuroReport, 11: 2057-2061, 2000.
Bulletin, 11: 113-114, 1914. VERNON PE. Synaesthesia in music. Psyche, 10: 22-39, 1930.
LEVITIN DJ. Absolute memory for musical pitch: Evidence from VIKE J, JABBARI and MAITLAND CG. Auditory-visual synesthesia:
the production of learned melodies. Perception and Report of a case with intact visual pathways. Archives of
Psychophysics, 56: 414-423, 1994. Neurology, 41: 680-681, 1984.
LEWKIWICZ D and TURKEWITZ G. Cross-modal equivalence in WARD J and SIMMER J. Lexical-gustatory synaesthesia: Linguistic
infancy: Auditory-visual intensity matching. Developmental and conceptual factors. Cognition, 89: 237-261, 2003.
Psychology, 16: 597-607, 1980. WARD WD. Absolute pitch. In D Deutsch (Ed), The Psychology of
MACALUSO E, FRITH CD and DRIVER J. Modulation of human Music (2nd ed.). London: Academic Press, 1999.
visual cortex by crossmodal spatial attention. Science, 289: WHIPPLE GM. Two cases of synaesthesia. American Journal of
1206-1208, 2000. Psychology, 11: 377-404, 1900.
MARKS L. On cross-modal similarity: Auditory-visual interactions WHITCHURCH AK. Synaesthesia in a child of three and a half years.
in speeded discrimination. Journal of Experimental Psycholoy: American Journal of Psychology, 33: 302-303, 1922.
Human Perception and Performance, 13: 384-394, 1987. ZIGLER MJ. Tone shapes: A novel type of synaesthesia. Journal of
MARKS LE. On associations of light and sound: The mediation of General Psychology, 3: 277-286, 1930.
brightness, pitch, and loudness. American Journal of
Psychology, 87: 173-188, 1974. Jamie Ward, Department of Psychology, University College London, Gower Street,
MARKS LE. On coloured-hearing synaesthesia: Cross-modal London WC1E 6BT, UK. e-mail: [email protected]