Language evolution: examining the link between cross-modality and aggression through the lens of disorders

Abstract

We demonstrate how two linguistic phenomena, figurative language (implicating cross-modality) and derogatory language (implicating aggression), both demand a precise degree of (dis)inhibition in the same cortico-subcortical brain circuits, in particular cortico-striatal networks, whose connectivity has been significantly enhanced in recent evolution. We examine four cognitive disorders/conditions that exhibit abnormal patterns of (dis)inhibition in these networks: schizophrenia (SZ), autism spectrum disorder (ASD), synaesthesia and Tourette's syndrome (TS), with the goal of understanding why the two phenomena altered reactive aggression and altered cross-modality cluster together in these disorders. Our proposal is that enhanced cross-modality (necessary to support language, in particular metaphoricity) was a result, partly a side-effect, of self-domestication (SD). SD targeted the taming of reactive aggression, but reactive impulses are controlled by the same cortico-subcortical networks that are implicated in cross-modality. We further add that this biological process of SD did not act alone, but was engaged in an intense feedback loop with the cultural emergence of early forms of language/grammar, whose high degree of raw metaphoricity and verbal aggression also contributed to increased brain connectivity and cortical control. Consequently, in conjunction with linguistic expressions serving as approximations/‘fossils’ of the earliest stages of language, these cognitive disorders/conditions serve as confident proxies of brain changes in language evolution, helping us reconstruct certain crucial aspects of early prehistoric languages and cognition, as well as shed new light on the nature of the disorders.

This article is part of the theme issue ‘Reconstructing prehistoric languages’.

1. Introduction

One major transition in human evolution concerns the emergence of modern cognition and behaviour, intricately linked to the ability to learn and use language(s) [1,2]. In contrast to the previous views claiming that language emerged quite suddenly, in a one-step process (e.g. [3–7]), ample evidence suggests that it builds instead on many different cognitive and behavioural components with an old evolutionary history (e.g. [8–10]). Moreover, current evidence supports the view that a complex feedback loop exists between our cognition, the languages we speak, the functions they perform, and the physical and cultural environment in which we live (e.g. [11–16]). Recently, the Self-domestication (SD) hypothesis of human origins has been posited, claiming that many of our characteristic physical, behavioural and even cognitive traits parallel those commonly found in domesticated mammals and that they resulted from selection for reduced aggression ([17], and references therein), leading specifically to a reduction of the reactivity to stress of the hypothalamic–pituitary–adrenal (HPA) axis, a major neuroendocrine system regulating a great number of bodily functions (see [18,19] for details). In this respect, bonobos are also considered to have undergone a SD process (e.g. [20,21]). Compared with chimpanzees, bonobos exhibit, among other features, more rounded skulls/brains, depigmentation, paedomorphic traits, broad cooperation, increased engagement in play and increased sensitivity to social cues (see [22], their table 1 for a useful summary). Regarding, specifically, aggression, compared with chimpanzees, bonobos exhibit reduced proactive aggression and reduced severity of reactive aggression [23], as well as stronger links between the anterior cingulate gyrus and the amygdala, a pathway involved in the inhibition of aggression [21]. Compared with chimpanzees, bonobos also show an enlarged dorsal amygdala [21], implicated in the activation of the HPA axis (e.g. [24,25]). Relevant for our purposes, the SD hypothesis has also been invoked to explain aspects of human language evolution (e.g. [26]), and in particular, aspects of the evolution of languages in prehistory [27,28].¹

In this paper, we reconsider two linguistic phenomena previously proposed to play a role in language evolution, sound symbolism (as found in e.g. ideophones) and vivid compounds (as found in e.g. derogatory language), under the framework of the SD hypothesis. More specifically, we argue that (clinical) conditions impacting upon these phenomena can be used as proxies for certain aspects of the previous stages in the evolution of language(s) (in line with [28,30,31]).² In this respect, we focus on the conditions such as synesthaesia, related to cross-modality (e.g. [32–34]), and Tourette's syndrome (TS), related to the management of aggression (e.g. [35]), and discuss their relevance and relatedness to other disorders, specifically, schizophrenia (SZ) and autism spectrum disorder (ASD), which also impact upon cross-modality and aggression and which have previously been analysed as useful proxies for shedding light on the evolution of human cognition and sociality (e.g. [36]). As noted, we do this in the light of the SD hypothesis, which provides a mechanism to bring directly together these two otherwise disparate phenomena (aggression and cross-modality), which, as we purport to show, are bound by a common neuronal substrate, and a common evolutionary cause. In this sense, our proposal offers a mechanism for bringing together biological and cultural considerations, both necessary for shedding light on language evolution. Our main argument will be that these four conditions implicate abnormal patterns of inhibition/disinhibition in the same brain circuits, simultaneously impacting upon both the realm of cross-modality (related to e.g. sound symbolism, but also to compounding) and the realm of aggression (related to derogatory language, as well as, more generally, to the management of emotions). Ultimately, we propose that these four conditions link to atypical degrees of SD. These considerations not only begin to draw a multidimensional mosaic of cognitive and behavioural functions affected in disorders, which is of interest for aetiopathology of disorders, but also illuminate some crucial steps in the evolution of language(s), as well as identify new possibilities for testing.

