Contemporary issues in synaesthesia: an overview and a meta-analysis Literature thesis

(1)

Contemporary issues in synaesthesia: an overview and a

meta-analysis

Literature thesis

Author: Anja Pahor

Student number: 6249515

Supervisor: Olympia Colizoli

Co-assessor: Romke Rouw

Date: 11.06.2012

Msc in Brain Cognitive Sciences

Cognitive Neuroscience track

University of Amsterdam

(2)

Introduction

What is synaesthesia?

Synaesthesia is a rare condition defined by its subjective nature in which a (sensory or conceptual) stimulus consistently and automatically evokes highly specific and physically unrelated experiences in the same sensory modality or in other modalities. The stimulus that triggers the synaesthetic experience has been termed the inducer, whereas the synaesthetic experience is known as the concurrent (Grossenbacher and Lovelace, 2001). For example, in auditory-visual synaesthesia different sounds elicit various visual experiences, such as colours, textures, and shapes (Goller et al., 2008). One of the most common types of synaesthesia is grapheme-colour synaesthesia, in which a given set of letters elicits specific colours that are constant over time (Ward et al., 2007). Each synaesthete has a unique set of concurrents that are mapped to inducers. For instance, for one synaesthete the letter “A” written in black ink is always perceived as “red”, whereas for someone else it is always perceived as “light blue”.

Congenital synaesthesia is typically experienced since early childhood after which it remains constant throughout a person’s life, given that no neurological damage occurs. It is prevalent in approximately 4% of the population in an equal number of men and women (Simner, Mulvenna, Sagiv et al., 2006). Congenital synaesthesia runs in families and there is some evidence that it might be inherited by an x-linked dominant trait (Baron-Cohen et al., 1996; Smilek et al., 2002; Ward and Simner, 2005). For example, Barnett and colleagues (2008) reported that in a sample of 53 congenital synaesthetes, 42% had a first-degree relative with synaesthesia and that different types of synaesthesia can be found within families. The focus of this paper will be on congenital synaesthesia, however it should be mentioned that other forms exist, such as acquired and narcotic–induced synaesthesia. The former occurs in adulthood as a result of brain damage (Beauchamp and Ro, 2008) or sensory deafferentation (Armel and Ramachandran, 1999), while the latter occurs temporarily due to consumption of psychedelic drugs (Afra, Funke, and Matsuo, 2009; Rogowska, 2011).

People with congenital synaesthesia mostly report that their additional sensations do not interfere with their everyday lives and some even describe them as pleasant (Day, 2005). Moreover, some experts have suggested that synaesthesia is linked to creativity (Sitton and

(3)

Pierce, 2004) and that it occurs more often in art students than in controls (Rothen and Meier, 2010). However, Ward and colleagues reported no direct link between synaesthesia and psychometric measures of creativity in a sample of 83 synaesthetes and 119 controls (Ward, Thompson-Lake, and Kaminski, 2008). Grapheme-colour synaesthesia has also been linked to developmental learning difficulties. For example, Green and Goswami (2008) demonstrated that children with grapheme-colour synaesthesia experience difficulties in numerical tasks when digits are presented in colours that are incongruent to their synaesthetic colours. In sum, the debate about relative positive and negative aspects of synaesthesia is complicated by its idiosyncratic nature.

To date there have been more than 60 types of synaesthesia recorded (Day, 2011), including many that are not purely perceptual such as grapheme personification (letters have specific personalities), emotion->smell, and mirror->touch (watching someone else being touched evokes tactile sensations on the watcher’s skin, Banissy and Ward, 2007). Among the most common types are number-form synaesthesia, day-colour synaesthesia, and grapheme-colour synaesthesia (Sagiv et al., 2006; Simner et al., 2006). Furthermore, it is likely that if a person has one type of synaesthesia he or she will also have another type of synaesthesia (Simner et al., 2006). This was also explored by Novich, Cheng, and Eagleman (2011) in a large scale analysis of subgroups in synaesthesia. The authors converged several types of analyses to show that there are five distinct clusters of synaesthesia forms: coloured sequence, coloured music, non-visual sequela, spatial sequence, and coloured sensation. Interestingly, they found that if a person has more than one type of synaesthesia, it is highly likely that the types belong to the same cluster rather than to different clusters.

Individual differences in synaesthesia

Even within the same type of synaesthesia there are a lot of individual differences that can account for the fact that some subjects perform better than others on synaesthetic perceptual and cognitive tasks (Hubbard et al., 2005). For instance, there is a lot of variation in self-reported location in space at which individuals experience their synaesthesia. Synaesthetes are thus often given questionnaires to determine whether they are “projectors”, who describe their synaesthetic concurrent as being “in external space”, or “associators”, who describe their concurrents as being present “in the mind’s eye” (Dixon, Smilek, and Merikle, 2004; Ward et al., 2007). An alternative explanation for individual

(4)

differences in synaesthesia is the higher-lower distinction, which focuses on differences in inducers rather than concurrents. In “lower” synaesthetes the synaesthetic experience is triggered at a perceptual level, whereas in “higher” synaesthetes it is evoked at a higher cognitive level (Ramachandran and Hubbard, 2001b). There is an abundant amount of studies that demonstrate the perceptual nature of synaesthesia (Palmeri et al., 2002; Weiss, Zilles, and Fink, 2005; Barnett et al., 2008). Likewise, there are many studies that show evidence in favour of conceptual origins of synaesthetic experiences. For example, Ward and Sagiv (2007) reported a case of “higher” synaesthesia in a subject that perceived colours when viewing numbers. Crucially, different physical presentations of a given number (digit, dice pattern, finger counting) elicited the same synaesthetic colour. The authors concluded that for this subject, synaesthetic colours arise from conceptual levels of representation. In lexical-gustatory synaesthesia, certain speech sounds consistently elicit specific tastes. It has been suggested that the development of this type synaesthesia is influenced by language and conceptual factors (Ward and Simner, 2003). Nevertheless, this does not necessarily imply that the taste experiences are triggered at a purely semantic level. Another type of synaesthesia that has been used to illustrate the conceptual nature of certain inducers is time-space synaesthesia, in which people associate time units (i.e. days, weeks, months) with specific spatial locations (Smilek et al., 2007). The reason why this type of synaesthesia is considered “higher” is because time units are examples of ordinal sequences, which belong to semantic-level properties (Ward et al., 2007).

This paper will focus on three aspects of synaesthesia that have divided experts for years. First, I will talk about the definition of synaesthesia and the three major issues that are associated with it. Secondly, I will evaluate existing neurobiological models of word-colour synaesthesia in order to determine which model can provide the most likely explanation of synaesthetic experiences. Thirdly, I will present the findings of a meta-analysis of grapheme-colour synaesthesia studies in which I focused on i) whether studies reported which modality (auditory, visual, or both) produces the synaesthetic experience and ii) what consistency criteria were used as inclusion criteria. By focusing on these three important topics rather than one I will present a holistic view of synaesthesia, which will include guidelines for defining synaesthesia, the best existing neurobiological model of word-colour

(5)

synaesthesia, and insight into methodological differences between studies on grapheme-colour synaesthesia, the most widely studied type of synaesthesia.

Defining synaesthesia

Many researchers recruit their subjects through websites that contain (1) a definition of synaesthesia, which might vary between research groups, and (2) a set of online questionnaires, which might be biased to the prototype of the “ideal synaesthete”. Below are some examples of definitions obtained from synaesthesia research websites.

