Orange Is Less Than Green: An Examination of Bidirectionality in Grapheme-Color Synesthesia

Abstract

Grapheme-color synesthetes experience a sense of color when viewing graphemes (e.g., digits and letters). Traditionally, these synesthetic perceptions are considered to be unidirectional, where viewing a grapheme elicits a nonveridical sensation of color, but viewing a color does not induce a reciprocal sense of a grapheme. A growing body of research has emerged that suggests the potential for bidirectional percepts, wherein color facilitates additional grapheme perception. We present here a novel paradigm in which we presented two sets of pure color patches, based on synesthete’s reported colors, side-by-side and asked participants to indicate the color patch with the greater affiliated magnitude. Results indicated that the odds of answering correctly on trials were significantly greater for synesthetes (80.2% accuracy) than nonsynesthetes (52.1% accuracy). These results are aligned with other reports that support the notion of inducing a sense of magnitude from color in synesthetes. These findings challenge the traditional model of synesthesia as a unidirectional phenomenon and have implications of the neuronal communications that underlie perception in general.

Keywords

bidirectionality color grapheme-color synesthesia

Synesthesia is a condition in which the perception of certain stimuli induces concurrent perception of a specific, nonveridical sensation (for a recent review of synesthesia, see Ward, 2013). A synesthete may experience a shape when tasting a particular flavor or experience a tactile sensation in response to hearing a particular word. Sensations such as these and others experienced by synesthetes have been presumed to be strictly unidirectional (Grossenbacher & Lovelace, 2001; Rich & Mattingley, 2002; but see Cytowic, 2002 for a notable exception). For example, hearing particular voice frequencies might induce the perception of red, but viewing red would not necessarily induce a reciprocal frequency (Moos, Simmons, Simner, & Smith, 2013). Similarly, grapheme-color synesthetes are individuals for whom graphemes elicit a sense of color, though colors are not thought to induce numbers (Mills, Boteler, & Oliver, 1999). However, there is a growing body of evidence suggesting that a bidirectional component to grapheme-color synesthesia may exist (e.g., Brugger, Knoch, Mohr, & Gianotti, 2004; Cohen Kadosh & Henik, 2007; Knoch, Gianotti, Mohr, & Brugger, 2005). Further examination of the directionality of synesthetic perceptions is critical not only to theoretical models of synesthesia but potentially to models of perception in typical brains as well; proponents of the disinhibited feedback theory of synesthesia (Grossenbacher & Lovelace, 2001) argue that there are no anatomical differences between synesthetic and nonsynesthetic brains. Although no definitive model has been elected within the field, this position suggests that synesthesia can serve as a unique window through which we can examine models of cognitive mechanisms in general (Cohen Kadosh & Henik, 2007), as elucidating bidirectional information flow in synesthetic brains would suggest the potential for similar information flow outside the realm of synesthesia.

Recent examinations of the assumption of unidirectionality in grapheme-color synesthesia include Brugger et al. (2004), who utilized a paradigm in which a set of colors was presented to participants based on reported colors experienced by synesthetes. Participants were initially trained to respond to half the colors with their left hands and to the other half with their right hands. Participants then repeated the task in a second set of trials with a reversal of the assigned responding hand. Results indicated that while controls showed no difference between the two conditions, synesthetes responded to colors associated with smaller numbers faster with their left hands and to colors associated with larger numbers faster with their right hands. These findings demonstrate that synesthetes can experience a classic SNARC (Spatial-Numerical Association of Response Codes) effect (Dehaene, Bossini, & Giraux, 1993) for synesthetic colors, what the authors term a “SCARC effect” (Spatial-Chromatic Association of Response Codes; Brugger et al., 2004).