2. Cross-modality and aggression: two dimensions of linguistic ‘fossils’

We build our arguments on the following recent proposals. First, we engage the proposals relying on cross-modality, considering that human cognition excels in transcending the boundaries of core cognitive systems (e.g. [37]). This enhanced capacity, which enables one to merge concepts and create new ones, has been specifically conflated with our ability for merging lexical items into phrases and sentences (e.g. [38]). Moreover, cross-modality is also expected to underlie our capacity for (pragmatic) inferencing, as well as our ability to metonymize and metaphorize (and more generally, to use figurative language), all of which are crucially involved in the complexification of languages via grammaticalization [39–41], as well as in conveying implicit meanings via implicatures [42]. Second, we engage the proposals couched within the SD hypothesis, claiming that during recent evolution significant changes in the management of aggression have occurred, with reactive aggression decreasing over time and with proactive, premeditated aggression gaining prominence quite late in our history. Third, we engage the proposals that human cognitive disorders, impacting upon both our ability for cross-modal thinking and the management of aggression, like SZ or ASD, can be linked to our evolutionary history, resulting from the altered expression of the same genes that have been selected in the line of descent of humans and that are expected to account for our cognitive and behavioural innovations [36,43,44]. Fourth, we specifically engage proposals that these conditions exhibit abnormal features of human SD, either exaggerated, as in SZ, or attenuated, as in ASD [30,31]. Finally, we build on our original proposal that the sort of cognitive and behavioural changes resulting in the human phenotype have engaged in an intense feedback loop with the early stages of language evolution, with the management of reactive aggression (and ultimately, our SD) being crucially involved. In brief, we contend that early stages of grammar, characterized by reduced syntactic complexity and reconstructed by relying on syntactic theory [45,46], were dominated by colourful metaphorical expressions, often derogatory, which provided an especially useful means for replacing physical aggression by verbal aggression (see [28,27,47] for details).

The rest of this section considers two types of such colourful linguistic expressions as found in present-day languages, namely, vivid compounds and ideophones, which we find of particular relevance for exploring the links between metaphoricity (and more generally, cross-modality), aggression (and more generally, the management of emotions in a social setting), and language complexity (and more generally, language evolution). We will build the argument that these phenomena are directly linked, via a common evolutionary cause, involving an increased global connectivity in the relevant brain networks.

(a). Vivid compounds (from English and Serbian)

• (1) kill-joy, turn-skin (cf. turn-coat), hunch-back, wag-tail, tattle-tale, scatter-brain, cut-throat, heck-wood, busy-body, cry-baby, fill-belly (glutton) spit-fire, tumble-dung (insect), crake-bone (crack-bone), shave-tail (shove-tail), fuck-wit
• (2) cepi-dlaka ‘split-hair’ (hair-splitter); guli-koža ‘peel-skin’ (who rips you off); vrti-guz ‘spin-butt’ (restless person, fidget); muti-voda ‘muddy-water’ (trouble-maker); jebi-vetar ‘fuck-wind’ (charlatan); vuci-guz ‘drag-butt’ (slow-moving person); kosi-noga ‘skew-leg’ (person who limps); lezi-baba ‘lie-old-woman’ (loose woman or man); podvi-rep ‘fold-tail’ (one who is crestfallen).

The grammar behind these compounds, coined by one single operation merging just one verb and one noun, has previously been reconstructed as a proxy of the earliest stages of grammar, based on the postulates of syntactic theory (e.g. [45–47]). Merging together two concrete, tangible concepts (like ‘fill’ and ‘belly’) in order to create an abstract concept (glutton) is an act of metaphorical creativity, an act that permeates almost every dimension of human language and cognition [48]. Metaphors thus involve a leap, or a crossing, typically from the concrete to the abstract, but metaphorical leaps (and cross-modality) can also involve a crossing of domains of different senses (e.g. ‘loud shirt’, ‘bitter cold’, ‘sharp cheese’), which will prove relevant to our discussion of synaesthesia in §3. Accordingly, if one considers the raw metaphoricity of these compounds, in the sense that they rely on crude images often depicting body parts and functions, frequently resulting in obscene, highly derogatory phrases (for many more colourful examples, see [45,46]), one can argue that they resemble the earliest verbal manifestations of cross-modal thinking. Importantly for our argument here, these compounds are mostly aimed at conveying/evoking emotions. Modern languages (continue to) show high levels of metaphoricity, but the modern metaphorical language tends to be subtler and more sophisticated, and less directly linked to the realm of emotions (more on this below). To take just one example, languages systematically conceptualize time (a more abstract concept) in terms of space (a more concrete, observable concept), as we talk about the problems being ‘behind’ us, or ‘ahead’ of us (for many more examples and patterns, see [48]).³

A second interesting feature of these compounds is their raw, visceral effect (e.g. [45,50]), and particularly, their great propensity for verbal aggression. In many cultures, comparable compositions are used in verbal duels/contests, whose purpose seems to be to replace physical fighting, thus contributing to a better survival of those who engage in verbal, as opposed to physical, duels (for more discussion, see e.g. [47,51] and references therein). In this sense, these verbal expressions would have been highly adaptive at the dawn of human evolution. Related to this, these compounds are also highly humorous. It is well established that humour provides relief from stress and tension (e.g. [52–54]), leading to a reduction in cortisol levels [55,56], and thus in reactive aggression levels. Recall that a reduction in the input to the HPA axis has been postulated as the key triggering factor (and one of the hallmarks) of domestication (see [57] for details).