“People with synaesthesia experience unusual sensations (e.g., of colour, of taste) when doing things that wouldn't usually trigger those sensations for non-synaesthetic people. In some cases, this means that a synaesthete might experience a sensation in one of the 5 senses (hearing, vision, taste, touch, smell) that is triggered by a different sense (e.g., sounds trigger tastes, smells trigger colours etc.).” Synaesthesia, University of Essex. Retreived May 7, 2012, from https://www.survey.bris.ac.uk/sussex/syn

“Synesthesia (or synaesthesia) is loosely defined as "senses coming together," which is just a translation of the Greek (etymology: syn - together, esthesia from aesthesis - sensation). At its simplest level, synesthesia means that when a certain sense or part of a sense is activated, another unrelated sense or part of a sense is activated concurrently. For example, when someone hears a sound, he or she immediately sees a color or shape in his or her "mind's eye." The Synesthesia Project. Boston University. Retreived May 7, 2012, from

http://www.bu.edu/synesthesia/

“Synaesthesia (American spelling: synesthesia) comes from the Greek 'syn' (together) + 'aesthesis' (perception), and describes a joining together of sensations that are normally experienced separately. Synaesthetes might experience colours or tastes when they read words or hear sounds, while others may experience any combination of tastes, smells, shapes, colours or touches.” Synesthesia Research Group. University of Edinburgh. Retreived May 7, 2012, from http://www.syn.psy.ed.ac.uk/

It should be noted that the last two websites also provide information about the developmental, automatic, and familial nature of synaesthesia.

Nature of synaesthetic associations

From the sample of definitions presented above we can see that websites tend to use a perceptual definition of synaesthesia. Indeed, several studies have demonstrated that synaesthetic associations have verifiable perceptual consequences (Hubbard and Ramachandran, 2005). For example, Ramachandran and Hubbard (2001) conducted an experiment in which synaesthetes and non-synaesthetes performed on an achromatic

(6)

crowding task. The task takes advantage of the fact that it is difficult to identify a target grapheme in peripheral vision that is surrounded by other, distracting graphemes. For non-synaesthetes the task becomes easier if the target is presented in a different colour than the distracting graphemes. Results showed that synaesthetes performed better than controls if the synaesthetic colours for targets and distracting graphemes differed. The authors interpreted the results as evidence for the claim that synaesthetic colours arise at an early perceptual stage. This was also supported by Palmeri and collegues (2002), who administered a visual search task to synaesthetes and non-syanesthetes. Subjects were asked to locate a target grapheme among a group of achromatic distracters. Synaesthetes were faster at rejecting distracters based on their synaesthetic colours than non-synaesthetes, which could reject them solely based on their identity.

However, some synaesthetes show no advantage in visual search over controls (Edquist et al., 2006), or show advantage only when the target is within focus of attention (Laeng, Svartdal, and Oelmann, 2004). Moreover, recent studies have reported no differences in performance between synaesthetes and non-synaesthetes on visual search tasks, suggesting that synaesthetic colours are likely to arise at a cognitive rather than a perceptual level (Sagiv, Heer, and Robertson, 2006; Gheri, Chopping, and Morgan, 2008). Moving beyond the visual search task, there is an abundance of studies that demonstrate the conceptual nature of synaesthetic experiences (War and Simner, 2003; Kadosh and Henik, 2006; Jansari, Spiller, and Redfern, 2006; Ward and Sagiv, 2007; Simner and Haywood, 2009). This brings about the question why do the definitions in our sample only describe the perceptual nature of synaesthesia? While such a description might be accurate for a subset of synaesthetes, some synaesthetes that have cognitive inducers or concurrents might be discouraged from registering for further online tests, believing that they are not adequate candidates. In fact one large-scale prevalence study demonstrated that the most common type of synaesthesia is coloured days (Simner et al., 2006), which deviates from the typical “merging of senses” definition. Another very common type of synaesthesia that does not fit this definition is grapheme-colour synaesthesia, since it occurs within just one modality - the visual modality. Perceptual definitions of synaesthesia also pose a great problem for less common types of synaesthesia, such as emotion->smell, personality->colour, mirror touch, and object personification. Experts are undoubtedly aware of both perceptual and cognitive

(7)

explanations of synaesthesia, however the general public may not be, thus it is important to present it as accurately as possible, especially when recruiting subjects.

Consistency

The most widely used diagnostic tool for synaesthesia is to test the consistency of synaesthetic experiences over time: synaesthetes are presented with a number of known inducers and are asked to report their synaesthetic concurrents. The test is repeated at least once after a substantial amount of time (Baron-Cohen, Wyke, and Binnie, 1987). Control subjects can also participate by reporting free associations to the stimuli used. Both groups are then compared based on the internal consistency of their reports (Ward, Simner, and Auyeung, 2005). Synaesthetes are more accurate in their reports than controls, as demonstrated by Sagiv and colleagues (2006), who showed that a sample of 100 grapheme-colour synaesthetes reporting grapheme-colour associations were 91.4% consistent over time, while the control sample was 33.4% consistent over time. Similarly, a group of 14 lexical-gustatory synaesthetes that were also included in the study were more consistent (86.2%) than their control group (31.4%).

Many researchers use The Synaesthesia Battery developed by dr. David Eagleman and colleagues (2007). It comprises of a freely accessible online collection of tests that enables standardised testing of synaesthesia. Anyone can take the battery – it starts with a questionnaire in which a person must indicate what type(s) of synaesthesia he or she has. There is also a textbox in which a person can write about his type of synaesthesia if it is not on the list. Based on the results of the questionnaire, the synaesthete is guided to a range of tests. For example, in a grapheme-colour consistency test the synaesthete views a grapheme and then chooses a colour that best matches his synaesthetic colour by navigating a 16.7 million colour palette. There is also a speeded congruency test that provides an additional way of distinguishing synaesthetes from non-synaesthetes. A great feature of the battery is that it also includes tests for other types of synaesthesia, such as time->colour and motion->sound synaesthesia. Moreover, standardised data can be pooled and compared across synaesthesia research groups. Recruiting websites invite people who think they might have synaesthesia to take either The Synaesthesia Battery or personally developed tests and questionnaires. Prior to conducting research, consistency tests are administered and only subjects that perform above a certain threshold are selected. As Simner (2012) has pointed

(8)

out in a recent article, the consistency criterion might be well suited to a subset of synaesthetes because research has been biased in selecting only consistent synaesthetes.

There are two major issues that are related to the consistency debate. Firstly, excluding people who report synaesthetic experiences, but fall in the middle range of 30%-80% consistency, might lead to the danger of focusing only on a narrow scope of synaesthesia. For example, in a study by Hupe, Bordier, and Dojat (2011) synaesthetes excelled on consistency tests and made few mistakes; critically these mistakes were later described by the synaesthetes as the second concurrent to a particular inducer, i.e. for these subjects one grapheme elicited 2 colours. It is possible that the reason why some synaesthetes are not highly consistent is because they experience multiple or complex 3D concurrents within the same modality or cognitive stream. Alternatively, the reason why some people report synaesthetic associations but underperform on consistency tests may be that their associations are genuine, yet transient over time. For example, in a questionnaire about alphabet spatial forms administered to 474 synaesthetes, 4% reported having a form, yet stated that its shape was not stable over time (Jonas et al., 2011). People with transient synaesthetic associations are generally not considered “truly” synaesthetic; however we should be careful not to make such decisions prematurely. On the other hand, it is possible that these people misinterpret the nature of synaesthetic associations and are not synaesthetic at all.

