c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Available online at www.sciencedirect.com

Journal homepage: www.elsevier.com/locate/cortex

Research report

What colour does that feel? Tactileevisual mapping and the

development of cross-modality
Vera U. Ludwig a,b,c,* and Julia Simner c,d
a

Department of Psychiatry & Psychotherapy, Charite Universitatsmedizin Berlin, Germany

Berlin School of Mind and Brain, Humboldt-Universitat zu Berlin, Germany
c
Department of Psychology, Humboldt-Universitat zu Berlin, Germany
d
Department of Psychology, University of Edinburgh, UK
b

article info

abstract

Article history:

Humans share implicit preferences for cross-modal mappings (e.g., low pitch sounds are

Received 13 July 2011

preferentially paired with darker colours). Individuals with synaesthesia experience cross-

Reviewed 18 October 2011

modal mappings to a conscious degree (e.g., they may see colours when they hear

Revised 15 January 2012

sounds). The neonatal synaesthesia hypothesis claims that all humans may be born with this

Accepted 11 April 2012

explicit cross-modal perception, which dies out in most people through childhood, leaving

Action editor Sergio Della Sala

only implicit associations in the average adult. Although there is evidence for decreasing

Published online 19 April 2012

cross-modality throughout early infancy, it is unclear whether this decline continues to

take place throughout childhood and adolescence. This large-scale study had two goals.

Keywords:

First, we aimed to establish whether human non-synaesthetes systematically map tactile

Synaesthesia

and visual dimensions e a combination that has rarely been studied. Second, we asked

Cross-modality

whether tactileevisual associations may be more pronounced in younger compared to

Neonatal synaesthesia hypothesis

older participants. 210 participants between the ages of 5e74 years assigned colours to

Touch

tactile stimuli. Smoothness, softness and roundness of stimuli positively correlated with

Colour

luminance of the chosen colour; and smoothness and softness also positively correlated
with chroma. Moreover, tactile sensations were associated with specific colours (e.g.,
softness with pink). There were no age differences for luminance effects. Chroma effects,
however, were found exclusively in children and adolescents. Our findings are consistent
with the neonatal synaesthesia hypothesis which suggests that all humans are born with
strong cross-modal perception which is pruned away or inhibited throughout development. Moreover, the findings suggest that a decline of some forms of cross-modality may
take place over a much longer time span than previously assumed.
2012 Elsevier Ltd. All rights reserved.

1.

Introduction

Humans share implicit cross-modal associations, which give

us inherent preferences for certain sensory combinations over

others (for a review, see Spence, 2011). For example, we match

higher pitch sounds with lighter colours, with smaller and
spikier shapes, and to locations higher in space than low pitch
sounds (Marks et al., 1987; Walker et al., 2010a; Ward et al.,

* Corresponding author. Division of Mind and Brain Research, Department of Psychiatry and Psychotherapy, Charite Universitatsmedizin
Berlin, Campus Mitte, Chariteplatz 1, D-10117 Berlin, Germany.
E-mail address: Vera.Ludwig@charite.de (V.U. Ludwig).
0010-9452/$ e see front matter 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cortex.2012.04.004

1090

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

2006). Recently, it has been shown that even chimpanzees

associate high pitch sounds with high luminance, suggesting
that some correspondences may constitute basic features of
the primate sensory system (Ludwig et al., 2011). Crossmodality has been demonstrated in a small number of
domains, mostly relating to vision and sound, although
humans are also known to map across other dimensions
(across taste and vision; Johnson and Clydesdale, 1982) and
from higher order triggers such as linguistic symbols. For
example, people share preferences for the colours of letters
(e.g., A tends to be red; Simner et al., 2005).
Cross-modal experiences are known to stem back into
childhood. Mondloch and Maurer (2004) asked 2.5e3-year-old
children about the sounds produced by two bouncing balls. All
children believed that the darker ball was making a lowpitched sound, and the lighter ball was making a highpitched sound, indicating a cross-modal mapping between
luminance and pitch, of a type also found in adults. Children
of a similar age also show sensory mappings to linguistic
symbols, by systematically associating certain letters with
certain colours (e.g., X black; Spector and Maurer, 2008,
2011). Children also show the classic bouba-kiki effect
(Ramachandran and Hubbard, 2001); like adults, they describe
spiky shapes with names such as kiki and rounded shapes
with names such as bouba (Maurer et al., 2006). Crossmodality has even been found back in infancy: for example,
3e4-month-old infants appear to associate high pitch with
visuo-spatial height and with visual sharpness (Walker et al.,
2010a). Moreover, 3e4-week-old infants treat certain intensities of white noise as equal to matched lightness intensities
(shown in a cardiac habituation paradigm) and these same
associations are found in adults in cross-modal matching
tasks (Lewkowicz and Turkewitz, 1980).
Maurer (1993) proposed the neonatal synaesthesia hypothesis
(see also Baron-Cohen, 1996; Maurer and Mondloch, 2005)
which suggests that all babies are born with extreme crossmodal connectivity, and so experience a type of explicit
cross-modal perception. This hyper-connectivity is, they
propose, slowly muted over time due to neural pruning (Maurer
and Maurer, 1988) and/or inhibitory processes (Grossenbacher
and Lovelace, 2001; see Spector and Maurer, 2009). This
pruning would leave only weak, implicit cross-modality in the
average adult. In a small number of people, however, the theory
claims that incomplete pruning/inhibition may cause unusually high cross-modal perception into adulthood, a condition
known as synaesthesia (e.g., some synaesthetes might see
colours when hearing sounds; Ward et al., 2006).
One piece of evidence for the neonatal hypothesis is the
apparent similarity in the experiences of synaesthetes and
non-synaesthetes. For example, both synaesthetes and nonsynaesthetes pair higher pitch sounds with lighter colours,
although synaesthetes experience these colours to a perceptual level (Ward et al., 2006). This suggests that synaesthetes
and non-synaesthetes lie on a continuum: their cross-modal
preferences are similar in their mappings, and differ only in
the degree to which they are experienced (e.g., see Harrison
and Baron-Cohen, 1997), and this is a key prediction of the
neonatal synaesthesia hypothesis. However, evidence for the
hypothesis outside synaesthesia has thus far largely been
limited to showing that very young children demonstrate the