Based on these considerations, we have previously proposed that these kinds of compositions played a key role in the reduction of reactive aggression, reinforcing a trend closely associated with SD, by providing a more adaptive way to compete, that is, the means for the gradual replacement of physical aggression by verbal aggression, and verbal behaviour more generally. In other words, our proposal postulates an intense feedback-loop between human SD and verbal behaviour, via the management of aggression, the biology–culture feedback loop, which would have significantly accelerated both SD and language evolution (see [28,27] for details). This reinforces the role of this vivid compound type as a useful probe into the earliest stages in the evolution of human languages.

(b). Ideophones

A second domain in which we observe a close connection between an increased ability for cross-modal associations and a management of emotions concerns ideophones. As defined by Dingemanse [58], ideophones are ‘marked words which vividly depict sensory events,’ showing a variety of phonological and semantic idiosyncrasies (see also [59,60]), and enabling a more precise, vivid expression of perceptions and experiences via different sensory domains through a single linguistic channel [33,61]. In some languages, such as Hmong, Korean and Japanese, ideophones are quite wide-spread and essential devices. Two examples from Hmong are: mî mê (mosquitoes buzzing) and plı̃-plǒn (empty bottle submerged in water filling up) [60]. Some illustrative examples from English include tick-tock, jibber-jabber, imitating a sound, and zig-zag, imitating a shape.

In the context of this paper, ideophones are of particular interest for several reasons. First, as shown in the examples above, ideophones often come in reduplicative pairs, and thus also involve binary compositions, fitting into the mould of the two-slot proto-grammar, illustrated above (see §2a) with vivid compounds in (1) and (2) (for this, and more types of binary compositions of evolutionary import, see [45]).⁴ The difference is that such ideophonic pairings do not involve typical words/morphemes that contribute compositionally to the meaning. Instead, the mapping between sound and meaning in ideophones is often iconic (i.e. partly imitative), and according to some authors, synaesthetic in nature [63]. This means that the interpretation (and the generation) of ideophones still demands enhanced cross-modal abilities. Third, ideophones entail a stronger connection between language and emotion, and it is thus not surprising that they can also be used for derogatory purposes, like the vivid compounds discussed above (see e.g. [64] for Gbeya), suggesting that they can also be of relevance for the management of reactive aggression. Overall, their structural simplicity, their relationship with less sophisticated forms of cross-modal thinking (mostly related to sensory information), and their stronger link to emotions (e.g. [65]) make ideophones another useful proxy for a stepping stone, an entry, into initial stages of grammar in human evolution. In truth, vivid compounds and ideophones may share other features of interest, including possibly a common neurobiological substrate. As shown by Progovac and colleagues [50]), two-slot compounds (the kill-joy type), in contrast with the more complex -er counterparts (i.e. joy-kill-er type), recruit the fusiform gyrus area (specifically BA37), the visual area that is also implicated in synaesthesia, a condition entailing abnormally enhanced cross-modal activity (more on this in the next section). If these kinds of compositions were among the earliest forms of language, then this finding is consistent with Ramachandran & Hubbard's [63]) suggestion that the beginning of cross-modality involved a cross-wiring in fusiform gyrus (see also [55]).

In the next section, we will show that this link between cross-modality, aggression and language complexity can also be observed in cognitive/language disorders, although in altered forms. Later we will rely on the SD hypothesis as the framework that can unify these apparently disparate phenomena. Also, because of the deep link that exists between evolution and abnormal development (with recently evolved genes in our species being enriched in candidate genes for human-specific cognitive conditions, as noted in §3), we will conclude that cognitive disorders (in particular the disorders/conditions we discuss in this paper) can serve as (even more) informative proxies for the stages in the evolution of language(s) that we have started to illuminate with the consideration of the ‘living fossils’ discussed in this section.

3. Cross-modality and aggression: two dimensions impaired in disorders implicating problems with language

In this section, we show how all these aspects highlighted by linguistic proxies, i.e. the emergence of cross-modal associations, the refinement of the management of aggression and the complexification of grammar, can be illuminated as well by examining (selected) cognitive disorders and conditions, as all these dimensions appear simultaneously affected in them. In §6, we will reconcile these seemingly disparate dimensions by invoking a common evolutionary cause, related to the SD hypothesis.

(a). Autism spectrum disorder and schizophrenia

Regarding cross-modality, it so happens that in most, if not all, highly prevalent cognitive disorders involving language deficits, the ability to understand and make use of figurative (metaphorical) language has been reported to be affected. For instance, individuals with ASD face difficulties with establishing connections between two elements of a compound whose combination requires metaphorical ‘stretching’ of meaning (e.g. [66,67]). Likewise, they face problems with understanding abstract concepts and metaphors in general [68, p. 47; 69]. In addition, Nagase [70] reported a U-shaped relationship between the severity of autistic traits and appreciation of humour (see also [71]). At the same time, subjects with ASD show an aptitude for rule-based systems, i.e. hyper-systemizing (e.g. [72,73]), with a positive impact in the domain of analytical, rule-governed language and thinking (see §4 for more details on this). Likewise, people with SZ can be significantly impaired in the comprehension of novel metaphors [74]. They also score significantly lower than controls in subtests measuring comprehension of implicit information, interpretation of humour, as well as discernment of emotional language prosody and comprehension of discourse [75]. At the same time, individuals exhibiting a mild, non-clinical manifestation of psychotic-affective conditions show relative strengths in the domain of pragmatics, when interpreting metaphors, emotions, humour and irony [76, p. 238]. In the same vein, relatives of people with SZ may have an advantage in artistic expression and originality [77].