In another examination of bidirectional coactivation of numbers from colors, Knoch et al. (2005) tasked participants with a random color generation task, modeled after the Mental Dice Task (Brugger, 1997). In this task, participants were asked to randomly name colors, and results revealed that synesthetes, but not controls, demonstrated a counting bias similar to that observed in a standard random number generating task. Knoch et al. assert that these results provide evidence for the stimulation of relational numeric properties by colors.

In addition, in a study examining number size and synesthetic colors, Cohen Kadosh et al. (2005) employed a modified version of the classic size congruency paradigm in which participants must identify the larger of a pair of numbers. Typically in such tasks, participants respond more quickly to pairs of numbers that have a greater magnitude difference (e.g., 2 and 8) compared to those that have a smaller magnitude difference (e.g., 4 and 5). Cohen Kadosh et al. report that synesthetes’ responses were modulated by the color the numbers were presented in: Pairs with a smaller magnitude difference presented in the colors of numbers with a large magnitude difference (e.g., 4 and 5 presented in colors associated with 2 and 8) resulted in decreased reaction time, suggesting color can influence magnitude perceptions. Results of other studies support Cohen Kadosh et al.’s results that synesthetes’ undergo a bidirectional flow of information for number-color percepts (Cohen Kadosh & Henik, 2006), letter-color percepts (Weiss, Kalckert, & Fink, 2009), and even in a bimodal presentation format (Paffen, van der Smaagt, & Nijboer, 2015). Together, these results provide preliminary support for the assumption that synesthetes undergo a bidirectional information flow.

Although these findings provide some evidence to bidirectional information relating to numbers and synesthetic colors, explicit task requirements to make comparisons between colors based on numerical magnitude properties have not been explored. In this study, we attempt to address this question by employing a novel paradigm in which participants were presented with two sets of adjacent color patches (see Figure 1) and instructed to indicate which of the sets was larger in magnitude. This task was performed under two different conditions in which the presentation time was either limited or unlimited in order to investigate the relationship between exposure to the stimulus and ascertained magnitude. Under both conditions, if color was capable of activating a sense of the inducer stimulus, synesthetes should have performed significantly better than controls, as demonstrated by a higher percentage of correct trials.

Figure 1.

Sample of color patches stimulus for (a) two-digit numbers and (b) three-digit numbers.

Method

Participants

Five grapheme-color synesthetes were recruited via flyers distributed throughout the California State University, Northridge campus. Subsequent age- and gender-matched controls were similarly recruited, though demographic and availability constraints resulted in a varied number of control participants matched to each synesthete. Initially, a sixth synesthete was recruited with five matched controls, though their data had to be excluded from analysis due to post hoc reporting by the synesthete that there were additional digits they felt were indistinguishable despite being separate on the color scale, resulting in an extremely limited number of unique synesthetic percepts (digits 2, 4, and 8). In addition, six control participants were excluded due to incomplete data sets, resulting in a varied number of controls per synesthete. To view demographic information for all participants, see Table 1.

Table 1.

Demographic Characteristics of Synesthetes and Gender- and Age-Matched Controls.

	Synesthete
	Gender	Age	Mean age	n
SYN1	M	22	22.2 ± 1.5	6
SYN2	F	22	19.0 ± 1.0	3
SYN3	F	23	21.0 ± 0.0	3
SYN4	F	30	26.7 ± 3.5	2
SYN5	F	23	22.2 ± 1.0	5

The size of our synesthetic sample (N = 5) was a result of the difficulty in recruiting individuals that exhibit sufficient amounts of unique synesthetic number-colors. Although this number may initially appear to be relatively low, the number of recruited participants is consistent with other examinations of synesthesia. Previous studies have contributed greatly to the understanding of percepts in synestheisa with reported sample sizes that range from case studies of N = 1 (i.e., Cohen Kadosh et al., 2005; Dixon, Smilek, Cudahy, & Merikle, 2000; Gevers, Imbo, Cohen, Fias, & Hartsuiker, 2010) to small groups (e.g., N = 2, Cohen Kadosh & Henik, 2006; N = 3, Brugger et al., 2004) up to larger groups (e.g., N = 13, Niessen, Fink, Schweitzer, Kluender, & Weiss, 2015; N = 15, Mattingley, Rich, Yelland, & Bradshaw, 2001).