type of cross-modality found in adults (see above) and has

tended to be restricted to the domains of vision and audition.
Although the framework does dictate adult-like crossmodality in children, it crucially predicts that children should
experience stronger associations than the average adult.
Recently, Wagner and Dobkins (2011) provided the first
evidence for such a decline of cross-modality with age in
infancy. These authors demonstrated that 2- and 3-month-old
infants show shapeecolour associations that are not present
in 8-month-olds or in adults.
Our current study had two goals. First, we aimed to
establish whether humans share systematic associations
between the tactile and the visual modality. Second, we aimed
to test whether such associations would be present to
different degrees in different age groups. In contrast to the
study in infants by Wagner and Dobkins, we tested older
children (5 years and above). We were interested in the
question of whether some cross-modal correspondences
would be more strongly expressed in older children or
adolescents compared to adults.
Our investigation is focused on touch for several reasons.
First, touchecolour synaesthesia exists (Day, 2005; Simner
and Ludwig, 2012; Ward et al., 2008; Steven and Blakemore,
2004; but see Brugger and Weiss, 2008); this suggests that
touchecolour correspondences should also be experienced by
non-synaesthetes e at least implicitly. Second, there is a small
amount of evidence for implicit touchecolour associations in
adult non-synaesthetes. Ward et al. (2008) showed that 12
adults systematically mapped low colour-luminance to rough
texture and to high pressure against the skin. Tactileevisual
associations of pressure are also shown by Martino and Marks
(2000), whose participants paired high vibrotactile frequencies
with the colour white, and low frequencies with black. In the
present study, we conducted the first large-scale investigation
of touchecolour associations, in both young and old populations. Given the neonatal synaesthesia hypothesis, we
predict that this cross-modality should already be evident at
a young age, and that certain mappings may be felt more
strongly in children.
In our study, child and adult participants between the ages
of 5 and 74 years palpated tactile stimuli constructed along
dimensions ranging from smooth to rough, soft to hard, and
pointed to round, and then indicated colours to match each
tactile sensation. We used tactile gradients because we aimed
to establish linear relations between tactile dimensions and
the basic visual dimensions of luminance (lightness) and
chroma (saturation; as found in the auditoryevisual domain,
e.g., Ward et al., 2006; Hubbard, 1996). We also analysed colour
categories as a third dependent variable to determine if
specific colours (e.g., red, green) were associated with specific
tactile sensations. Finally, we assessed whether tactileevisual
mappings differed between age groups.

2.

Method

2.1.

Participants

We tested 210 non-synaesthetes (116 female; 5e74 years;

M 17.3). Most data collection took place at the National

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Museum of Scotland as part of the Edinburgh International

Science Festival 2008. Some adult participants were recruited
from existing subject pools and tested in the Department of
Psychology. For the age analysis, we split our participants into
three groups: 5e9-year-olds (mean 8, n 87, 51 female),
10e18-year-olds (mean 12, n 68, 30 female), and 19e74year-olds (mean 38.6, n 55, 35 female). Informed
consent was obtained and testing was approved by the local
ethics committee. Participants confirmed they were not
touchecolour synaesthetes with reference to a verbal
description and poster. The poster gave pictorial examples of
different synaesthesias to ensure that both child and adult
participants understood the nature of the condition and could
accurately indicate their own status.

2.2.

Materials

We created 18 tactile stimuli based on three scales, of roughness, hardness, and roundness:
Roughesmooth. Six flat surfaces (23 cm  28 cm) ranged
from rough to smooth. Roughness was quantified by the ISO
grit value of commercially available sandpapers. Low grit
values correspond to a rough surface (fewer, but bigger,
grains) and high values to a smoother surface (more, but
smaller, grains). Surface values were graded exponentially
yielding perceptually approximately equidistant degrees of
rough / smoothness (revealed by piloting): P60 (roughest
stimulus), P120, P240, P600, P1200, and no grits (smoothest
stimulus). Surfaces were regularly replaced to avoid wear.
Hardesoft. Stimuli comprised six cubes (150 cm 
100 cm  75 cm) ranging from hard to soft, each made from
foam covered in identical black material. Piloting again revealed
that the six foam cubes were perceived as approximately equidistant in terms of hardness. The hardness of foam was established from the mean hardness values (in Newton) provided by
the supplier. These values were 270 (hardest), 215, 175, 132.5, 85,
and 40 (softest).
Pointederound. Six wooden 3-dimensional shapes (10 cm
high) ranging from pointed to round were manufactured by

1091

a professional wood turner, following mathematical formulae

that incrementally altered the shape from pointed to round
(see Fig. 1).

2.3.

Procedure

Participants placed their left hand through the opening in

a fabric screen in order to feel 18 individual objects, hidden
from view. Note that tactile acuity is equivalent for the
dominant and non-dominant hand so that it is acceptable that
participants used their left hand to feel the stimuli (Sathian
and Zangaladze, 1996). Before the experiment began, participants were told they would be feeling objects varying from
rough to smooth, hard to soft, and pointed to round, and they
first felt the two extremes of each scale once to establish the
range of sensation. All 18 objects were then presented in one
of 40 pseudo-randomized orders, ensuring that no more than
two objects of the same type (e.g., foam) were presented in
succession. Participants were instructed to feel each object
(without lifting, to avoid a weight confound) and to choose
a colour for each object from an on-screen electronic colour
wheel which seems to fit the way the object feels. Participants predominantly used the tips of all of their fingers to feel
the roughesmooth stimuli. They used their whole hands
(including the palm) to feel the hardesoft and pointederound
stimuli. Participants made their selection using a mouse in
their right hand, and also selected luminance via an on-screen
luminance bar. They were told not to guess the real colour of
the objects, but to match the colour to the sensation experienced when feeling the object, that there was no right or wrong
answer. The experiment lasted 5e10 min.

2.4.

Data preparation

Colour responses were initially recorded as RGB (red, green,

blue) values and were then converted into CIE L*a*b* colour
space, which was established by the commission internationale
de leclairage (CIE). In this perceptually real colour space, the
Euclidian distance between two points reflects their visual

Fig. 1 e Equations (left) used to obtain graphs (top-right) varying from triangle [function f1(x)] to half-circle [function f6(x)],
provided as plans for wood turner to create the wooden 3D shapes used as stimuli (shown bottom-right) for the
pointederound gradient.

1092

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

difference as perceived by a human observer (Tkalcic and Tasic,

2003). Locations were then coded as CIE L*C*h coordinates in
this same colour space, where L* describes the lightness of
colour (our first dependent variable) and C* describes the
chroma of colour (our second dependent variable; equivalent to
saturation). Since h (hue) is an angle running from 0 to 360 ,
this dimension cannot be meaningfully analysed in a standard
way (e.g., see Ward et al., 2006). Therefore, colour choices were
instead categorized by the two authors independently into the
11 irreducible colour terms of English (i.e., red, green; yellow,
blue, orange, pink, purple, brown, black, white, grey; Berlin and
Kay, 1969). This was our third dependent variable. In the first
coding round, Cohens kappa coefficient for inter-rater agreement was .72 ( p < .0001; Cohen, 1960), where values between .61
and .80 reflect substantial agreement (Landis and Koch, 1977). In
a second coding round, including only colours with discrepant
judgements in the first round, kappa was .67 ( p < .0001).
Remaining codings after the second round (7% of all colours)
were resolved by discussion.

2.5.