These findings can be ascribed to a partial inhibition (ASD) or disinhibition (SZ) of cross-modality connections necessary for metaphorical language, with milder manifestations of the latter condition conferring some adaptive advantages. In fact, as discussed in Benítez-Burraco [78], the problems that the affected individuals exhibit as well with literal meanings might also result in part from the same altered pattern of inhibition/disinhibition, considering the fuzzy boundaries between figurative and non-figurative language, with the same basic cognitive mechanisms (such as metaphorization or metonymization) being involved in both domains. Ultimately, as discussed in the previous section, the affected mechanisms boil down to our species-specific enhanced ability for cross-modal associations. This postulated connection between the emergence of our enhanced cross-modal abilities and (clinical) conditions in which these abilities are found impaired is in line with the significant enrichment of, specifically, candidates for ASD and SZ in genes that have been positively selected in our species (e.g. [44,79,80]), particularly, in genes involved in the evolution of our distinctive skull/brain (see [81, Table 2] and in our species-specific distinctive patterns of brain activity (see [82, Table 1].

Importantly, ASD and SZ present as well with socialization and behavioural disturbances that, at least in part, reduce to an inappropriate management of aggression. Accordingly, reactive aggressive behaviour is reported to be significantly higher in children, specifically boys, with ASD, correlating with lower cognitive outcomes [83,84]. In addition, children with ASD display more reactive than proactive aggression attitudes [85]. With regards to SZ, aggressive behaviour is slightly, but significantly, more frequent in patients than in the typical population [86,87], although no positive correlation seems to exist between physical aggression and neuropsychological performance in affected people [88]. Importantly, in contrast to ASD, subjects with SZ and with psychopathic features engage in aggression that is more premeditated, i.e. proactive [89]. In other words, while both conditions, ASD and SZ, show deviation from the typical population, when compared with SZ, ASD shows markedly higher reactive aggression and more marked deficits in metaphorical and even in the non-metaphorical language domain.

These findings in the domains of cross-modal thinking and of aggression management can be viewed as particular manifestations of the ‘diametric brain hypothesis' [36], according to which humans evolved a suite of non-social brain adaptations, as well as a suite of social brain adaptations, which tend to exhibit trade-offs (in this respect, see also §§4–6). Hence, SZ would involve a(n) (maladaptively) over-developed social-brain phenotype in conjunction with, to some degree, a(n) (maladaptive) under-development of non-social brain phenotype, whereas ASD would entail the opposite. Importantly, SZ and ASD also implicate the opposite features of human SD, with ASD showing some features of a hypo-domesticated human phenotype, and SZ some features of a hyper-domesticated phenotype [30,31], the significance of which will be discussed in §6. As discussed further in §4, ASD and synaesthesia/SZ can thus be seen as complementary conditions, with trade-offs in systematicity versus creativity, acting as poles of otherwise continuous variability attested in humans.

This proposal regarding ASD and SZ is significantly enriched by considering two additional conditions that also entail abnormal patterns of cross-modality and an altered management of aggression: synaesthesia and Tourette's syndrome (TS). Similarly to what we find in SZ and ASD, these two conditions also exhibit an imbalance between inhibition and disinhibition, impacting upon both cross-modality and aggression.

(b). Tourette's syndrome

TS is a hereditary tic disorder that is sometimes, not always, accompanied by involuntary production of obscene speech and derogatory remarks/insults (coprolalia). The following findings regarding TS are of direct relevance for our proposal: (i) the linguistic nature of the derogatory language produced by a subset of TS individuals; (ii) the abnormal processing of figurative language by subjects with TS; (iii) the uncontrollable production of derogatory language (coprolalia) by a subset of TS individuals; (iv) the higher incidence of physical aggression/violence among TS individuals; and (v) the heightened reactivity of the HPA axis, as well as the postulated anxiolytic (anxiety-relieving) property of tics in TS.

With regards to the linguistic nature of the derogatory speech identified as coprolalia (i), Van Lancker & Cummings [90] provide many examples, which feature a preponderance of single-word utterances (the most frequent among them being fuck, shit, bitch, which also predominate as elements of verb–noun compounds illustrated in §2; see [45,46]), as well as some simple phrases/compounds (e.g. fuck you, motherfucker, asshole). Concerning the processing of figurative language by individuals with TS (ii), Eddy and colleagues [91] found evidence of a reduced ability for interpreting nonliteral remarks, such as sarcasm, metaphor, and indirect requests (see also [92]). By contrast, people with TS do not seem to exhibit a significant impairment of structural aspects of language, i.e. literal language (but see e.g. [93] for some problems with expressive language). Instead, subjects with this condition have been reported to show enhanced grammar processing ability, specifically when it comes to rule-governed, rather than idiosyncratic/irregular, linguistic knowledge [94], possibly resulting from an enhanced procedural memory [95]. Recall that this advantage in systematic, rule-governed processing ability is also attested in ASD, to be further discussed in §4.