All participants were screened by completing the Synesthesia Battery provided by Eagleman, Kagan, Nelson, Sagaram, and Sarma (2007). Synesthetic participants were verified for having grapheme-color synesthesia, whereas control participants reported experiencing no form of synesthesia. All participants gave informed written consent approved by the institutional review board at California State University, Northridge. This study followed the principles of the Declaration of Helsinki.

Preliminary Session

Each synesthete underwent a preliminary session in which they completed an assessment of color photisms using MATLAB software (The MathWorks, Inc., Natick, MA) and a synesthesia toolbox provided by Eagleman et al. (2007). The vector color values reported for each grapheme were then used to construct the color patches for the stimuli, resulting in a unique set of stimuli for each synesthete.

Stimuli

The stimuli consisted of sets of color patches that matched the color photisms each synesthete experienced when presented with a particular grapheme. The color patches corresponded with a preselected set of number combinations, which included all possible two-digit combinations as well as a subset of possible three-digit combinations wherein each possible digit appeared at least once in each possible position. A set of color patches (each 1.149° of visual angle) appeared on the left side of a centrally located crosshair (0.775° of visual angle), with its mirrored representation (e.g., 12 vs. 21) presented on the right side. In addition, each slide had its own inverted version (e.g., 21 + 12 vs. 12 + 21). Synesthetic and nonsynesthetic participants were instructed to indicate which set of colors, left or right, represented the larger number. Because of the mirrored nature of the stimuli, numbers were constrained such that the first and last digits could not be the same number (e.g., 77 or 939); otherwise, the mirror presentation would be indistinguishable from the original. We omitted the digits “0” and “1” as these graphemes correspond visually to the letter graphemes “O” and “l” and could potentially lead to confusion when synesthetes viewed the associated color patch. Consequently, single-digit numbers were not included, as they would require a “0” to be represented in the first of two color patches (e.g., 07). Post hoc, we also omitted digits for which a synesthete experienced a color indistinguishable from that of another digit (e.g., both 7 and 8 being “bluish”), as these color patches would be ambiguous (see Figure 2 for grapheme number sets in perceived colors). This was done upon learning, after data collection, that similar shades of a color could be confusing for some synesthetes.

Figure 2.

Number sets for recruited synesthetes.

Procedure

The experiments were administered with the SuperLab Software (Cedrus, San Pedro, CA) and utilized a unique set of stimuli for each synesthete and their matched controls.

Both synesthetic and nonsynesthetic participants were positioned in front of a monitor (1024 × 769 pixels) and instructed to place their fingers on the “Z” and “/” keys. They were then instructed to indicate which set of colors “felt larger” by means of a button press with their left or right hand. All stimuli appeared twice in random order under each of the following presentation conditions.

Fast presentation session

In the fast presentation session, each stimulus appeared for 150 ms before disappearing, to prevent intentional saccades to the target. The participant was then presented with a crosshair and instructed to indicate which set of color patches represented the number of greater magnitude. The response window was not restricted, and the subsequent stimulus was not presented until the participant had submitted a response.

Slow presentation session

Following the fast presentation session, participants completed a slow presentation session, a procedure identical to the fast presentation with the exception that participants could view the stimulus for an unrestricted amount of time, and that the stimulus disappeared only after a participant responded.

Normalization

Each synesthete was presented with a unique set and number of stimuli based on their associated colors. In addition, due to experimenter error, one synesthete and their matched controls received a truncated set of stimuli. Due to the varied number of presentable stimuli inherent to each set of synesthetes and matched controls, we examined the number of correct responses across all trials for analysis and computed a mixed-effects model. Mixed-effects models are appropriate when there are unequal numbers of trials per subject per condition, as was the case in this study.