Statistical analyses

The data for luminance, chroma, and colour categories were

analysed in different ways due to the nature of the variables.
For luminance, we conducted repeated-measures ANOVAs
with the outcome variable luminance and the within-subject
factor tactile scale (from roughesmooth, hardesoft, and
pointederound) with increments on each scale coded from 1
to 6. Degrees of freedom were corrected using GreenhouseeGeisser estimates since the assumption of sphericity
was violated. Effect sizes were calculated for linear contrasts
using the following formula (following Rosenthal et al., 2000;
equation p. 38; interpretation: r .10: small; r .30: medium;
r .50: large; Cohen, 1988).
s
F1; dfR
r
F1; dfR dfR
For chroma, we used Friedmans ANOVA (the nonparametric equivalent of the repeated-measures ANOVA)
because the chroma data were not normally distributed and
could not be normalised using standard transformations. This
was due to a relatively large number of achromatic colours
(black, white, and grey) e in particular for the smoothness
scale. These achromatic colours are a meaningful part of the
data and constitute instances of colours that were chosen to
have the least possible colourfulness. Therefore, we included
all chroma data in our analyses. Since no linear contrasts could
be calculated, we followed up on this analysis by the nonparametric Wilcoxon signed-rank test to compare the chroma
of the extremes of each tactile scale (e.g., the roughest vs the
smoothest stimulus). Effect size was calculated by r Z/N1/2 for
the outcome of this test (Rosenthal, 1991).
To analyse the number of choices for colour categories (e.g.,
red, green), we used Cochrans Q test. Each colour category was
analysed separately, as done in previous research (Rich et al.,
2005; Simner et al., 2005). Cochrans Q test is appropriate for
analysing three or more sets of dependent frequencies or
percentages (Cochran, 1950). There are six frequencies for
each colour for each tactile scale (e.g., 9, 11, 17, 16, 12, and 29

choices for yellow for the stimuli of the roughesmooth scale,

sorted from roughest to smoothest, see Table 1). These
frequencies are related because each subject chooses a colour
for each stimulus of each tactile scale. Therefore, Cochrans
Q test is appropriate here. We hence tested for each colour
whether choices for this colour were unevenly distributed
across the levels of the tactile scales. For example: are choices
for pink unevenly distributed across the levels of the
hardesoft scale? If yes (significant Cochrans Q test), this
shows that the hardesoftness of a stimulus influences the
numbers of choices for pink. We carried out separate
Cochrans Q tests for each of the 11 colours and each of the
three tactile scales. We followed up each Cochrans Q test with
a continuity-corrected McNemars test, which is a suitable
post-hoc test for Cochrans Q test (e.g., see Morris and
Sloutsky, 1998). Using this test, we compared the frequencies
of choices for each colour between the two extremes of each
tactile scale (e.g., the frequency of red choices for the
smoothest stimulus vs the roughest stimulus). The resulting c2
statistics was converted to effect size phi (4) using the formula
4 (c2/N )1/2 (e.g., Osborn, 2006). 4 is interpreted like Pearsons
r (.5: large effect, .3: moderate effect, .1: small effect). We corrected all p-values using HolmeBonferroni correction within
each tactile (i.e., for 11 comparisons; Holm, 1979).

3.

Results

Below, we first establish that humans systematically map

touch and colour, by reporting the tactileevisual associations
generated by all participants irrespective of age for the visual
dimensions of luminance, chroma, and colour categories. We
also show that our luminance and chroma effects are independent of those concerning colour categories. Second, we
assess whether our effects differed in the three age groups
(5e9-year-olds, 10e18-year-olds, and adults).

3.1.

Tactileevisual cross-modality in all participants

3.1.1.

Luminance

Participants systematically chose more luminant (lighter)

colours as the stimulus became smoother (Fig. 2A, top): i.e., for
roughesmooth stimuli, F(4.71, 983.31) 67.835, p < .0001. This
was a linear relationship, shown by polynomial withinsubject contrasts [F(1, 209) 296.833, p < .0001] with a large
effect size (r .77). Participants also chose lighter colours as
the stimulus became softer; i.e., for hardesoft stimuli, F(4.74,
990.29) 47.520, p < .0001 (Fig. 2A, middle). Again, the effect
size for the linear contrast was large (r .67, p < .0001). Finally,
participants systematically chose lighter colours as the stimulus became rounder: i.e., for pointederound stimuli F(4.54,
947.92) 4.774, p < .001 (Fig. 2A, bottom), again a linear relationship with a medium effect size (r .25, p < .001).

3.1.2.

Chroma

There was a highly significant effect of roughesmoothness on

chroma [c2(5) 26.58, p < .0001; Fig. 2B, top]. Chroma values for
the smoothest stimulus were significantly higher
(Mdn 41.57) than for the roughest (Mdn 33.86, Z 2.81,
p < .01), with an effect size in the small to medium range

1093

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Table 1 e Total number of choices for particular colours for each stimulus for all participants (n [ 210) with associated
p-values for Cochrans Q test, as well as p-values and effect sizes for McNemars tests comparing the extremes of each
tactile scale.
Stimulus

Number of choices
Red

Blue

Green

Yellow

Black

White

Brown

Grey

Orange

Pink

Purple

RougheSmooth
1
(Roughest)
2
3
4
5
6
(Smoothest)
Cochrans Q p-value
McNemar p-value
McNemar effect size 4

9
13
10
10
5
7
.47
.80
.02

36
21
39
37
45
43
.02
.46
.05

28
29
41
56
53
38
**
.23
.08

9
11
17
16
12
29
*
*
.22

45
45
26
11
6
15
***
***
.27

1
0
1
3
11
23
***
***
.30

34
38
21
18
18
5
***
***
.32

24
32
24
26
27
18
.33
.41
.06

8
11
15
11
9
9
.63
1.00
.00

4
2
8
8
16
16
**
.01
.18

12
8
8
14
8
7
.48
.36
.06

HardeSoft
1
(Hardest)
2
3
4
5
6
(Softest)
Cochrans Q p-value
McNemar p-value
McNemar effect size 4

24
15
13
14
12
9
.08
.02
.17

63
59
62
53
41
40
.01
.01
.01

26
40
37
42
43
38
.22
.13
.10

2
6
9
11
18
25
***
***
.29

34
26
23
14
14
12
***
**
.22

1
3
1
3
7
7
.03
.08
.12

16
12
17
13
10
6
.19
.04
.14

14
10
7
13
10
7
.36
.17
.09

9
9
6
2
6
10
.25
1.00
.00

7
11
14
22
30
35
***
***
.29

14
19
21
23
19
21
.71
.28
.07

PointedeRound
1
(Pointiest)
2
3
4
5
6
(Roundest)
Cochrans Q p-value
McNemar p-value
McNemar effect size 4