With regards to the uncontrollable production of derogatory language (coprolalia) by a subset of TS individuals (iii), coprolalic words/phrases are pronounced as brief, loud outbursts, with abnormal, exaggerated prosody, starkly contrasting with the rest of TS speech. Because of its impulsive and spontaneous nature, coprolalia in TS is characterized as ‘reactive language’ [96, p. 88], being comparable to the outbursts referred to as automatic speech, or automatisms in aphasic studies (e.g. [62,97,98]). This type of language relates to the altered inhibition of specific brain networks (see §5 for details) and contrasts with ‘intentional’ language, which gets impaired in e.g. Broca's and global aphasias as the result of damage in other brain areas, particularly those involved in working memory processing. Concerning the higher incidence of physical aggression/violence among TS individuals (iv), Ganos and colleagues [99] reported that explosive outbursts/rage attacks are commonly observed in TS (see also [100–102]). When it comes to incidence of aggression with specifically those TS individuals exhibiting coprolalia, Freeman and colleagues [103] found that anger control problems are associated with nearly 35% of cases. This kind of aggressive behaviour is clearly reactive and can be ascribed to the lack of restraint/control that characterizes typical populations, as discussed in §5. Finally, concerning the anxiolytic property of tics (v), there are numerous reports stating that individuals with TS experience an increase in their tics under stressful conditions, which are accompanied by a sense of compulsion/urge that gets relieved by tic performance (e.g. [104–107]). It is also of interest that a subset of TS patients exhibit heightened reactivity to stress of the HPA axis [108], which is an important dimension of SD, as discussed in §2. Children with TS show higher cortisol levels in response to stressors, which is also indicative of an enhanced HPA responsivity to stress [109]. However, contrary to what is observed in ASD, they seem to return to normal diurnal cortisol rhythm of the axis following the acute stress response [110]. As discussed in Progovac & Benítez-Burraco [28], coprolalia, as manifested in some TS individuals, may provide a unique window into the ancient struggles to channel stress/tension into verbal (as opposed to physical) aggression, illuminating an important transition in human prehistory.

(c). Synaesthesia

Synaesthesia is a condition (not typically considered a disorder) where stimulation of one sensory or cognitive pathway leads to automatic, involuntary experiences in another sensory or cognitive pathway [32,111]; e.g. a synaesthete may experience a specific colour whenever he/she hears a particular tone (e.g. C-sharp may be blue), sees a particular number (e.g. ‘5’ may be green and ‘6’ red), or reads a particular word or letter (e.g. ‘T’ may be blue). In other cases, certain words can evoke specific tastes, such as the word ‘basketball’ inducing the taste of waffles, or the word ‘blue’ tasting ‘inky’ [112]. Accordingly, one can identify different types of synaesthesia, including, but not limited to, grapheme–colour; sound–colour and lexical–taste synaesthesia [113].

The following findings regarding synaesthesia are of direct relevance for our proposal: (i) the uncontrollable (involuntary) nature of synaesthetic experiences; (ii) the enhanced cross-modal abilities exhibited by synaesthetes, including language abilities; and (iii) the impact of synaesthesia on social abilities. With regards to the uncontrollable nature of synaesthetic experiences (i), it should be stressed that synaesthesia results in automatic, involuntary experiences in another sensory or cognitive domain. Concerning cross-modality (ii), as noted by Cuskley & Kirby [33, p. 871], ‘synesthetes should be viewed as super cross-modal associators, with unusually strong, stable, and specific cross-modal biases.’ Synaesthetes exhibit a variety of advantages when it comes to memory, including an enhancement of both visual and verbal memory [114,115], as well as much better performance on episodic memory ([116]; see also [117] for a recent review). These advantages seem to have a positive effect on their language abilities, too. In this respect, Papadopoulos and colleagues [118] found that synaesthetes are faster in their lexical decisions than controls, positively impacting upon semantic processing (see also [119]). Although this difference did not reach statistical significance in their study, it is in line with previous findings that synaesthetes benefit from sensations in other domains (e.g. colours) in auditory lexical decision tasks, reinforcing their superior verbal memory (e.g. [120]), as well as showing an enhanced performance for expressive and receptive vocabulary compared with controls [121]. Likewise, synaesthetes seem to excel on verbal creativity tasks, with scores on synaesthesia correlating with scores on the Remote Associates Test [122].

Relevant for our discussion here, sound-symbolic effects (such as those found in bouba–kiki experiments discussed below) have been considered to be synaesthetic in nature [63]. This extends to naturally occurring sound-symbolic words, i.e. ideophones. In an EEG experiment, Lockwood & Tuomainen [123] found that Japanese ideophones elicit a specific brain response not found in non-ideophonic adverbs [124,125], which is consistent with abundant linguistic literature reporting on the ‘vivid experience’ of ideophones (e.g. [126]). Accordingly, after recognizing a word as an ideophone based on its phonological salience, a second sensory system is activated by association that results in an integration process, which is regarded as (quasi-)synaesthetic, but that some authors regard as plainly synaesthetic (e.g. [118,127]). It is thus not surprising that synaesthetes would have superior understanding of unfamiliar sound-symbolic foreign words, as found in Bankieris & Simner's [128] experiment, suggesting that sound symbolism relies on the type of cross-modal integration that characterizes synaesthesia to an exaggerated degree. Overall, the enhanced cross-modality in synaesthesia has been ascribed to the hyper-connectivity of a synaesthete's brain ([114,115,117,129]; more on this in §5).