Results

Logistic Regression

A mixed-effects binary logistic regression was used to determine the extent to which accuracy rates across trials could be predicted by (a) synesthete status, (b) presentation duration, and (c) the interaction between synesthete status and presentation duration. Synesthete status, presentation duration, and the interaction between the two were entered into a model as fixed effects. A subject variable was entered as a random effect, although the estimated variability for the subject effect was small (see Table 2). A “null” model (predicting accuracy scores with no interaction term) was also computed to test the significance of the interaction term in predicting accuracy scores. Coefficients in each model were tested for significance using Wald’s z test estimates (see Table 2 for model coefficients). The statistical software program R was used to conduct the analyses (R Core Team, 2014), using the lme4 package (Bates, Maechler, Bolker, & Walker, 2015).

Table 2.

Binary Logistic Mixed Model Coefficients and p Values From Wald’s Test Predicting Accuracy Across Trials.

	Model with no interaction term			Model with interaction term
	Coefficients (SE)	p	Model AIC	Coefficients (SE)	p	Model AIC
Fixed effects
Intercept	0.14 (0.13)	.306	8,462.10	0.08 (0.14)	.54	8,429.90
Synesthete status	1.31 (0.29)	<.001		1.82 (0.31)	<.001
Duration	−0.07 (0.05)	.177		0.05 (0.06)	.404
Synesthete status × Duration				−0.92 (0.16)	<.001
Random effect		Variance (SD)		Variance (SD)
Subject (intercept)		0.30 (0.55)		0.32 (0.56)

Note. Blanks indicate term not included in model. AIC = Akaike information criterion; SD = standard deviation; SE = standard error.

As predicted, the odds of answering correctly were significantly greater for synesthetes than nonsynesthetes, z = 5.83, p < . 001. Synesthetes answered 80.2% of all trials correctly, whereas nonsynesthetes answered 52.1% of trials correctly (see Table 3 for number of trials per synesthete). Critically, a Likelihood Ratio Test suggests that there was a significant interaction between synesthete status and presentation duration when compared with a model that contained no interaction term, χ²(1) = 34.18, p < .001. Examining the coefficient for the interaction, the odds of answering correctly are about 40% lower for synesthetes in the fast condition than in the slow condition. The bar graph in Figure 3 confirms this pattern. Synesthetes were more accurate in the slow trials (85.85% correct) than the fast trials (74.59% correct), whereas nonsynesthetes were at about chance levels in the slow (51.50%) and fast (52.65%) trials.¹

Table 3.

Number of Trials per Participant by Presentation Speed.

Participant	Fast	Slow
Syn1	120	120
Syn2	268	268
Syn3	196	196
Syn4	76	76
Syn5	68	68

Note. Synesthetes and matched controls completed the same number of trials.

Figure 3.

Total proportion of correctly answered trials for synesthetes and nonsynesthetes as a function of trial presentation duration. Chance performance falls at 0.5.

Descriptive Data

Although our mixed-effect analyses account for within-subjects error variance, we present descriptive statistics for each synesthete compared to their matched controls as a function of fast or slow trial presentation (Figure 4), aligning with other papers on synesthesia (e.g., Cohen Kadosh et al., 2005). The patterns of frequency data match the mixed-effect analyses, such that overall, synesthetes were more accurate on slow compared to fast trials, whereas the matched controls showed smaller differences in accuracy between fast and slow trials.² The performance of the control subjects was not statistically significantly different from chance (50% accuracy) in the slow (71.67%, z = 1.91, p = .056) or the fast (41.67%, z = 1.08, p = .280) conditions. It should be noted that one synesthete (SYN 5) had a particularly restricted set of stimuli consisting of merely 68 trials, and this limited set may have contributed to observed chance-level performance (51% and 41% on the slow and fast presentation trials, respectively).

Figure 4.