Secondly, some synaesthetes have few inducers, but are very consistent in their reports. These synaesthetes are often excluded from research, which brings about two issues. The first is that there is no agreement on what is the “sufficient” number of inducers a subject with a particular type of synaesthesia should have. The second issue is whether it might be worth including this group in research, thereby gaining new and interesting insights about different forms of synaesthesia. For example, Amin and collegues (2011) administered questionnaires to 34 grapheme personification synaesthetes and reported that 64% personified 20 or more graphemes, 27% personified between 10 and 20 graphemes, and 9% personified fewer than 10 graphemes. Eleven of these subjects were also tested for consistency of gender and personality associations. Unfortunately, it is not clear how many graphemes these subjects personified. For personality consistency, the synaesthetes scored an average of 70.4% (range 42-97%) while the control group scored an average of 49.9%.

(9)

There were no significant differences on measures of gender consistency between synaesthetic and control groups.

Automaticity

Several studies have used a modified Stroop paradigm to demonstrate that synaesthesia is automatic. In the most common version of this task, synaesthetes take a longer time to name the print colour of a grapheme that is incongruent to their individual synaesthetic colour, than the print colour of a grapheme that matches their synaesthetic colour (Wollen and Ruggiero, 1983; Mills, 1999; Mattingley et al., 2001; Smilek et al., 2001; Lupiañez and Callejas, 2005; Laeng, Hugdahl, and Specht, 2011). The fact that a synaesthetic colour can interfere with grapheme or word naming suggests that binding of colour to form occurs automatically and cannot be ignored (Dixon et al., 2000). It should be noted that even though automaticity is considered one of the defining features of synaesthesia, the modified Stroop task is not used as a diagnostic tool. This reflects the fact that the synaesthetic stroop effect has been demonstrated in non-synaesthetes with explicitly learned colour-grapheme associations (Elias et al., 2003). Similarly, Meier and Rothern (2009) showed that a week of colour-letter association training in non-synaesthetes can produce the synaesthetic stroop effect. However, the subjects reported that letters triggered memories of colour associations, not the experience of colour itself. These findings also support the claim that the modified Stroop task cannot be used to demonstrate the perceptual nature of synaesthesia (Ramachandran and Hubbard, 2001). Synaesthetic experiences can also be modulated by selective attention, as demonstrated by Rich and Mattingley (2003). They used a variant of the modified Stroop task that was comprised of alphanumeric local-global stimuli: the local character elicited one synaesthetic colour while the global character elicited a different synaesthetic colour. When synaesthetes actively attended to the congruently coloured local stimulus instead of the incongruently coloured global stimulus, the stroop effect was diminished compared to trials in which they were not asked to use selective attention. The same effect was observed when congruency was switched. All in all, scientists agree that synaesthetic experiences arise involuntarily, but can be modified by selective attention (Mattingley, 2009).

(10)

Summary

We have shown that the task of defining synaesthesia is related to three major issues: the nature of synaesthetic associations (perceptual, conceptual), consistency of synaesthetic associations, and automaticity of synaesthetic associations. Firstly, many would agree that capturing the complex and variable experience of synaesthesia in a single definition is not a straightforward task. As there are numerous types of synaesthesia, researchers should aim to generate a definition that applies to all of them. This can be achieved by stressing the importance of both perceptual and conceptual aspects of synaesthesia. With regards to consistency, researchers are advised to (1) investigate why certain subjects perform in the middle range on consistency tests before they decide whether to include them in research, (2) consider whether subjects with few but consistent inducers should be excluded from research and if that is the case, contemplate what threshold should be used. Importantly, these decisions (along with rationales) should be reported in articles, thereby aiding other scientists who are faced with similar issues. The third issue, automaticity of synaesthetic associations, is not questioned per se, however, researchers should bear in mind that the modified Stroop task has certain limitations. While it adequately demonstrates automaticity, it cannot (1) be used to distinguish synaesthetes from non-synaesthetes and (2) cannot be used to infer whether interference occurs at a perceptual or conceptual level.

Neurobiology of synaesthesia

Since the beginning of the investigation into the neural basis of synaesthesia the main debate has been whether it arises from cross-activation (Ramachandran and Hubbard, 2001b) or from disinhibition of feedback (Grossebacher and Lovelace, 2001). Other models have been proposed since: the two-stage model (Hubbard, 2007), the cascaded cross-tuning model (Brang, 2010b), and the multisensory model (Goller, Otten, and Ward, 2008). I will briefly describe these models, along with evidence that supports or rejects them. Due to the fact that the majority of studies focus on word-colour synaesthesia, which can be divided into grapheme-colour, phoneme-colour, and lexical-colour synaesthesia (Weiss et al., 2001); I will limit the discussion to this group of synaesthesia. For an overview of the various existing definitions of grapheme-colour synaesthesia, see the meta-analysis in the next section.

(11)

Cross-activation model

The finding that the visual word form area (VWFA) lies in the middle portion of the left fusiform gyrus (Cohen and Dehaene, 2004) adjacent to colour processing area V4/V8 (Bartels and Zeki, 2000; Wade et al., 2002) has led to the proposal that grapheme colour synaesthesia may arise from cross-wiring between these areas (Ramachandran and Hubbard, 2001b). Since synaesthesia runs in families the authors suggested that a genetic mutation could cause defective pruning or stabilization of connections between brain maps. Moreover, they proposed that synaesthetic cross-wiring is analogous to cross-wiring between hand and face areas of amputees with phantom limbs. This explains why a phantom limb patient experiences touch simultaneously on his face and on his phantom limb when his face is brushed with a swab (Ramachandran and Hirstein, 1998).

Disinhibited feedback model

The disinhibited feedback theory posits that synaesthesia arises from disinhibited feedback from higher order multisensory areas, such as the superior temporal sulcus (STS), or the tempero-parietal-occipital junction, to lower visual areas. According to this model, non-synaesthetes are able to inhibit top-down signalling thus no concurrents are activated. While the cross-activation model suggests that synaesthesia arises from abnormal neural connections, the disinhibited model proposes that it is based on normally existing neural networks. The latter can be supported by findings that some non-synaesthetes experience synaesthesia when taking psychedelic drugs (Grossebacher and Lovelace, 2001). However, it has been reported that drug-induced synaesthesia involves more complex visual scenes than typically experienced in congenital synaesthesia, indicating that they might arise from different mechanisms (Shanon, 2002; Afra, Funke, and Matsuo, 2009).

Cascaded cross-tuning model

Brang and colleuges (2010) proposed an updated version of the cross-activation model, termed the cascaded cross-tuning model (CCT), which assumes that there are horizontal and feedback connections between posterior temporal grapheme areas and V4. The authors based their model on findings from grapheme processing research, namely hierarchical feature analysis. Thus, cascaded activation between feature-level and letter-level processes are responsible for low-level activations of letter representations and low level activations of

(12)

colour representations in V4. The model is supported by the authors’ magnetoencephalography (MEG) study that will be described in detail later.

Two-stage model

In light of a number of studies that provided support either for cross-activation or disinhibited feedback, Hubbard (2007) proposed an integrated model, also known as the two-stage model, in which synaesthetic colours are elicited in the fusiform gyrus via cross-activation, followed by hyperbinding in the parietal cortex. We will now consider evidence in favour of these models, bearing in mind that evidence that supports either cross-activation or disinhibited feedback, but does not disprove the opposing model, can also be interpreted as evidence for the two-stage model.