Finally, there is evidence that synaesthesia can impact upon social cognition and behaviour (iii). For example, synaesthetes experiencing grapheme–colour synaesthesia outperform controls in fine visual discrimination of facial identity and emotion [130]. Interestingly, a social bouba–kiki effect has been identified as well, in which people associate proper names featuring rounded sounds (e.g. Bob) with round-faced (versus angular-faced) individuals [131]. As noted by the authors, this effect might have some social relevance, as it entails that social judgement involves not only amodal application of stored information, but also cross-modal integration of perceptual and bodily input. Likewise, in mirror-sensory synaesthesia, in which the pain or touch that synaesthetes observe in other people is mirrored on their own bodies, imitation is also enhanced [132]. Interestingly for our discussion here, synaesthesia can interact with the management of stress/aggression. For instance, synaesthetic experiences have been linked to (and seemingly triggered by) stressful experiences, as in the association found between grapheme–colour synaesthesia and post-traumatic stress disorder (PTSD) [133,134], as well as an association found between synaesthesia and anxiety (e.g. [117,135]).⁵

4. Cross-modality and aggression: overlaps and intersections among the disorders

Still, the picture is more complicated than the one depicted so far. Far from being independent conditions with clear-cut, condition-specific features, SZ, ASD, TS and synaesthesia can co-occur in patients. In other words, each of these conditions is complex, with comorbidities, and each grades into neurotypicity. Beginning with synaesthesia, it has been reported to be comorbid with ASD, with more than chance probability [117,137–139]. ASD shares with synaesthesia bias towards detail processing, as well as atypical sensory sensitivity, such as hyper- or hypo-sensitivity to certain sounds, textures or smells, visual stimuli, and pain and temperature [73,139]. On the other hand, whereas subjects with ASD typically show lack of imagination, synaesthetes exhibit noteworthy creativity and imagination. Differences can be observed as well in the domain of language, as synaesthesia is not linked to deficits in linguistic functioning, the way ASD is, the latter entailing problems both with certain structural aspects of language and with figurative language, as mentioned in §3.

This suggests some parallels in turn between synaesthesia and SZ, given that vivid imagination is a hallmark of SZ as well, although with this condition it seems more severely blurred with reality. In this sense, various kinds of hallucinations experienced by individuals with SZ may be considered as more heightened states of synaesthesia.⁶ Conversely, as noted by Meier & Rothen [113], the more vivid imagery visualizer style in synaesthesia is consistent with reports of hallucinations and fantasizing (see also [140,141]), and difficulties in dissociating dream imagery from reality [113, p. 3]. In fact, Banissy and colleagues [140] found that synaesthetes score higher on positive schizotypy. In this respect, recent genetic analyses have found a very slight association between SZ polygenic scores and synaesthesia, although no significant association was found between synaesthesia and scores related to ASD [142]. Finally, synaesthesia and TS have also occasionally been found to co-occur in patients, with synaesthetic experiences interacting with, provoking, or being driven by tics and obsessionality [143].

Continuing with TS, there is also some evidence of comorbidity between this condition and SZ, particularly, childhood-onset SZ [144]. Similarly, TS and ASD frequently co-occur in patients, with rates of comorbidity ranging from 3 to 30%, and with affected people exhibiting several clinical features of interest, most notably speech abnormalities (e.g. echolalia), sensory abnormalities (positing some parallels with synaesthesia, too) and social deficits, including increased aggressive behaviour [145,146]. Concerning specifically the language domain, high-functioning autists resemble individuals with TS in their enhanced propensity for hyper-systematicity (e.g. [147]), i.e. for producing rule-governed forms, including over-regularized past-tense forms (e.g. slip–slipped, plim–plimmed, bring–bringed). Regarding the social sphere, people with TS experience disturbances in social functioning that resemble those experienced by subjects with ASD ([148]; but see [149] for some differences in this respect).⁷

Very broadly speaking then, we can see trade-offs, i.e. complementary effects, among these four disorders, which exhibit some exaggerated effects of the otherwise beneficial traits for language evolution, including enhanced cross-modal thinking and enhanced rule-governed systematicity.⁸ Specifically, while TS and ASD seem to show certain advantages with systematic, rule-governed language, synaesthesia (and possibly SZ, at least relatives of those with SZ) shows advantages with metaphorical extensions and creativity with language. It is important to emphasize here that both of these dimensions are critical for language, and that these conditions only represent the (visible) poles/edges of the otherwise continuous cline of variability that exists in human populations, arguably as a result of the continued selection for facility with language.⁹ If so, then the understanding of these disorders, as well as the understanding of human evolution, needs to take into account these linguistic considerations. In §§5 and 6, we interpret the findings regarding this wide-ranging variability in terms of an altered balance between inhibition and disinhibition in the cortico-striatal brain networks.

5. Cross-modality and aggression: a delicate balance between inhibition versus disinhibition

It is beyond the scope of this paper to discuss in any detail the aetiological basis of the (partial) comorbidity found among the four disorders considered here, as multiple genetic, neurobiological and even environmental factors have been invoked to account for the shared symptoms (but also for the clinical differences). Nonetheless, in this section we wish to highlight one important aspect of this (partially) shared aetiopathology, relevant for illuminating aspects of language(s) evolution: the altered balance between inhibition and disinhibition observed in all these disorders, in the same brain circuits implicated in aggression, cross-modality and language processing.

Beginning with synaesthesia and ASD, an enhanced interaction between locally close sensory areas has been suggested to lead to an altered perception in both conditions [138], with this hyper-connectivity seemingly resulting from abnormal neuronal pruning (for synaesthesia, see [154,155]; for ASD, see [156]). More generally, individuals with ASD differ from typical individuals by being biased towards local, at the expense of global, conceptual processing, leading to difficulties in generalizing information, i.e. to a weakened drive to detect meaning by looking at the ‘big picture’ [157]. In other words, ASD can be characterized not solely as involving exuberant local connectivity, but also as involving impoverished global connectivity [139,158].