34
31
17
13
15
18
**
.03
.15

31
36
28
36
40
34
.66
.79
.02

41
41
50
58
58
57
.10
.07
.13

25
19
25
20
22
21
.88
.63
.03

18
12
8
7
5
4
*
.01
.19

1
5
1
2
1
4
.16
.37
.06

19
12
23
14
19
22
.27
.71
.03

8
8
8
8
9
5
.93
.58
.04

13
16
24
16
15
18
.38
.42
.06

12
11
15
17
19
10
.40
.82
.02

8
19
11
19
7
17
.03
.11
.11

Note. For each colour separately, Cochrans Q test analyses whether choices are evenly distributed across the six stimuli of the tactile scales.
As a follow-up, for each colour separately, McNemars test analyses whether the frequency of choices for that colour differ between the
extremes of each scale (i.e., 1st vs 6th stimulus). The numbers of choices for each colour per stimulus are printed in normal font, while p-values
and effect sizes are printed in italics. Interpretation of 4: .5: large effect, .3: moderate effect, .1: small effect. ***p < .001, **p < .01, *p < .05 (all
HolmeBonferroni corrected within each tactile scale, i.e., for 11 comparisons). For non-significant effects, the uncorrected p-value is shown.

(r .19). There was also a significant effect of the hardesoft

scale on chroma, [c2(5) 21.99, p < .001; Fig. 2B, middle]. Here,
chroma values for the softest stimulus were significantly
higher (Mdn 55.64) than those for the hardest stimulus
(Mdn 44.10, Z 3.52, p < .001) with a small to medium
effect size (r .17). The pointederound scale did not exert an
effect on chroma [c2(5) 5.1, p .41; Fig. 2B, bottom].

3.1.3.

Colour categories

Table 1 shows the total number of choices for each colour

category for all stimuli. Moreover, for each colour and tactile
scale it shows the p-value for Cochrans Q test, the p-value for
a McNemars test between the two extremes of each tactile scale
and effect size 4. For the roughesmooth scale, Cochrans Q test
was significant for green, yellow, black, white, brown, and pink.
McNemars tests showed that yellow and white were chosen
significantly more frequently for the smoothest stimulus
compared to the roughest stimulus. The same was true for pink
at an uncorrected level (uncorrected p .01). In contrast, black
and brown were chosen significantly more often for the

roughest stimulus compared to the smoothest stimulus. Green

was unevenly distributed across the roughesmooth scale with
no particular directionality (i.e., McNemars test was nonsignificant). For the hardesoft scale, Cochrans Q test was
significant for yellow, black, and pink. Yellow and pink were
chosen significantly more often for the softest stimulus
compared to the hardest stimulus. In contrast, black was chosen
significantly more often for the hardest stimulus than for the
softest stimulus. For the pointederound scale, Cochrans Q test
was significant for red and black. McNemars tests showed that
both red and black were chosen more frequently for the pointiest stimulus than for the roundest stimulus, although this was
only significant at an uncorrected level.

3.1.4. Pale or pink? Luminance/chroma effects within colour

categories
Colours inherently differ in their luminance and chroma. For
example, the average luminance of pink in our study was 62.84
(SD 15.01) while the average luminance of brown was only
40.39 (SD 18.49). Therefore, it is possible that our cross-modal

1094

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Fig. 2 e Tactileevisual correspondences for all participants (n [ 210). Error bars depicted (95% confidence interval). (A) Effects
of smoothness, softness, and roundness (from top to bottom) on CIE lightness. (B) Effects of smoothness, softness, and
roundness (from top to bottom) on CIE chroma.

effects on luminance and chroma (reported in sections 3.1.1 and

3.1.2) might take their roots from the underlying preferences for
colour categories (reported in section 3.1.3). For example, the
overall mapping of softness to high luminance may be nothing
more than the consequence of a mapping of softness to pink.
However, if our luminance and chroma effects can still be found
within individual colour categories, then these effects can be
considered independent of those categories. For example, if
people are more likely to choose pink for soft stimuli than for
hard stimuli, but they also choose lighter pinks for softer stimuli
and darker pinks for harder stimuli, then we would conclude that
softnesseluminance mapping is not a simple artefact from the
softnessepinkness mapping alone.
Here we checked whether our five significant luminance and
chroma effects (smoothnesseluminance, smoothnesse

chroma, softnesseluminance, softnessechroma, and roundnesseluminance) still hold within the 11 colour categories. For
this, we first calculated the average luminance and the average
chroma for each stimulus for each colour category. We then
correlated these luminance and chroma values with the
respective tactile scales (from 1 to 6) for each colour separately,
with corrected alpha levels.
The softness of stimuli still significantly correlated with
luminance within 5 of the 11 colour categories (within blue:
r .98; green: r .96; brown: r .96; grey: r .95; and pink:
r .93; all HolmeBonferroni corrected for 11 comparisons,
overall type I error rate a .05; Holm, 1979). Therefore we can
conclude that there is indeed a softnesseluminance mapping
effect which is independent of the preference of pairing
softness with specific colours. Similarly, a significant

1095

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Table 2 e Effects involving chroma per age group.

Effect

Age group

Friedmans ANOVA
c2-statistic

Roughesmooth on chroma

Hardesoft on chroma

Pointederound on chroma

5e9 years
10e18 years
19 years
5e9 years
10e18 years
19 years
5e9 years
10e18 years
19 years

c (5)
c2(5)
c2(5)
c2(5)
c2(5)
c2(5)
c2(5)
c2(5)
c2(5)

24.17
12.83
2.38
5.73
14.31
5.14
.92
4.51
9.92

Wilcoxon signed-rank test

between extremes of each scale
p

Effect size r

***
.03
.80
.33
*
.40
.97
.48
.08

.36
.10
.04
.22
.33
.17
.04
.08
.20

**
.41
.74
.04
*
.21
.72
.51
.14

Note. ***p < .001, **p < .01, *p < .05 (all HolmeBonferroni corrected). For non-significant effects, the uncorrected p-value is shown. Interpretation
of effect sizes: r .10: small; r .30: medium; r .50: large (Cohen, 1988).

smoothnessecorrelation was found within a colour category

(grey: r .96; HolmeBonferroni corrected for 11 comparisons;
a .05) and this correlation was also evident in a number of
other colour categories, although early significance in these
latter analyses did not survive our extremely stringent
correction (green: r .91; blue: r .89; orange: r .86; red:
r .83, all uncorrected p-values < .05). Hence, overall, the
smoothnesseluminance mapping also appears to be independent of mappings of smoothness to specific colours.
The other three effects, softnessechroma, smoothnessechroma, and roundnesseluminance mappings, were not
significantly present within individual colour categories when
controlling for multiple comparisons. However, a number of
trends in this direction (i.e., uncorrected p-values < .05) with
very high r-values were again seen here: the softnessechroma
correlation was as high as r .89 within brown; the smoothnessechroma correlation was r .91 in white,1 r .86 in purple,
and r .86 in green. Also roundnesseluminance mapping was
still evident as a trend in the blue responses, r .83. Hence,
these latter three effects also show clear trends within colour
categories although they cannot survive our rigorous correction, which had relatively low statistical power (each n 6) and
an alpha corrected by a multiple of 11.
From our significant findings, we can now be confident that
softnesseluminance and smoothnesseluminance mappings
are independent effects from the mapping of softness and
smoothness to specific colours, with additional trends suggesting the same may also be true for softnessechroma and
smoothnessechroma. The weakest evidence for such independence exists for roundnesseluminance mapping. In the
next section, we report our analysis of age effects in tactileevisual mappings.