Neuroimaging data

Functional imaging studies of grapheme-colour synaesthesia show that visual area V4/V8, which has a role in colour processing, is activated in synaesthetes for letters and words but not for tones or other linguistic stimuli. Critically, this activation is not observed in non-synaesthetic controls (Nunn et al., 2002; Hubbard et al., 2005; Sperling et al., 2006; Leeuwen et al., 2010; Laeng et al., 2011). On the other hand, some studies report increased activation in the parietal lobes of synaesthetes (while they experience synaesthetic colours) compared to non-synaesthetic controls. Specifically, increased activation has been reported in the superior parietal lobe (Paelesu et al., 1995; Weiss et al., 2005; Laeng et al, 2011) and in the left inferior parietal lobe (Nunn et al., 2002; Weiss et al., 2005; Rouw and Scholte, 2010). A recent review of brain areas in synaesthesia suggests the involvement of a network of areas: sensory regions that are normally evoked during the perception of “real” stimuli, parietal areas, and frontal areas, particularly the right dorsolateral prefrontal cortex (Rouw et al., 2011).

Interestingly, an fMRI study by Gashler-Markefski et al. (2011) showed that auditory-visual synaesthetes show similar activation in the left superior temporal gyrus (auditory and language areas) for words and for pure tones, whereas non-synaesthetes show significantly larger activation for words than tones. It should be noted that for the synaesthetes words elicited reliable colour percepts, while tones only occasionally elicited colours. Increased tone activation compared to controls was also found in the right occipital lobes, the fusiform gyrus, and in the left middle temporal gyrus. The authors interpreted the abnormally high

(13)

activations to tones in synaesthetes in auditory and visual areas as a result of abnormally strong feedback from a multisensory area. However, it is also possible that these activations arise from direct auditory-visual cross-wiring.

EEG and MEG data

Functional imaging studies are useful for investigating what brain areas are active during the experience of synaesthesia, however, in order to disentangle which neural model provides the best explanation of synaesthesia, one must use techniques that have higher temporal resolution, such as electroencephalography (EEG) and magnetoencephalography (MEG). The cross-activation model predicts that V4 will be activated early (Ramachandran and Hubbard, 2001b), whereas the disinhibited feedback model proposed by Grossebacher and Lovelace (2001) predicts that V4 will be activated after a considerable delay, as information must first reach higher level multisensory areas before it is propagated back to V4.

Beelli and colleagues (2008) conducted an ERP study using electric brain tomography (high and low resolution) in which they investigated the time-course of synaesthesia in participants who experienced colours in response to spoken words and tones. Results showed that the posterior temporal occipital area (PIT) and area V4 were active during the perception of synaesthetic colours induced by auditory stimuli, thus replicating certain functional imaging studies (Nunn et al., 2002; Hubbard et al., 2005; Sperling et al., 2006 Leeuwen et al., 2010; Laeng et al., 2011). These brain areas were activated just 122 ms after the onset of the auditory stimulus, indicating that rapid cross-activation could be the basis of synaesthetic experiences. However, ERP has a substantially lower spatial resolution than fMRI, thus there is no guarantee that precisely PIT and V4 areas were isolated. Early ERP components (N1 and P2) that occur between 100 and 200 ms after stimulus onset have also been reported by other grapheme-colour synaesthesia studies (Sagiv and Ward, 2006; Brang et al., 2008; Brang et al., 2010a). Further evidence in favour of the cross-activation theory and particularly the cascaded cross-tuning theory comes from Brang and colleagues (2010b), who conducted an MEG study in a sample of four projector grapheme-colour synaesthetes. They compared the time-courses of activation of V4 and posterior temporal grapheme areas (PTGA), predicting that these areas will be activated nearly simultaneously in response to perceiving an inducer. The alternative hypothesis was that V4 will be activated only after a feedback sweep occurs. Results showed that V4 and PTGA were activated between 105 and

(14)

115 ms, providing evidence against disinhibited feedback - at least in projector synaesthetes. To our knowledge, there are no EEG or MEG studies that support the disinhibited feedback model; however, there are a number of transcranial magnetic stimulation (TMS) studies that do.

TMS data

Esterman and colleagues (2006) applied repetitive TMS, which is used for noninvasive stimulation of targeted brain areas, to the posterior parietal cortices (PPC) of two projector grapheme-colour synaesthetes while they performed on a colour-naming task. When rTMS was applied over right PPC, interference that normally arises in incongruent trials was attenuated. The authors proposed that rTMS over right PPC interfered with colour-form binding, thus diminishing the mismatch between the real colour of a letter and the synaesthetic colour. These results are in line with the idea that feedback from multisensory areas, such as the posterior parietal cortex, is important for synaesthetic perception. Similarly, Muggleton and colleagues (2007) applied TMS on various sites of the parietal lobes of five grapheme-colour synaesthetes (1 projector and 4 associators) while they performed on a colour-priming task. In this task, the real colour of a grapheme must be indicated while the synaesthetic colour must be ignored. Results showed that TMS over the right parietal-occipital junction, but not over other parietal sites, reduced interference thereby replicating the findings by Esterman et al. (2006). However, based on TMS studies, the existence of direct connections between lower sensory areas cannot be excluded. Thus it can be argued that these studies provide support for both the disinhibited feedback model and the two-stage model.

Connectivity data

A recent functional study by Neufeld et al (2012) was able to shed light on connectivity between brain areas involved in synaesthesia. In this study, auditory-visualsynaesthetes and controls were presented with different types auditory stimuli during which fMRI data was collected. Functional connectivity analysis revealed (1) no evidence of increased connectivity between the auditory and visual cortices of synaesthetes compared to controls and (2) increased connectivity in synaesthetes between the left IPC and primary auditory (A1) and visual areas (V1). The latter results suggest that synaesthetic experiences are accompanied with increased interaction between these areas, possibly reflecting an information flow from

(15)

A1 through left IPC to V1. These results provide support for the disinhibited feedback theory; however, further research is required to determine the exact nature of the information flow and to extend these findings to other types of synaesthesia. Moreover, it is necessary to take into account that there are individual differences among synaesthetes that could be coupled with distinct neural mechanisms.

Rouw and Scholte (2007) conducted a diffusion tensor imaging study (DTI) in which they examined microscopic properties of cerebral white matter in grapheme-colour synaesthetes and non-synaesthetic controls. Importantly, they identified projector and associator synaesthetes and interpreted the results according to this distinction. Two main findings emerged: (1) increased connectivity in parietal and frontal areas in synaesthetes compared to controls and (2) increased connectivity in the inferior temporal cortex near the fusiform gyrus in synaesthetes compared to controls. In addition, the strength of structural connectivity correlated with brain activity during synaesthetic experience (measured by the BOLD signal) in a cluster in the temporal cortex. Turning the focus on differences between projectors and associators, results showed that structural connectivity in the inferior temporal cortex correlated with the tendency to see colours in external space (projector). In contrast, structural connectivity within parietal and frontal areas did not correlate with individual differences among synaesthetes.

Leeuwen, Ouden, and Hagoort (2011) also focused on altered connectivity patterns between projector and associator synaesthetes. However, they used a different technique - dynamic causal modelling for fMRI. They found that for projectors, modulation of a bottom-up activation pathway to V4 best explained cross-activation data, whereas for associators modulation of a top-down pathway to V4 was the best explanation. The findings by Rouw and Scholte (2007) and Leeuwen et al. (2011) suggest that some of the conflicting evidence on cross-activation and disinhibited feedback could be attributed to individual differences, namely the projector-associator distinction.