By contrast, the so-called two-stage model of synaesthesia proposes that a combination of both increased local connectivity between sensory brain regions and modulation of these connections by higher-order areas may be a mechanism of synaesthesia [159,160]. According to Grossenbacher & Lovelace [161], synaesthesia is a result of disinhibited cortical sensory feedback. Cytowic [162] emphasizes limbic mediation and cortical depression as a mechanism for synaesthesia. One possible explanation for this hyper-connectivity in synaesthesia is to invoke a failure of adequate pruning of exuberant brain connections both locally and globally, the pruning that typically takes place in ontogeny (e.g. [154]). In other words, one possibility for accounting for synaesthetic experiences is to ascribe them to a failure to inhibit the early hyper-connectivity among brain regions, which then survives into adulthood [63,163–165].

The increased local and global connectivity of the brain characterizing synaesthesia seems to be causing the enhancement of both visual and verbal memory and possibly other cognitive and linguistic functions as well, as discussed in §3. These differences in local versus global connectivity apparently explain both the features shared with and the differences between ASD and synaesthesia, for example, why both ASD and synaesthesia are advantageous for the attention to detail (correlated with enhanced local connectivity), whereas only syneasthesia seems to also show advantages in creativity, metaphoricity and imagination (correlated with abnormally enhanced global connectivity).¹⁰ In this respect, SZ seems to involve an even more profound failure of inhibition of global connectivity than synaesthesia. In particular, Silbersweig and colleagues [166] ascribe SZ hallucinations in both auditory and visual modalities to the abnormal disinhibition of cortical–subcortical circuits.¹¹ In sum, both the vulnerabilities and the advantages exhibited by individuals with these conditions in the domain of cross-modality seem to result from an altered pattern of inhibition/disinhibition, either locally, or globally, or both.

A similar pattern emerges with regards to social functions, notably including aggression. Specifically, TS and ASD have been found to exhibit an interneuron dysfunction that gives rise to an altered degree of inhibition of specific cortico-striatal circuits, resulting in reduced control of striatal activity by cortical structures ([168]; see also [169,170]). The lack of proper control seems to contribute to atypical degrees of reactive aggression in both ASD and TS, as well as to involuntary verbal aggression (including coprolalia in TS; see below for speech automatisms in aphasia).¹² It is of special interest that both enhanced cross-modality and (the suppression of) reactive aggression, including verbal aggression, rely on a precise degree of (dis)inhibition of connectivity in the cortico-striatal brain circuits, the same circuits that are also essential for the processing of grammar, and language more generally. This leads to our proposal that improved cross-modality, as observed in present-day humans, was partly a result, or a side-effect, of the SD processes that initially favoured specifically the dense connectivity in this network, enabling better control of reactive aggression by the higher, cortical structures of the brain (see §§6 and 7 for a more complete characterization, which takes into account the feedback loop between these biological developments and the cultural emergence of simple forms of language/grammar). This begins to explain why the two phenomena, altered reactive aggression and altered cross-modality, cluster together in these cognitive disorders.

Roth & Strüber [173] found that reactive aggression is mostly associated with volume reductions and/or activity decreases of frontal brain structures together with the hyperactivity of the amygdala. Worbe and colleagues [174] have related TS to cortico-basal ganglia network immaturity, the network also implicated in physical aggression and language processing (see also [35,175]). The damage to these networks also leads to a specific kind of disinhibition in aphasia, which results in the production of speech automatisms, including expletives, i.e. verbal aggression [98,176,177]. According to Van Lancker & Cummings [90], the clinical-linguistic observations suggest that there is release from inhibition of emotionally and sexually charged behaviours in TS (see also [107]), which is consistent with Ludlow and colleagues' [178] suggestion that TS may involve an imbalance between cortical and limbic systems, where the stereotypic speech and other behaviours get produced without inhibition by the cortical system.

As noted in §2, the emergence of our ability for cross-modal associations (upon which our language abilities largely rely) was previously hypothesized to have resulted from some changes in brain wiring leading to an increased connectivity of distant regions (see [38] for details). However, as also discussed by Boeckx & Benítez-Burraco [38], a completely unrestricted merging ability is cognitively unhelpful, and therefore requires inhibition. As suggested by Kircher & Glendenning [179], it is the ability to inhibit the cortex that has resulted in increased ‘intelligence’ in humans. We can now directly link this hypothesized (re)wiring in the brain to the suppression of reactive aggression, as directly targeted by the forces of SD, as well as to the emergence of simple forms of language/grammar, as explained in the following sections.

The optimal (or typical) balance between inhibition and disinhibition is achieved by a higher degree of cross-modal connectivity in childhood, beneficial for language learning, but a higher degree of inhibition of such connectivity (possibly via pruning) in adulthood, which can be beneficial for consolidating knowledge; overall this pattern may be the crux of the existence of a critical period for language acquisition (e.g. [180]; see, however, [181] for a critical review of related notions, as well as for further references). In this respect, Watson and colleagues [182] found that synaesthesia is more prevalent in populations in which language learning is prolonged for some reason, whether it is because they are multilingual, or because their writing systems involve complex phoneme–grapheme mappings. Because there are trade-offs, this delicate balance of inhibition versus disinhibition, i.e. proliferation versus pruning, may be tipped in either direction in atypical conditions (e.g. ASD, SZ, TS, synaesthesia), resulting in different cognitive outcomes, where typical and atypical outcomes are on a graded scale/cline.