Proportion of correct responses on slow and fast presentation trials for each synesthete (left panels) and their matched controls on gender and age (right panels) with chance performance falling at 0.5.

Discussion

The purpose of this experiment was to investigate whether viewing color stimuli could trigger a sense of numbers, allowing grapheme-color synesthetes to perform better than nonsynesthetes on a forced-choice magnitude task. As is evidenced by the data, in general, synesthetes were more accurate at slow presentations compared to fast presentations of the stimuli, though performances in both conditions were greater than the chance levels observed in nonsynesthetes. The superior performances by synesthetes overall compared to nonsynesthetes suggest that color stimuli are able to involve magnitude information in the absence of additional cues. As noted before, however, one synesthete (SYN5) performed similarly to nonsynesthetes. Although this could be an artifact of their restricted data set, another possible interpretation for this outcome could be that this person does not qualify as a bidirectional synesthete but is instead a unidirectional synesthete. This result would be consistent with suggestions that there may be subclassifications of unidirectional and bidirectional synesthetes (Cohen Kadosh & Henik, 2007).

The present findings are consistent with and add to previous reports by Brugger et al. (2004) and Cohen Kadosh and Henik (2006) that there appears to be a bidirectional flow of information relating to magnitude when viewing colors. These findings further suggest that the observed bidirectional information flow may be a robust effect as evidenced by the similar pattern of outcomes despite differing tasks (e.g., reaction times, magnitude comparison tasks).

Results of at least one other study suggest that as exposure time increases, larger effects of bidirectional information flow are observed. However, Brugger et al. (2004) found this pattern for much shorter exposure times (30 ms) than used in this study (150 ms). Given that the paradigm utilized by Brugger et al. compared reaction times for response hands to a single stimulus, and our paradigm required participants to consider and identify the larger of two sets of colors to measure accuracy, we believe that it is difficult to draw direct conclusions when comparing presentation times between these studies. Aligning with Brugger et al., we observed that longer presentation times produced larger effects of bidirectional information flow. Thus, our findings further support a trend of bidirectional information flow relating to magnitude as observed by Brugger et al. and perhaps may speak to a more robust feature of this information flow as different types of tasks (e.g., a comparison task vs. reaction times to a single stimulus) may take different processing times. Future studies could investigate the magnitude of an observed effect as related to duration of the exposure time.

The difference observed exclusively in synesthetic participants across presentation times suggests that this bidirectional communication is better performed when given sufficient time to process the stimuli. Although several interpretations likely exist, one outstanding interpretation is that the magnitude task demanded a high degree of attentional effort or directed attention. If the magnitude task were automatic, it seems unlikely that increased processing time would improve accuracy. However, as is evidenced by the data, an increase in the presentation duration of stimuli was accompanied by an overall increase in performance in the synesthetes only, suggesting the involvement of an attentional mechanism. It is possible that synesthetes are required to explicitly retrieve the appropriate number during the magnitude task, a retrieval process that demands attentional resources. This finding would be consistent with previous reports of synesthesia being susceptible to modulation by top-down effects (Ramachandran & Hubbard, 2001). However, the potential attentional mechanism underlying the observed pattern of responses is beyond the scope of this article; our focus was exclusively on investigating whether color stimuli could elicit a magnitude response in the absence of additional cues. Further studies investigating the possibility of an attentional mechanism could restrict response time for participants during the decision-making task or include a mask following stimulus presentation to restrict iconic memory.