Structural data

Structural imaging studies generally support the two-stage model proposed by Hubbard (2007). For example, Weiss and Fink (2009) conducted a voxel-based morphometry study (VBM) that examined grey matter (GM) differences in synaesthetes and in controls. Results

(16)

showed that synaesthetes had increased GM volumes in the left intraparietal cortex and in the right fusiform gyrus compared to controls. Somewhat similar results were reported by Jancke et al. (2009), who found that the most prominent differences in cortical structures of grapheme-colour synaesthetes compared to controls were in the anterior part of the fusiform gyrus, an area involved in colour, letter, and word processing (Bartels and Zeki, 2000).Rouw and Scholte (2010) reported that grapheme-colour synaesthetes had increased GM in the left superior parietal cortex and decreased gray GM in the cingulate gyrus compared to controls, regardless of projector/associator type. The finding that synaesthetes have increased GM in the superior parietal cortex is in line with previous evidence that implicates the role of this area in synaesthetic experiences. The authors also examined structural differences between projectors and associators. Projector synaesthetes had increased GM in the left visual cortex (V17 and V18), in and near the right primary auditory cortex, and in the left precentral gyrus (premotor cortex). In contrast, associator synaesthetes had increased GM (and increased brain activation) in and near the hippocampus, which has a role in memory. Associators also had increased GM bilaterally in the angular gyrus, an area involved in multi-modal integration. The authors thus demonstrated that subjective reports about synaesthetic experiences are related to the functions of involved brain areas. In addition, the results can be interpreted as support for the two-stage model of synaesthesia.

Multisensory model

Lastly, we will consider a model based on current findings from multisensory research domains. Traditionally, multisensory research assumed that sensory-specific areas feed information to higher-level areas in which multimodal neurons combine and integrate information from different senses. However, this view has been replaced by a general understanding that even early sensory areas have a role in multisensory processing (Driver and Noesselt, 2008). This has been supported by electrophysiological findings that there are neurons in primary and secondary sensory cortices that respond to stimuli from two or more modalities, which has been shown in monkeys (Wang et al., 2008; Lemus et al., 2010) and rats (Brett-Green et al., 2003; Wallace, Ramachandran, and Stein, 2004; Menzel et al., 2005; Hirokawa et al., 2008). Thus, another possible explanation of synaesthesia, proposed by Goller, Otten, and Ward (2008), is that synaesthetes might have more bimodal neurons than

(17)

unimodal neurons in early sensory areas compared to non-synaesthetes. Since synaesthesia cannot be studied in animals and there are no techniques that can non-invasively measure single cell activity in humans, there is no direct evidence for this premise. Moreover, it is limited to “perceptual” forms of synaesthesia with no explanation about the origins of synaesthesia that arises at a conceptual level. Nevertheless, it is useful to keep in mind that animal research suggests that multisensory integration might occur in earlier stages of information processing than previously implicated.

Summary

We presented evidence for various models of word-colour synaesthesia and grouped them according to the technique used. In general, neuroimaging studies support cross-activation, disinhibited feedback and two-stage models, EEG and MEG studies support cross-activation and cascaded cross-tuning models, while TMS studies support the disinhibited feedback model and, to a certain extent, the two-stage model. Overall, two conclusions can be made: first, the model that has received the most widespread support is the two-stage model, proposed by Hubbard (2007). At present, it seems likely that word-colour synaesthesia is characterized by cross-activation of areas in the fusiform gyrus that are dedicated to processing colour and graphemes/phonemes and by hyperbinding in the parietal cortex. Secondly, findings from structural imaging and functional connectivity studies suggest that differences related to the subjective experience of synaesthesia are characterized by distinct neural mechanisms. Namely, people who experience synaesthetic colours in external space (projectors) have (1) increased white matter coherence in the inferior temporal cortex compared to associators (Rouw and Scholte, 2007), (2) increased grey matter density and volume in modality-specific sensory areas and in the premotor cortex (Rouw and Scholte, 2010), and (3) their synaesthetic colours are likely to arise from bottom-up cross-activation from grapheme-processing areas in the fusiform gyrus (Leeuwen et al., 2011). In contrast, people who experience synaesthetic colours in “the mind’s eye” (associators) have (1) increased grey matter and increased brain activation in and near the hippocampus, increased grey matter bilaterally in the angular gyrus (Rouw and Scholte, 2010) and (2) their synaesthetic colours are likely to arise from top-down modulation of activation from parietal areas (Leeuwen et al., 2011). Evidence suggests that synaesthetic experiences arise from a combination of cross-activation and disinhibited feedback mechanisms. However, it is

(18)

possible that in some people, one of the mechanisms is more prominent, thereby giving rise to distinct synaesthetic experiences. At the extremes of a continuum, the best explanation for projectors might be cross-activation while the best explanation for associators might be disinhibited feedback.

Meta-analysis of grapheme-colour synaesthesia studies

Grapheme-colour synaesthesia is the most commonly studied type of synaesthesia, yet there are discrepancies in the way different studies define it. We conducted a meta-analysis to explore the ambiguity of the term “grapheme-colour” synaesthesia. Since we discussed the issue of consistency earlier in the paper, we also included consistency scores of synaesthetes and non-synaesthetes, if they were reported. With the results of the meta-analysis we hope to stress the importance of providing detailed accounts of subjects’ synaesthetic experiences. First, we will consider some examples of definitions of grapheme-colour synaesthesia:

1) Only visually presented graphemes elicit synaesthetic colours

“In this study, we focus on colour-graphemic synaesthesia, a condition in which synaesthetes report vivid colour experiences induced by written letters.” Retrieved from Sperling et al. (2006). “In one of the most common forms, viewing numbers or letters (graphemes) elicits the percept of a specific color, known as grapheme-color synesthesia.” Retreived from Brang et al. (2010).

2) Visual and auditory presentations of graphemes elicit synaesthetic colours

“Individuals with grapheme-color synaesthesia experience vivid colors whenever they see, hear, or just think of ordinary letters and digits.” Retreived from Smilek et al.(2007).

3) Only auditory presentations of graphemes elicit synaesthetic colours (coloured-hearing)

“Colored-hearing synesthetes experience colors when hearing tones or spoken utterances.”

Retreived from Beeli et al.(2008).

Using different versions of the same definition would be less problematic if the authors would clearly state the nature of grapheme-colour synaesthesia in their subjects. If there is no clear specification whether they experience synaesthetic colours only to written letters or numbers or also to spoken letters and numbers, one must return to their initial definition, assuming that it accurately describes the synaesthetic experiences of all participants. Table 1 shows a summary of the meta-analysis of grapheme-colour synaesthesia studies.

(19)

Table 1: Meta-analysis of grapheme-colour synaesthesia studies First author, year Technique Sensitive to visual modality? Sensitive to auditory modality? Number of participants /controls Consistency in synaesthetes Consistency in non-synaesthetes Svartdal 1989

Behavioural yes not

reported

2/3 not reported no reported value

Paulesu 1995 PET no yes 6/6 above 90% no reported

value

Mills 1999 Behavioural yes yes 1/0 “consistent”,

no reported value

not applicable

Mattingley 2001

Behavioural yes not

reported 15/15 “consistent”, no reported value no reported value Ramachan-dran 2001

Behavioural yes not

reported 2/20 “consistent”, no reported value no reported value

Nunn 2002 MRI not

reported yes 12/27 “consistent”, no reported value no reported value

Gray 2002 MRI yes yes 48/10 “consistent”,

no reported value

Smilek 2001 Behavioural yes yes 1/7 “consistent”,

no reported value

Dixon 2004 Behavioural yes not

reported

12/0 “consistent”, no reported value

not applicable

Kadosh 2005 Behavioural yes not

Ward 2005 Behavioural yes not

reported

7 /8 mean 95%, SD= 3

mean 31%, SD= 14

Simner 2005 Behavioural yes not

reported 70/317 mean letters: 92% mean numbers: 93% mean letters: 36% mean numbers: 35%

Ward 2005 Heredity yes yes 85/48 mean 92%,

SD=10.8%

mean 33% SD=14.2%

Weiss 2005 MRI yes not

reported 9/0 mean 96.4% ± 3.6% not applicable Hubbard 2005

MRI yes not

reported 6/6 “consistent”, no reported value no reported value Mattingley 2006

Behavioural yes not

reported 14/14 mean 93%, SD=9% no reported value Esterman 2006 TMS yes not reported 2/0 “consistent”, no reported value not applicable