3.2.

Age group analysis of tactileevisual mappings

In order to assess age effects, different analyses were carried

out for the three variables. For luminance, we carried out
mixed ANOVAs for each tactile scale including age group as
a between-subject factor with three levels (5e9-year-olds,
10e18-year-olds, adults). The within-subject variable was
1

Colours categorised as white still contained some variation in

chroma, albeit relatively close to 0.

again the respective tactile scale (from roughesmooth,

hardesoft, and pointederound) with increments on each scale
coded from 1 to 6. For luminance, there were no significant
interactions of age group with any of the three tactile scales; for
roughesmooth: F(9.40, 973.08) .350, p .96; for hardesoft:
F(9.49, 981.75) .916, p .51; for pointederound: F(9.14,
945.86) 1.532, p .13. This suggests that the luminance
effects were similarly strong in all age groups.
For chroma, a different procedure was chosen because
Friedmans ANOVA does not allow for the inclusion of
between-subject factors. We repeated Friedmans ANOVAs
(as in the main analysis) in the three age groups separately,
controlling family-wise error rate by HolmeBonferroni
correction within each stimulus category (Holm, 1979).
The results are shown in Table 2. The effect of smoothness
on chroma was only present in the youngest age group of
5e9-year-olds with a medium-to-large effect size. A
tendency for this effect existed in 10e18-year-olds (uncorrected p-value < .05 for the Friedmans ANOVA, see
Table 2), which, however, did not survive correction for
multiple comparisons. The effect of softness on chroma
was only significantly present in the 10e18-year-olds with
a medium effect size. There was a tendency for this effect
in 5e9-year-olds (uncorrected p < .05 for the Wilcoxon
signed-rank test in Table 2). In adults there was no indication at all for the effects of smoothness on chroma, or of
softness on chroma. Consistent with the overall analysis,
there was no effect of roundness on chroma in any group.
Note that the effect sizes take into account different
sample sizes of the groups.
The finding that chroma effects were detected in children
but not in adults does not necessarily mean that the effects
were significantly stronger in children compared to adults. To
assess this, we used a standard procedure to compare effect
sizes r between independent samples (Howitt and Cramer,
2008; p. 399; Howell, 2010). The effect sizes r (see Table 2)
were first transformed to Fishers-Zr-values. Afterwards, the
three groups were compared to each other one by one using
the following formula:
Zr1  Zr2
Z r
1
1

N1  3 N2  3

1096

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

We used one-tailed p-values since we had the clear

hypothesis that younger participants would have larger effect
sizes than older participants. Since there were three
comparisons for each tactile scale (5e9-year-olds vs 10e18year-olds, 5e9-year-olds vs adults, 10e18-year-olds vs
adults), p-values need to be below .05/3 .017 to survive
Bonferroni correction. At an uncorrected level only, the effect
size for smoothnessechroma mapping was significantly
larger in the 5e9-year-olds compared to the 10e18-year-olds,
Z 1.674, p .048 (one-tailed); and it was also larger in the
5e9-year-olds compared to the adults, Z 1.891, p .029 (onetailed). The other comparisons between groups did not come
close to significance (all uncorrected one-tailed ps > .15).
Finally, we also repeated our colour category analysis
within each age group separately. Again we corrected for
multiple comparisons within each tactile scale using
HolmeBonferroni correction (for 33 comparisons, i.e., 11
colours  3 age groups). Colour associations per age group and
effect sizes are shown in Table 3. We did not conduct direct
statistical comparisons between groups for colour categories,
since this would not have been meaningful here. Associations
with colour categories were relatively similar across age
groups (see Table 3). An interesting finding was that the
conventional mapping of softness with pink was only found in
adults with a large effect size.

4.

Discussion

This study shows that all people (i.e., not just synaesthetes)
systematically associate tactile sensations with visual dimensions (as has been found in other sensory domains; Marks et al.,
1987; Ward et al., 2006). In cross-modal matching, smoothness,
softness and roundness positively correlated with the luminance of colour; and smoothness and softness also positively
correlated with chroma. Additionally, certain tactile sensations
were associated with specific colours (e.g., roughness with
brown). In our age analysis we found that luminance effects
were similarly strong in all age groups. Chroma effects,
however, were exclusively found in children and adolescents.

This is consistent with claims for innateness in cross-modality,

of the type proposed by the neonatal synaesthesia hypothesis
(e.g., Maurer and Mondloch, 2005). Moreover, our findings
suggest that a decline of some forms of cross-modality does not
only take place during infancy (Wagner and Dobkins, 2011) but
rather over a much longer time span.
Below we discuss the following three issues in more detail:
(4.1) explanations and possible mechanisms for touchecolour
associations in humans (irrespective of age); (4.2) explanations for the developmental progression of touchecolour
associations with age; and (4.3) the relation of our findings to
the condition of touchecolour synaesthesia.

4.1.
Mechanisms for touchecolour associations in
humans
We found that humans systematically associate certain tactile
sensations with certain visual sensations. A subset of our
associations might reflect real-world associations, and therefore indicate learning. For example, we found a significant
association between softness and pinkness, which may stem
from the Western convention of associating pinkness with
infancy and femininity, an effect reinforced in children from
a young age (Pomerleau et al., 1990). However, colour category
effects (e.g., softnessepinkness) tended to be independent of
chroma/luminance effects (e.g., softnesseluminance). For
example, although both pinkness and luminance were associated with softness, our softnesseluminance mappings were
found even within the category of pink suggesting two independent
effects. Softnesseluminance mappings (and other mappings
linked with chroma/luminance) do not appear to reflect realworld associations. For example, there is no clear real-world
link between softness and high luminance, which can be seen
by considering any number of dark natural objects that are soft
rather than hard (e.g., mud, humus, hair, fur). Hence, for many
of our effects, real-world associations are an unsatisfactory
explanation (but see below for one exception mediated by
language).
On a neuronal level, tactileevisual correspondences might
be mediated by communication of somatosensory and visual

Table 3 e Effects involving colour categories per age group.