The results of this study are incongruous with models that propose synesthesia to be unidirectional (Ramachandran & Hubbard, 2003), and the need to consider the underlying neuronal mechanisms for bidirectional percepts has become increasingly evident. Neuroimaging studies have revealed the color-processing area of the brain, hV4, to be responsive to graphemes in grapheme-color synesthetes (Hubbard, Arman, Ramachandran, & Boynton, 2005). In addition, areas in the prefrontal and parietal cortices (for overview, see Nieder & Dehaene, 2009) have been associated with numerical processing, with intraparietal sulcus activity playing a critical role in numerical comparison (Chochon, Cohen, van de Moortele, & Dehaene, 1999) and coding of numerical magnitude (Piazza, Pinel, Le Bihan, & Dehaene, 2007). Recently, Shum et al. (2013) reported areas of the brain that are responsive to visual numerals, located in the inferior temporal gyrus. These may be the areas through which bidirectional percepts are manifested, though future study is warranted as any such speculation is beyond the scope of this article.

In addition, this study proposes a method for examining the potential for color stimuli to elicit bidirectional information flow relating to numbers. In a pilot study, we have utilized this same paradigm replacing numbers with potential words and required participants to identify which set of color patches represented a word (e.g., cat vs. tac). Although preliminary data suggest that some synesthetes are capable of completing this task, the limitations on the number of admissible synesthetes due to incomplete, unique synesthetic alphabets necessitates further study with a greater number of synesthetes.

These results provide methodological and theoretical contributions to the existing literature, demonstrating a novel paradigm that can be utilized for understanding the bidirectional relationship between color stimuli and bidirectional information flow related to magnitude in grapheme-color synesthesia.

Footnotes

Acknowledgements

The authors would like to thank Dr. John Mayberry at the University of the Pacific for his help with selection and interpretation of analyses.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by an internal grant award from the California State University, Northridge.

Notes

References

Bates

Maechler

Bolker

Walker

(2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67, 1–48. doi:10.18637/jss.v067.i01.

Brugger

(1997). Variables that influence the generation of random sequences. An update. Perceptual and Motor Skills, 84, 627–661.

Brugger

Knoch

Mohr

Gianotti

L. R. R.

(2004). Is digit-color synesthesia strictly unidirectional? Preliminary evidence for an implicitly colored number space in three synesthetes. Acta Neuropsychologica, 2, 252–258.

Chochon

Cohen

van de Moortele

P. F.

Dehaene

(1999). Differential contributions of the left and right inferior parietal lobules to number processing. Journal of Cognitive Neuroscience, 11, 617–30. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10601743.

Cohen Kadosh

Henik

(2006). When a line is a number: Color yields magnitude information in a digit-color synesthete. Neuroscience, 137, 3–5. doi:10.1016/j.neuroscience.2005.08.057.

Cohen Kadosh

Henik

(2007). Can synaesthesia research inform cognitive science? Trends in Cognitive Sciences, 11, 177–184. doi:10.1016/j.tics.2007.01.003.

Cohen Kadosh

Sagiv

Linden

D. E. J.

Robertson

L. C.

Elinger

Henik

(2005). When blue is larger than red: Colors influence numerical cognition in synesthesia. Journal of Cognitive Neuroscience, 17, 1766–1773.

Cytowic

R. E.

(2002). Olivier Messiaen (1908-1992). In: Cytowic

R. E.

(ed.) Synesthesia: A union of the senses, 2nd ed. Cambridge, MA: MIT Press, pp. 308–311.

Dehaene

Bossini

Giraux

(1993). The mental representation of parity and number magnitude. Journal of Experimental Psychology: General, 122, 371–396.

10.

Dixon

M. J.

Smilek

Cudahy

Merikle

P. M.

(2000). Five plus two equals yellow. Nature, 406, 365.

11.

Eagleman

D. M.

Kagan

A. D.

Nelson

S. S.

Sagaram

Sarma

A. K.

(2007). A standardized test battery for the study of synesthesia. Journal of Neuroscience Methods, 159, 139–145. doi:10.1016/j.jneumeth.2006.07.012.

12.

Gevers, W., Imbo, I., Cohen Kadosh, R., Fias, W., & Hartsuiker, R. J. (2010). Bidirectonality in synesthesia: Evidence from a multiplication verification task. Experimental Psychology, 57(3), 178–184. doi:10.1027/1618-3169/a000022.