(20)

2006 reported no reported value Hubbard

2006

Behavioural yes not

reported 1 /12 “consistent”, no reported value no reported value Sperling 2006

MRI yes no 4/0 “consistent”,

no reported value

not applicable

Dixon 2006 Behavioural yes not

reported

1/0 “consistent”, no reported value

not applicable

Simner 2006 Behavioural yes yes 1/ 0 mean 92% not applicable

Edquist 2006 Behavioural yes not

reported

14/14 mean 88%, SD=8.2%

mean 26% SD=15.3%

Rouw 2007 DTI yes not

reported

18/18 minimum 90% required

no reported value

Beeli 2007 Behavioural no yes 19/0 “consistent”,

no reported value not applicable Muggleton 2007 TMS yes not reported 5/0 mean 96%, range = 86-100% not applicable Johnson 2007

Behavioural yes yes in 6/10 subjects

10/10 mean r=0.95, SD=0.07

no reported value

Nikolic 2007 Behavioural yes not

reported (but colored-music in 3/6 subjects) 6/12 mean 98% no reported value

Ward 2007 Behavioural yes yes 14/0 mean 96%,

range = 82-100%

no reported value

Brang 2008 ERP yes not

Barnett 2008 EEG yes not

reported

15/15 mean 95.4%, SE=1.2

no reported value

Hong 2008 Behavioural yes not

reported 4/0 “consistent”, no reported value not applicable Carriere 2009

Behavioural yes no 2/6 “consistent”,

no reported value

Jancke 2009 MRI yes not

Gebuis2009 Behavioural yes not

Weiss 2009a MRI yes not

reported

18/18 mean 90.2% ±7.6%

no reported value

(21)

Weiss2009b Behavioural yes not reported 10/10 mean 91.5 ± 6.9% no reported value

Rothen 2009 Behavioural yes not

reported

13/13 r =0.94 for

hue

r=0.21 for hue

Brang 2010a EEG yes not

Brang 2010b MEG yes not

Ward 2010 Behavioural yes not

reported 36/36 mean 96% in subset of 18 subjects range= 80-100% mean 33% (SD=14.2) taken from Ward, 2005 Rothen 2010a

Behavioural yes not

reported 44/0 mean r = 0.82 for hue not applicable Rothen 2010b TMS yes not reported 36/0 mean r = 0.77 for hue not applicable

Rouw, 2010 MRI yes yes in 25

out of 35 questioned 42/42 “consistent”, no reported value no reported value van Leeuwen 2010

MRI yes yes, n=8

(sound-colour) 21/19 mean 91%, SD= 7.5% mean 32% SD= 18%

Rich 2010 Behavioural yes not

reported

12/12 mean 85%, SD= 10%

no reported value Gevers 2010 Behavioural yes (only

digits) yes (speech sounds) 1/1 “consistent”, no reported value no reported value Ghirardelli 2010

Behavioural yes not

reported 1/15, 1/19, 1/18 in three experi-ments “consistent”, no reported value no reported value Bridgeman 2010

Behavioural yes not

reported

2/0 100% not applicable

Tsutomu 2010

EEG yes not

Brang 2011a Behavioural yes not

Brang 2011b Behavioural yes no 7/25 “consistent”,

no reported value

Hänggi 2011 MRI yes not

Specht 2011 MRI yes not

Nijboer 2011 Behavioural yes not

reported

12/24 “consistent”, no reported

no reported value

(22)

value van Leeuwen

2011

MRI yes not

reported

19/19 mean 91% mean 32.2%

Hupe 2011 MRI yes not

reported 10/25 “consistent”, no reported value no reported value Radvansky 2011

Behavioural yes yes, n=3 (speech sounds) 10/48 mean 72% (strict), or 0.88% (liberal) no reported value

Jäncke 2011 EEG no yes

(coloured hearing) 12/13 “consistent”, no reported value no reported value

Gross 2011 Behavioural yes yes, in

“many” (speech sounds) 9/23 all scores above 85% score range: 40-60% Terhune 2011 TMS, TDCS yes no 6/6 “consistent”, no reported value no reported value

Paffen 2011 Behavioural yes not

Banissy 2012 MRI yes yes 9/42 “consistent”,

no reported value

Watson 2012 Behavioural yes not

reported

54/not reported

not reported not applicable

Dovern 2012 MRI yes yes, in

11/12

12/12 mean 84.8% ± 14.5%

no reported value

Neufeld 2012 MRI no yes, in all 14/14 dimensionless

consistency: 1.26, SD=0.55

1.85, SD=0.50

Niccolai 2012 Behavioural 22% only VIS, 76% both VIS and AUD 18% only AUD, 76% both VIS and AUD 41 grapheme-color, 63 total /0 52% scored above 91% not applicable

Erskine 2012 Behavioural yes no 5/0 speed

congruency: above 85%

(23)

Table 2: Summary of meta-analysis of grapheme-colour synaesthesia studies

Visual inducers in synaesthetes (N=68) Auditory inducers in synaesthetes (N=68)

yes no not reported yes no not reported

63 4 1 21 5 42

Reported nature of inducers

The results of the meta-analysis indicate that the majority of studies (42) did not specifically mention whether synaesthetes that have visual inducers also have any auditory inducers. In some cases one can only speculate whether this is the case based on how the authors define grapheme-colour synaesthesia. However, 31% of the studies (21) reported that auditory presentations of graphemes trigger synaesthetic colours in their subjects (see table 2). Within the latter group, 16 studies reported visual and auditory inducers in all subjects or in a subgroup of subjects. Four studies reported auditory but no visual inducers (Paelesu et al., 1995; Beeli et al., 2007; Jancke et al., 2011, and Neufeld et al., 2011) and one study reported auditory inducers but did not specifically mention the absence of visual inducers (Nunn et al., 2002). Only 5 out of 68 studies specifically stated that their subjects experienced synaesthetic colours in response to visual inducers, but not to auditory inducers (Sperling et al., 2006; Carriere et al., 2009; Brang et al., 2011b; Terhune et al., 2011; Erskine et al., 2012).

Reported consistency

Regarding consistency of reported colour associations in synaesthetes, 36 out of 68 studies stated that synaesthetic performance was “consistent”, but did not report values, 2 studies did not mention consistency at all, while 30 studies did report consistency values (see table 2). Among the latter group, seventeen studies reported comparable data (average consistency), of which the average was 91.7%. In this sample, the lowest average score was 72% (Radvansky et al., 2011), whereas the highest average score was 100%, reported by Bridgeman et al. (2010). For non-synaesthetic controls, 41 out 68 studies did not report values, 17 did not have control subjects, while 10 did report consistency values. Among the studies that reported consistency values, 6 provided comparable data (average consistency),

Consistency in synaesthetes (N=68) Consistency in non-synaesthetes (N=68) consistent,

reported value

consistent, no reported value

not applicable reported value no reported value not applicable 30 36 2 10 41 17

(24)

the average of which was 31.62%. The lowest reported average score was 26% (Edquist et al., 2006), while the highest reported average score was 36% (Simner et al., 2005). Note that this range does not include studies that reported alternative measures of consistency, or did not report the average consistency. For example, Gross et al. (2011) reported that non-synaesthetes performed on a consistency task in the range of 40-60%.