Tactile scale

Age group

Associations with colour categories

Roughesmooth

5e9 years
10e18 years

Black* (4 .30), white* (4 .24), green (p .003, 4 .09), grey (p .007, 4 .15), pink (p .02, 4 .24)
Brown* (4 .37), black (p .008, 4 .11), yellow (p .008, 4 .29), white (p .03, 4 .22),
pink (p .02, 4 .13)
Black* (4 .35), white* (4 .36), brown (p .008, 4 .37), grey (p .04, 4 .06), pink (p .01, 4 NA)
Blue (p .03, 4 .08)
Black* (4 .33), brown (p .04, 4 .00), yellow (p .02, 4 .24)
Pink* (4 .48), yellow (p .004, 4 .36), black (p .05, 4 .27)
Orange (p .005, 4 .03)
Black (p .004, 4 .21), brown (p .01, 4 .19)
Brown (p .04, 4 .20)

Hardesoft

Pointederound

19 years
5e9 years
10e18 years
19 years
5e9 years
10e18 years
19 years

Note. Shown are all colours with a Cochrans Q test p-value < .05. Marked with * are associations that survive HolmeBonferroni correction within
each tactile scale (i.e., for 33 comparisons; overall a .05). For those that do not survive correction, uncorrected p-values are given. Effect size 4 is
calculated for McNemars tests comparing the two extremes of each tactile scale. In contrast, Cochrans Q test is calculated across all levels of
the tactile scales, explaining possible discrepancies between p and 4. The direction of the association of the colours with tactile scales is not
indicated. NA: McNemars test cannot be carried out since no participant in this group chose pink for the first or last stimulus. Interpretation of
4: .5: large effect, .3: moderate effect, .1: small effect.

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

brain areas at an early stage of processing (Ramachandran and

Hubbard, 2001). To our knowledge, direct connections from
somatosensory regions to colour-specific visual regions have
not yet been identified. However, it is plausible that they exist
since there is increasing evidence for communication between
early sensory areas which were previously believed to be
unimodal. For example, in the monkey brain there are direct
connections originating in visual cortex projecting into
somatosensory cortex, as well as somatosensoryeauditory, and
visuoeauditory projections (e.g., Cappe and Barone, 2005;
Falchier et al., 2002; Rockland and Ojima, 2003). In humans,
neuroimaging and transcranial magnetic stimulation (TMS)
studies have demonstrated the direct involvement of visual
cortex in certain tactile tasks, especially during visual deprivation from blindness or blindfolding (Merabet et al., 2007; Sadato
et al., 2004; Zangaladze et al., 1999).
One may speculate on why humans possess and readily
form touchecolour and other cross-modal correspondences
and whether such associations might serve some function. One
possibility is that cross-modality constitutes a necessary byproduct of the way the brain processes multi-sensory information. This is supported by the finding that also chimpanzees
associate high pitch with high luminance, as do humans
(Ludwig et al., 2011). In humans, cross-modal correspondences
might, however, have a function for language. It has been
shown, that children learn sound symbolic words better (Imai
et al., 2008). Sound symbolic words are words for which the
sound of the word seems to fit its meaning (Hinton et al.,
1994). For example, the word mil sounds more high-pitched
than the word mol. Therefore most people judge mil to better
fit a white or small object than a black or large object (Newman,
1933; Sapir, 1929). Touchecolour associations are, of course,
less evidently linked to language. However, they may nonetheless be a by-product of an overall propensity for cross-modal
matching which might be particularly pronounced in languagelearning children. Touchecolour associations may also represent the basic stages of cross-modality that give way to more
sophisticated visualehaptic matchings in spatial stimulus
information, or may simply be a more general function of
abstract thought. What is clear, however, is that certain stages
of cross-modal mapping are more evident in children than
adults, perhaps suggesting an innateness of cross-sensation
which dies out with age. We discuss this in the next section.

4.2.
Explanations for the development of touchecolour
associations with age
In our age analysis we found that some effects were clearly
present in children and adolescents, but not in adults. First,
smoothnessechroma mapping was only and strongly found
in the youngest age group of 5e9-year-olds. There was a trend
for this effect in the 10e18-year-olds but no effect at all in the
adults. This is exactly the type of development one would
expect for a cross-modal mapping of the type that the
neonatal synaesthesia hypothesis proposes. Second, softnessechroma mapping was only found in 10e18-year-olds old
with a trend in the 5e9-year-olds. Again there was no indication at all for this effect in adults. It has to be noted that
direct comparisons of effect sizes between children and adults
did not survive Bonferroni correction for multiple

1097

comparisons. However, Howell (2010) pointed out that it is

very difficult to find significant differences between r-values
unless sample sizes are very large. At least in the case of
smoothnessechroma mappings, our data are overall still very
indicative for stronger effects in children compared to adults.
One needs to keep in mind that age-comparison studies often
show the reverse pattern, where increased noise in childrens
data tends to arise from distractibility or poor task comprehension. Our findings hence indicate that some forms of
cross-modal correspondences might be more common and
pronounced in children and adolescents than in adults.
Gori et al. (2008) showed that children (8e10 years) do not
yet optimally integrate information from haptic and visual
modalities when perceiving spatial information conveyed by
both senses (i.e., size and orientation). This might seem at odds
with our evidence of early cross-modality in the touchevision
domain. Nonetheless, the type of cross-modal integration
shown by Gori et al. is very different from the low-level crossmodal correspondences found in our own study. It is therefore
possible that when visualehaptic cross-modal integration start
to become optimal after 10 years (Gori et al., 2008) some
inherent low-level synaesthesia-like correspondences weaken.
This may be because higher-order cross-modal processes start
to dominate over, and further attenuate, inherent low-level
correspondences (as suggested here for smoothnessechroma
mappings, for example).
Other effects found in our study (i.e., luminance effects)
were remarkably consistent across age groups. Similar ageresistant or even age-enhanced cross-modality has been
shown by Marks et al. (1987; for pitchebrightness and loudnessebrightness, in matching between sound and light). Why
do these associations not decrease through age, as predicted by
the neonatal synaesthesia hypothesis? One explanation is that
they are simply stronger than associations that do diminish.
Another explanation is that certain associations are strengthened rather than muted during development due to interactions with real-world experiences and learning (Marks et al.,
1987). For example, linguistic conventions might cause or
strengthen a mapping of softness with luminance: soft objects
are less dense, and so lighter in terms of weight. It may be this
linguistic notion of lightness of weight which mediates the
association with lightness of colour (see Walker et al., 2010b).
A limitation of our study is that the adult sample was much
more heterogeneous in terms of the range of ages
(19e74 years) than the children groups (5e9 years,
10e18 years). This is due to the nature of our data collection
(mainly volunteers at a science festival). Future studies should
test equally sized groups of younger and older adults in order
to determine whether a decline of cross-modal mapping may
still take place during adulthood. Moreover, the adult group in
our study was slightly smaller than the two other groups.
However, for the smoothnessechroma and the softnesse
chroma effects there were not even trends present in the
adults. Therefore it is unlikely that these effects would have
been found with a larger sample size.