13.

Grossenbacher

P. G.

Lovelace

C. T.

(2001). Mechanisms of synesthesia: Cognitive and physiological constraints. Trends in Cognitive Sciences, 5, 36–41. doi:10.1016/S1364-6613(00)01571-0.

14.

Hubbard

E. M.

Arman

A. C.

Ramachandran

V. S.

Boynton

G. M.

(2005). Individual differences among grapheme-color synesthetes: Brain-behavior correlations. Neuron, 45, 975–985. doi:10.1016/j.neuron.2005.02.008.

15.

Knoch

Gianotti

L. R.

Mohr

Brugger

(2005). Synesthesia: When colors count. Cognitive Brain Research, 25, 372–374. doi:10.1016/j.cogbrainres.2005.05.005.

16.

MATLAB (Version 8.0.0.783) [Computer software]. Natick, MA: MathWorks.

17.

Mattingley

J. B.

Rich

A. N.

Yelland

Bradshaw

J. L.

(2001). Unconscious priming eliminates automatic binding of colour and alphanumeric form in synaesthesia. Nature, 410, 580–582.

18.

Mills

C. B.

Boteler

E. H.

Oliver

G. K.

(1999). Digit synaesthesia: A case study using a Stroop-type test. Cognitive Neuropsychology, 16, 181–191.

19.

Moos, A., Simmons, D., Simner, J., & Smith, R. (2013, September). Color and texture associations in voice-induced synesthesia. Frontiers in Psychology, 4, 568.

20.

Nieder

Dehaene

(2009). Representation of number in the brain. Annual Review of Neuroscience, 32, 185–208. doi:10.1146/annurev.neuro.051508.135550.

21.

Niessen

Fink

G. R.

Schweitzer

Kluender

Weiss

P. H.

(2015). Implicit interactions between number and space in digit-color synesthesia. Cortex, 64, 225–234.

22.

Paffen

C. L. E.

van der Smagt

M. J.

Nijboer

T. C. W.

(2015). Cross-modal, bidirectional priming in grapheme-color synesthesia. Consciousness and Cognition, 33, 325–333.

23.

Piazza

Pinel

Le Bihan

Dehaene

(2007). A magnitude code common to numerosities and number symbols in human intraparietal cortex. Neuron, 53, 293–305. doi:10.1016/j.neuron.2006.11.022.

24.

Ramachandran

V. S.

Hubbard

E. M.

(2001). Synaesthesia—A window into perception, thought and language. Journal of Consciousness Studies, 8, 3–34. Retrieved from http://cbc.ucsd.edu/pdf/Synaesthesia - JCS.pdf.

25.

Ramachandran

V. S.

Hubbard

E. M.

(2003). The phenomenology of synaesthesia. Journal of Consciousness Studies, 10, 49–57.

26.

R Core Team. (2014). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.r-project.org/.

27.

Rich

Mattingley

(2002). Anomalous perception in synaesthesia: A cognitive neuroscience perspective. Nature Reviews Neuroscience, 3, 43–52. doi:10.1038/nrn702.

28.

Shum

Hermes

Foster

B. L.

Dastjerdi

Rangarajan

Winawer

Parvizi

(2013). A brain area for visual numerals. The Journal of Neuroscience, 33, 6709–6715. doi:10.1523/jneurosci.4558-12.2013.

29.

Superlab (Version 4.5.4) [Computer software]. San Pedro, CA: Cedrus.

30.

Ward

(2013). Synesthesia. Annual Review of Psychology, 64, 49–75. doi:10.1146/annurev-psych-113011-143840.

31.

Weiss

P. H.

Kalckert

Fink

G. R.

(2009). Priming letters by colors: Evidence for the bidirectionality of grapheme-color synesthesia. Journal of Cognitive Neuroscience, 21, 2019–2026. doi:10.1162/jocn.2008.21166.