Conclusions

The range of different types of synaesthesia and the differences among individuals within each type of synaesthesia pose a problem for those who wish to capture its complexity in a simple definition. A good definition of synaesthesia should stress its perceptual and conceptual characteristics, its persistence over the life span of a given individual, and its consistency and automaticity. However, the last two features should be dealt with caution, particularly when using them to select participants for research. In this process, it might be wise to assume that even less consistent and automatic synaesthetic experiences could, if studied, enlighten the field of synaesthesia research. Specifically, we advise researchers to expand the approach normally used to set inclusion criteria. Once subjects have been tested for consistency, the subjects that fall in the middle range of scores should not be instantly excluded, but should enter further evaluation. Open-ended questionnaires might bring insight as to why they are not entirely consistent in their reports. On the basis of these results, researchers might find it easier to determine whether their synaesthetic associations are real and whether they would be valuable subjects for research. Furthermore, if subjects are excluded on the basis of insufficient number of inducers, the reasons for this should be reported in order to aid fellow researchers who are faced with a similar task. Researchers are advised to make a balanced decision between the requirements of their experimental design and the possibility of focusing on a narrow scope of synaesthesia. A clear definition of synaesthesia should be accompanied by transparent inclusion criteria. Individual differences among synaesthetes need to be carefully examined all the while embracing the notion that less than typical performance on standard synaesthetic tests does not necessarily imply lack of synaesthesia.

In the second part of the paper, we discussed the neurobiology of word-colour synaesthesia. We believe that once the neurological correlates of synaesthesia are determined, they could

(25)

be used as one of the defining criteria. However, at this point the neural basis of word-colour synaesthesia, let alone synaesthesia in general, is unclear. We discuss several models, including the cross-activation model (Ramachandran and Hubbard, 2001), the disinhibition of feedback model (Grossebacher and Lovelace, 2001), the two-stage model (Hubbard, 2007), the cascaded cross-tuning model (Brang, 2010b),and the multisensory model (Goller, Otten, and Ward, 2008). Based on evidence from various techniques, we conclude that the two-stage model offers the best explanation for word-colour synaesthesia. Furthermore, findings show that there are structural brain differences between projectors and associators (Rouw and Scholte, 2007; Rouw and Scholte, 2010) and that their experience of synaesthesia is related to distinct neural mechanisms (Leeuwen et al., 2011). The two-stage model can be interpreted in line with this distinction: the neural mechanism for “projecting” synaesthetic colours might be related to cross-activation, while the neural mechanism for “associating” might be related to disinhibited feedback. It should be noted that the projector-associator distinction might not be reliable. For example, Edquist et al. (2006) reported that out of 9 subjects who were asked about projecting/associating again after 12 months, 3 subjects changed their answers. It is possible that they misunderstood the abstract concept of the question. On the other hand, it is also possible that their synaesthetic associations are experienced in a complex and interchangeable manner. Thus we advise researchers to test for consistency of projector-associator reports over time. This would not be time-consuming as the projector-associator questionnaire can be administered at the same times as classical consistency tests. With these results it would be possible to distinguish “strong” projectors/associators from those who fall in the continuum between the two extremes. We predict that the synaesthetic experiences of these three groups are related to distinct neural mechanisms.

Finally, the results of our meta-analysis indicate that there are inconsistencies in the way in which grapheme-colour synaesthesia is presented in scientific articles. Out of 68 studies that were analyzed, 62% did not specifically report whether subjects also had any auditory inducers. We were able to demonstrate that incoherent definitions of this type of synaesthesia can lead to confusion at a very basic level: what is the nature of the inducer that produces a concurrent? We also demonstrated that 52.9% of studies stated that synaesthetes performed consistently, but did not report specific values. If possible, we

(26)

suggest that researchers provide the following information per subject: (1) the nature of inducers, (2) the number of inducers, (3) the nature of synaesthetic experiences, (4) any additional types of synaesthesia that might be present, (4) consistency of synaesthetic experiences and (5) consistency of projector-associator reports. Presenting this information will resolve at least some of the issues that are at the heart of synaesthesia research.

(27)

References

Afra, P., Funke, M., & Matsuo, F. (2009). Acquired auditory-visual synesthesia: A window to early cross-modal sensory interactions. Psychology Research and Behavior Management , 2, 31-37.

Amin, M., Olu-Lafe, O., Claessen, L. E., Sobczak-Edmans, M., Ward, J., Williams, A. L., et al. (2011). Understanding grapheme personification: A social synaesthesia? Journal of Neuropsychology , 5 (2), 255-282.

Armel, K. C., & Ramachandran, V. S. (1999). Acquired synesthesia in retinitis pigmentosa. Neurocase , 5 (4), 293-296.

Banissy, M. J., & Ward, J. (2007). Mirror-touch synesthesia is linked with empathy. Nature Neuroscience , 10 (7), 815-816.

Banissy, M. J., Stewart, L., Muggleton, N. G., Griffiths, T. D., Walsh, V. Y., Ward, J., et al. (2012). Grapheme-color and tone-color synesthesia is associated with structural brain changes in visual regions implicated in color, form, and motion. Cognitive Neuroscience , 3 (1), 29-35.

Barnett, K. J., Foxe, J. J., Molholm, S., Kelly, S. P., Shalgi, S., Mitchell, K. J., et al. (2008). Differences in early sensory-perceptual processing in synesthesia: A visual evoked potential study. NeuroImage , 43 (3), 605-613.

Baron-Cohen, S., Burt, L., Smith-Laittan, F., Harrison, J., & Bolton, P. (1996). Synaesthesia: Prevalence and familiarity. Perception , 25 (9), 1073-1079.

Baron-Cohen, S., Wyk, M. A., & Binnie, C. (1987). Hearing words and seeing colours: an experimental investigation of a case of synaesthesia. Perception , 16 (6), 761-767.

Bartels, A., & Zeki, S. (2000). The architecture of the colour centre in the human visual brain: new results and a review. European Journal of Neuroscience , 12 (1), 172-193.

Beauchamp, M. (2008). Neural Substrates of Sound–Touch Synesthesia after a Thalamic Lesion. Journal of Neuroscience , 28 (50), 13696-702.

Beeli, G., Esslen, M., & Jancke, L. (2007). Frequency Correlates in Grapheme-Color Synaesthesia. Psychological Science , 18 (9), 788-792.

Brang, D., Edwards, L., Ramachandran, V., & Coulson, S. (2008). Is the Sky 2? Contextual Priming in Grapheme-Color Synaesthesia. Psychological Science , 19 (5), 421-428.

Brang, D., Hubbard, E., Coulson, S., Huang, M., & Ramachandran, V. (2010b).

Magnetoencephalography reveals early activation of V4 in grapheme-color synesthesia. NeuroImage, 53 (1), 268-274.

(28)

Brang, D., Kanai, S., Ramachandran, V. S., & Coulson, S. (2010a). Contextual priming in grapheme-color synesthetes and yoked controls: 400 msec in the life of a synesthete. J. Cognitive Neuroscience , 23 (7), 1681-1696.

Brang, D., Rouw, R., Ramachandran, V., & Coulson, S. (2011a). Similarly shaped letters evoke similar colors in grapheme-color synesthesia. Neuropsychologia , 49 (5), 1355-1358.

Brang, D., Williams, L. E., & Ramachandran, V. S. (2012b). Grapheme-color synesthetes show enhanced crossmodal processing between auditory and visual modalities. Cortex , 48 (5), 630-637.

Brett-Green, B., Fifkova, E., Larue, D. T., Winer, J. A., & Barth, D. S. (2003). A multisensory zone in rat parietotemporal cortex: Intra- and extracellular physiology and thalamocortical connections. The Journal of Comparative Neurology , 460 (2), 223-237.

Bridgeman, B., Winter, D., & Tseng, P. (2010). Dynamic phenomenology of grapheme-color synesthesia. Perception , 39 (5), 671-676.