4.3.

Relation to touchecolour synaesthesia

We found here evidence of decreasing cross-modality with

age, a key prediction of the neonatal synaesthesia hypothesis.

1098

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Another important prediction of that account is that the

cross-modal associations of non-synaesthetes should
resemble those experienced by synaesthetes. We therefore
compared the associations of our non-synaesthetes to those
of our case-study, synaesthete EB (Simner and Ludwig, 2012),
who experiences conscious perceptual colours whenever she
touches objects. Using the same stimuli as in the present
study, we have shown that EBs tactileevisual associations
also rely on the same type of rule system as the nonsynaesthetes described here, systematically linking the
tactile qualities of smoothness and softness with the luminance and chroma of associated colours. This adds further
support for the hypothesis that synaesthesia and cross-modal
associations in non-synaesthetes share the same roots (see
Simner et al., 2005; Ward et al., 2006). Moreover, our developmental data here suggest that that these roots may indeed
lie in extreme cross-modality in early childhood. Hence, the
degree to which adults experience cross-modal correspondences (strongly/consciously in synaesthetes vs weakly/
implicitly/non-perceptually in non-synaesthetes) may
depend on the extent of pruning and/or inhibitory processes
which occur during development (Cohen Kadosh and Henik,
2007; Hubbard and Ramachandran, 2005; Spector and
Maurer, 2009).
Future studies should determine how the phenomenology
of childhood cross-modality resembles the phenomenology of
adult synaesthesia. The present study indicates that children
may sometimes map tactile and visual sensations more
systematically than adults. Do children also experience
touchecolour associations more vividly and consistently than
non-synaesthete adults? In future studies, we might also ask
the functional reasons for a decline in cross-modality. On the
psychological level, it is possible that, with advancing age,
people may rely increasingly on cognitive strategies when
mapping cross-modally, which might in turn override innate
correspondences. On the biological level, future studies must
determine whether pruning, inhibitory processes, or
a combination of these factors contributes to this decline. We
finish with a methodological recommendation for those
researchers who are mainly interested in colour categories
rather than correlations with the basic visual dimensions of
luminance and chroma. In the present study, we have established the nature of tactileevisual interactions in a very
detailed way using tactile gradients (six stimuli per tactile
scale). This design is ideal to describe associations between
tactile dimensions and basic visual dimensions (luminance
and chroma, e.g., see Ward et al., 2006). However, it renders
the analysis of colour categories somewhat complex.
Researchers interested in colour categories should therefore
opt for simple stimuli sets, for example by using just two
stimuli per tactile scale (e.g., one rough and one smooth).

5.

Conclusion

To conclude, our large-scale study shows that people make

systematic associations between touch and colour. Moreover,
some effects were only present in the younger age groups.
This is the first evidence that cross-modality might be more
pronounced in older children and adolescents compared to

adults. It is consistent with a proposal that cross-modal

hyper-connectivity in the infantile brain gives rise to
extreme cross-modality, and that this connectivity is subsequently slowly muted during normal ageing to leave only
weaker, implicit cross-modality in the average adult.

Acknowledgements
V.L. was supported by a Huygens Scholarship Programme
award by the Dutch Ministry of Education, Culture and
Science. We thank Sean Roberts for writing the software
programme used in this study. We are also grateful to our
anonymous reviewers for their helpful comments.

references

Baron-Cohen S. Is there a normal phase of synaesthesia in

development? Psyche, 2(27) 1996.
Berlin B and Kay P. Basic Colour Terms: Their Universality and
Evolution. Berkeley, CA: University of California Press, 1969.
Brugger P and Weiss PH. Dermo-optical perception: The nonsynesthetic palpability of colors a comment on Larner (2006).
Journal of the History of the Neurosciences, 17(2): 253e255, 2008.
Cappe C and Barone P. Heteromodal connections supporting
multisensory integration at low levels of cortical processing in
the monkey. European Journal of Neuroscience, 22(11):
2886e2902, 2005.
Cochran WG. The comparison of percentages in matched
samples. Biometrika, 37(3/4): 256e266, 1950.
Cohen J. A coefficient of agreement for nominal scales. Educational
and Psychological Measurement, 20(1): 37e46, 1960.
Cohen J. Statistical Power Analysis for the Behavioral Sciences. New
York: Academic Press, 1988.
Cohen Kadosh R and Henik A. Can synaesthesia research inform
cognitive science? Trends in Cognitive Sciences, 11(4): 177e184,
2007.
Day S. Some demographic and socio-cultural aspects of
synesthesia. In Robertson LC and Sagiv N (Eds), Synesthesia:
Perspectives from Cognitive Neuroscience. New York: Oxford
University Press, 2005: 11e33.
Falchier A, Clavagnier S, Barone P, and Kennedy H. Anatomical
evidence of multimodal integration in primate striate cortex.
The Journal of Neuroscience, 22(13): 5749e5759, 2002.
Gori M, Del Viva M, Sandini G, and Burr DC. Young children do not
integrate visual and haptic form information. Current Biology,
18(9): 694e698, 2008.
Grossenbacher P and Lovelace G. Mechanisms of synesthesia:
Cognitive and physiological constraints. Trends in Cognitive
Sciences, 5(1): 36e41, 2001.
Harrison JE and Baron-Cohen S. Synesthesia: A review of
psychological theories. In Baron-Cohen S and Harrison JE
(Eds), Synesthesia: Classic and Contemporary Readings. Oxford:
Blackwell, 1997: 109e122.
Hinton L, Nichols J, and Ohala JJ. Sound Symbolism. Cambridge, UK:
Cambridge University Press, 1994.
Holm S. A simple sequentially rejective multiple test procedure.
Scandinavian Journal of Statistics, 6(2): 65e70, 1979.
Howell DC. Statistical Methods for Psychology. Belmont, CA:
Cengage Wadsworth, 2010.
Howitt D and Cramer D. Introduction to Statistics in Psychology.
Essex: Pearson Education Limited, 2008.

c o r t e x 4 9 ( 2 0 1 3 ) 1 0 8 9 e1 0 9 9

Hubbard EM and Ramachandran VS. Neurocognitive mechanisms

of synesthesia. Neuron, 48(3): 509e520, 2005.
Hubbard TL. Synesthesia-like mappings of lightness, pitch, and
melodic interval. American Journal of Psychology, 109(2):
219e238, 1996.
Imai M, Kita S, Nagumo M, and Okada H. Sound symbolism
facilitates early verb learning. Cognition, 109(1): 54e65, 2008.
Johnson J and Clydesdale FM. Perceived sweetness and redness in
colored sucrose solutions. Journal of Food Science, 47(3):
747e752, 1982.
Landis JR and Koch GG. The measurement of observer agreement
for categorical data. Biometrics, 33(1): 159e174, 1977.
Lewkowicz DJ and Turkewitz G. Cross-modal equivalence in early
infancy: Auditory-visual intensity matching. Developmental
Psychology, 16(6): 597e607, 1980.
Ludwig VU*, Adachi I*, and Matsuzawa T. Visuoauditory
mappings between high luminance and high pitch are shared
by chimpanzees (pan troglodytes) and humans. Proceedings of
the National Academy of Sciences of the United States of America,
108(51): 20661e20665 [*equal contribution], 2011.
Marks LE, Hammeal RJ, and Bornstein MH. Perceiving similarity
and comprehending metaphor. Monographs of the Society for
Research in Child Development, 52(1): 1e102, 1987.
Martino G and Marks LE. Cross-modal interaction between vision
and touch: The role of synesthetic correspondence. Perception,
29(6): 745e754, 2000.
Maurer D. Neonatal synesthesia: Implications for the processing
of speech and faces. In de Boysson-Bardies B, de Schonen S,
Jusczyk P, McNeilage P, and Morton J (Eds), Developmental
Neurocognition: Speech and Face Processing in the First Year of Life.
Dordrecht: Kluwer Academic Publishers, 1993: 109e124.
Maurer D and Maurer C. The World of the Newborn. New York: Basic
Books, 1988.
Maurer D and Mondloch CJ. Neonatal synesthesia: A reevaluation.
In Robertson LC and Sagiv N (Eds), Synesthesia: Perspectives from
Cognitive Neuroscience. New York: Oxford University Press,
2005: 193e213.
Maurer D, Pathman T, and Mondloch CJ. The shape of boubas:
Sound-shape correspondences in toddlers and adults.
Developmental Science, 9(3): 316e322, 2006.
Merabet LB, Swisher JD, Mcmains SA, Halko MA, Amedi A,
Pascual-Leone A, et al. Combined activation and deactivation
of visual cortex during tactile sensory processing. Journal of
Neurophysiology, 97(2): 1633e1641, 2007.
Mondloch CJ and Maurer D. Do small white balls squeak? Pitchobject correspondences in young children. Cognitive, Affective
& Behavioral Neuroscience, 4(2): 133e136, 2004.
Morris BJ and Sloutsky VM. Developmental differences in young
childrens solutions of logical vs. empirical problems. In:
Proceedings of the Twenty First Annual Conference of the
Cognitive Science Society. Mahwah, NJ: Erlbaum, 1998:
432e437.
Newman SS. Further experiments in phonetic symbolism. The
American Journal of Psychology, 45(1): 53e75, 1933.
Osborn CE. Statistical Applications for Health Information Management.
Sudbury, MA: Jones and Bartlett Publishers, Inc., 2006.
Pomerleau A, Bolduc D, Malcuit G, and Cossette L. Pink or blue:
Environmental gender stereotypes in the first two years of life.
Sex Roles, 22(5e6): 359e367, 1990.
Ramachandran VS and Hubbard EM. Synaesthesia e A window
into perception, thought and language. Journal of Consciousness
Studies, 8(12): 3e34, 2001.

1099

Rich AN, Bradshaw JL, and Mattingley JB. A systematic, largescale study of synaesthesia: Implications for the role of early
experience in lexical-colour associations. Cognition, 98(1):
53e84, 2005.
Rockland KS and Ojima H. Multisensory convergence in calcarine
visual areas in macaque monkey. International Journal of
Psychophysiology, 50(1e2): 19e26, 2003.
Rosenthal R. Meta-Analytic Procedures for Social Research. Newbury
Park, CA: Sage, 1991.
Rosenthal R, Rosnow RL, and Rubin DB. Contrasts and Effect Sizes in
Behavioural Research: A Correlational Approach. Cambridge:
Cambridge University Press, 2000.
Sadato N, Okada T, Kubota K, and Yonekura Y. Tactile
discrimination activates the visual cortex of the recently blind
naive to braille: A functional magnetic resonance imaging
study in humans. Neuroscience Letters, 359(1e2): 49e52, 2004.
Sapir E. A study in phonetic symbolism. Journal of Experimental
Psychology, 12(3): 225e239, 1929.
Sathian K and Zangaladze A. Tactile spatial acuity at the human
fingertip and lip: Bilateral symmetry and interdigit variability.
Neurology, 46(5): 1464e1466, 1996.
Simner J and Ludwig VU. The color of touch: A case of tactilevisual synaesthesia. Neurocase, 18(2): 167e180, 2012.
Simner J, Ward J, Lanz M, Jansari A, Noonan K, Glover L, et al.
Non-random associations of graphemes to colours in
synaesthetic and non-synaesthetic populations. Cognitive
Neuropsychology, 22(8): 1069e1085, 2005.
Spector F and Maurer D. The colour of os: Naturally biased
associations between shape and colour. Perception, 37(6):
841e847, 2008.
Spector F and Maurer D. Synesthesia: A new approach to
understanding the development of perception. Developmental
Psychology, 45(1): 175e189, 2009.
Spector F and Maurer D. The colors of the alphabet: Naturallybiased associations between shape and color. Journal of
Experimental Psychology: Human Perception and Performance,
37(2): 484e495, 2011.
Spence C. Crossmodal correspondences: A tutorial review.
Attention, Perception, & Psychophysics, 73(4): 971e995, 2011.
Steven MS and Blakemore C. Visual synaesthesia in the blind.
Perception, 33(7): 855e868, 2004.
Tkalcic M and Tasic JF. Colour spaces: Perceptual, historical and
applicational background. EUROCON 2003. Computer as a tool.
The IEEE Region, 8(1): 304e308, 2003.
Wagner K and Dobkins KR. Synaesthetic associations decrease
during infancy. Psychological Science, 22(8): 1067e1072, 2011.
Walker P, Bremner JG, Mason U, Spring J, Mattock K, Slater A,
et al. Preverbal infants sensitivity to synaesthetic crossmodality correspondences. Psychological Science, 21(1): 21e25,
2010a.
Walker P, Francis BJ, and Walker L. The brightness-weight
illusion: Darker objects look heavier but feel lighter.
Experimental Psychology, 57(6): 462e469, 2010b.
Ward J, Banissy MJ, and Jonas CN. Haptic perception and
synaesthesia. In Grunewald M (Ed), Human Haptic Perception.
Basel: Birkhauser, 2008: 259e265.
Ward J, Huckstep B, and Tsakanikos E. Sound-colour
synaesthesia: To what extent does it use cross-modal
mechanisms common to us all? Cortex, 42(2): 264e280, 2006.
Zangaladze A, Epstein CM, Grafton ST, and Sathian K.
Involvement of visual cortex in tactile discrimination of
orientation. Nature, 401(6753): 587e590, 1999.