Carriere, J. S., Eaton, D., Reynolds, M. G., Dixon, M. J., & Smilek, D. (2009). Grapheme-color synesthesia influences overt visual attention. J. Cognitive Neuroscience , 21 (2), 246-258.

Cohen, L., & Dehaene, S. (2004). Specialization within the ventral stream: the case for the visual word form area. NeuroImage , 22 (1), 466-476.

Day, S., Robertson, L. C., & Sagiv, N. E. (2005). Some demographic and socio-cultural aspects of synesthesia. V L. C. Robertson, & N. Sagiv (Edit.), Synaesthesia Perspectives from Cognitive Neuroscience (pp. 11-33). Oxford University Press.

Dixon, M. J., Smilek, D., Cudahy, C., & Merikle, P. M. (2000). Five plus two equals yellow: Mental arithmetic in people with synaesthesia is not coloured by visual experience. Nature , 406, 365.

Dixon, M. J., Smilek, D., Duffy, P. L., Zanna, M. P., & Merikle, P. M. (2006). The Role of Meaning in Grapheme-Colour Synaesthesia. Cortex , 42 (2), 243-252.

Dixon, M., Smilek, D., & Merikle, P. (2004). Not all synaesthetes are created equal: Projector versus associator synaesthetes. Cognitive, Affective, & Behavioral Neuroscience , 4, 335-343.

Dovern, A., Fink, G. R., Fromme, A. C., Wohlschlager, A. M., Weiss, P. H., & Riedl, V. (2012). Intrinsic Network Connectivity Reflects Consistency of Synesthetic Experiences. The Journal of Neuroscience , 32 (22), 7614-7621.

Driver, J., & Noesselt, T. (2008). Multisensory Interplay Reveals Crossmodal Influences on Sensory-Specific Brain Regions, Neural Responses, and Judgments. Neuron , 57 (1), 11-23.

(29)

Edquist, J., Rich, A. N., Brinkman, C., & Mattingley, J. B. (2006). Do Synaesthetic Colours Act as Unique Features in Visual Search? Cortex , 42 (2), 222-231.

Elias, L. J., Saucier, D. M., Hardie, C., & Sarty, G. E. (2003). Dissociating semantic and perceptual components of synaesthesia: behavioural and functional neuroanatomical investigations. Cognitive Brain Research , 16 (2), 232-237.

Erskine, H., Mattingley, J. B., & Arnold, D. H. (2012). Synaesthesia and colour constancy. Cortex, In press.

Esterman, M., Verstynen, T., Ivry, R. B., & Robertson, L. C. (2006). Coming Unbound: Disrupting Automatic Integration of Synesthetic Color and Graphemes by Transcranial Magnetic Stimulation of the Right Parietal Lobe. Journal of Cognitive Neuroscience , 18 (9), 1570-1576.

Gaschler-Markefski, B., Szycik, G. R., Sinke, C., Neufeld, J., Schneider, U., Baumgart, F., in drugi. (2011). Anomalous Auditory Cortex Activations in Colored Hearing Synaesthetes: An fMRI-Study. Seeing and Perceiving , 24 (4), 391-405.

Gebuis, T., Nijboer, T. C., & van der Smagt, M. J. (2009). Of Colored Numbers and Numbered Colors. Experimental Psychology , 56 (3), 180-187.

Gevers, W., Imbo, I., Cohen Kadosh, R., Fias, W., & Hartsuiker, R. J. (2010). Bidirectionality in synesthesia: Evidence from a multiplication verification task. Experimental Psychology , 57(3), 178-184.

Gheri, C., Chopping, S., & Morgan, M. (2008). Synaesthetic colours do not camouflage form in visual search. Proceedings of the Royal Society B: Biological Sciences , 275 (1636), 841-846.

Ghirardelli, T. G., Mills, C. B., Zilioli, M. K., Bailey, L. P., & Kretschmar, P. K. (2010). Synesthesia Affects Verification of Simple Arithmetic Equations. The Journal of General Psychology , 137 (2), 175-189.

Goller, A. I., Otten, L. J., & Ward, J. (2008). Seeing Sounds and Hearing Colors: An Event-related Potential Study of Auditory-Visual Synesthesia. Journal of Cognitive Neuroscience , 21 (10), 1869-1881.

Gray, J. C., & Baron-Cohen, S. (2002). Implications of Synaesthesia for Functionalism: Theory and Experiments. Implications of Synaesthesia for Functionalism: Theory and Experiments , 9 (12) , 5-31.

Green, J. A., & Goswami, U. (2008). Synesthesia and number cognition in children. Cognition , 106 (1), 463-473.

(30)

Gross, V. C., Neargarder, S., Caldwell-Harris, C. L., & Cronin-Golomb, A. (2011). Superior encoding enhances recall in color-graphemic synesthesia. Perception , 40 (2), 196-208.

Grossenbacher, P. G., & Lovelace, C. T. (2001). Mechanisms of synesthesia: cognitive and physiological constraints. Trends in Cognitive Sciences , 5 (1), 36-41.

Hanggi, J., Wotruba, D., & Jancke, L. (2011). Globally Altered Structural Brain Network Topology in Grapheme-Color Synesthesia. The Journal of Neuroscience , 31 (15), 5816-5828.

Hirokawa, J., Bosch, M., Sakata, S., Sakurai, Y., & Yamamori, T. (2008). Functional role of the secondary visual cortex in multisensory facilitation in rats. Neuroscience , 153 (4), 1402-1417.

Hong, S. W., & Blake, R. (2008). Early visual mechanisms do not contribute to synesthetic color experience. Vision Research , 48 (8), 1018-1026.

Hubbard, E. M., & Ramachandran, V. (2005). Neurocognitive Mechanisms of Synesthesia. Neuron , 48 (3), 509-520.

Hubbard, E. M., Arman, A. C., Ramachandran, V. S., & Boynton, G. M. (2005). Individual Differences among Grapheme-Color Synesthetes: Brain-Behavior Correlations. Neuron , 45 (6), 975-985.

Hubbard, E. M., Manohar, S., & Ramachandran, V. S. (2006). Contrast Affects the Strength of Synesthetic Colors. Cortex , 42 (2), 184-194.

Hubbard, E. (2007). Neurophysiology of synesthesia. Current Psychiatry Reports , 9, 193-199.

Hupe, J.-M., Bordier, C., & Dojat, M. (2011). The Neural Bases of Grapheme-Color Synesthesia Are Not Localized in Real Color-Sensitive Areas. Cerebral Cortex , In press.

Jancke, L., & Langer, N. (2011). A strong parietal hub in the small-world network of coloured-hearing synaesthetes during resting state EEG. Journal of Neuropsychology , 5 (2), 178-202.

Jancke, L., Beeli, G., Eulig, C., & Hanggi, J. (2009). The neuroanatomy of grapheme-color synesthesia. European Journal of Neuroscience , 29 (6), 1287-1293.

Jansari, A. S., Spiller, M. J., & Redfern, S. (2006). Number Synaesthesia: When Hearing Four Plus Five Looks Like Gold. Cortex , 42 (2), 253-258.

Jarick, M., Dixon, M. J., Stewart, M. T., Maxwell, E. C., & Smilek, D. (2009). A different outlook on time: Visual and auditory month names elicit different mental vantage points for a time-space synaesthete. Cortex , 45 (10), 1217-1228.

Johnson, A., Jepma, M., & De Jong, R. (2007). Colours sometimes count: Awareness and bidirectionality in grapheme-colour synaesthesia. The Quarterly Journal of Experimental Psychology , 60 (10), 1406-1422.