Our proposal is consistent with the view that recent evolutionary innovations are more susceptible to ontogenetic damage because of their reduced resilience [183,184], given the so-called ‘last in, first out’ principle, according to which what is acquired last is the most shallow/fragile layer that is the easiest to lose (e.g. [176,185]). According to Pattabiraman and colleagues [184], our unique mental abilities arise in part from changes in the organization and function of pre-existing neural circuits, and these human-specific brain features are implicated in neurodevelopmental and neurodegenerative disease risk. In this respect, significant strengthening of neuronal connectivity of cortico-striatal networks, found altered in conditions like TS, ASD and SZ, occurred in relatively recent evolution, in the line of descent of humans, implicating FOXP2 and other genes (e.g. [153,186–189, 190, p. 37]). Such connectivity seems to be instrumental not only for exerting control over various physical impulses/reactivity, but also for language processing in general, reinforcing the main idea that the evolution of language, including cross-modality, was intimately linked to the taming of reactive aggression, i.e. to the SD processes, as further highlighted in the following two sections.

6. What accounts for the intimate link between cross-modality and aggression in language evolution?

As shown so far, there are multiple dimensions along which these four disorders can be mapped, including cross-modality, non-figurative and figurative language abilities, brain connectivity (global or local), and aggression (reactive or proactive; verbal or physical), and all of them ultimately point to altered inhibition/disinhibition patterns, involving the same cortico-subcortical brain circuits, in particular cortico-striatal networks. Table 1 summarizes how the dimensions considered in this paper present across SZ, ASD, TS and synaesthesia. To these considerations, we also add gender differences, which implicate sexual dimorphism, also impacting upon brain anatomy and function. This is relevant because a reduction in sexual dimorphism is a frequent outcome of domestication processes [191], and sexual dimorphism seems to have attenuated in humans, in particular during the period we focus on here, that is, the period starting around 200 ka or earlier, and stretching to around 50–40 ka ([192]; for the reasons, see [28]). Needless to say, this table is only a preliminary, partial attempt at deconstructing these cognitive conditions into their (evolutionary) primitives.

Table 1.

Features of interest across the conditions examined.

	cross-modality:global	cross-modality:local	literal language	metaphorical language	reactive aggression	proactive aggression	social function	sexual dimorphism
ASD	reduced	increased	impaired	impaired	enhanced	reduced	reduced	high male prevalence
SZ	increased	possibly reduced	somewhat impaired	impaired	slightly enhanced	enhanced	somewhat enhanced	only slight male prevalence
TS	reduced	possibly increased	not impaired, possibly enhanced in some domains	impaired	enhanced	(not enough data)	reduced	high male prevalence
synaesthesia	increased	increased	not impaired, enhanced in several domains	not impaired, possibly enhanced	(not enough data)	(not enough data)	somewhat enhanced	possible female prevalence, but not high

Our contention here is that the crucial dimensions along which the cognitive disorders/conditions discussed in the paper can be mapped are unified under the postulates of the SD hypothesis, thus helping complete the complex puzzle only partially explained by Crespi & Badcock's [36] model (§3). In our previous work, we proposed that ASD can be characterized as exhibiting some features of a hypo-domesticated condition [30], contrasting in this respect with SZ, in which features of human SD appear more prominent [31]. In this paper we have considered two additional conditions, namely, synaesthesia and TS, which add important detail to this scenario, enriching the analysis. Based on the evidence presented in §§3–5 we contend that TS can be construed as a hypo-domesticated type, in that it shows some parallels with ASD, while synaesthesia can be construed as a hyper-domesticated type, coming closer to SZ in this sense. In this respect, TS exhibits the following attenuated features of SD: increased reactive aggression, both physical and verbal; social deficits; an increase in the reactivity of the HPA axis; as well as pronounced gender bias/sexual dimorphism in favour of males; all of which are hallmarks of hypo-domestication.¹³ On the other hand, synaesthesia does not show any handicaps of hypo-domestication, but instead shows advantages in language, memory and certain social aspects, with a propensity for imagination and creativity, although in some cases exhibiting the blurring of what is real versus imagined, which brings it closer to SZ, another hyper-domesticated phenotype.

In this respect, conditions such as ASD and TS can provide an excellent window into the earliest, initial stages of language evolution (with a high degree of reactive aggression, with diminished cross-modality, and with high levels of sexual dimorphism), while synaesthesia (and to some extent SZ) provide a window into a somewhat later stage, when high levels of connectivity in these brain networks would have resulted in exaggerated, super cross-modality, among a significant portion of the population.¹⁴ As shown in §5, TS and ASD can be seen as involving a lesser degree of connectivity in these global networks, correlating with a higher degree of reactive aggression (due to reduced cortical control over the subcortical regions), and with a lower degree of cross-modality, affecting figurative language and sound–meaning associations, especially in comparison with SZ and synaesthesia. On the opposite end of the continuum, synaesthesia and SZ can be seen as exhibiting an increased degree of connectivity in these networks, correlated with better control over reactive aggression, and with exuberant cross-modality. This is the sense in which these disorders are diametrically opposed to each other, and thus complementary to each other, as also seen in §4.

As SD was targeting the reduction/suppression of reactive aggression, it gradually contributed to enhanced global connectivity of the brain (so that cortical structures can exert better control over the subcortical structures). The emergence of cross-modality can be hypothesized to be, at least partly, a side-effect of the enhanced connectivity in these networks due to SD. Nonetheless, the path towards cross-modality is likely to have also been facilitated by the selective advantages of the earliest, simple forms of language with high emotional load, such as those introduced in §2. Such forms relied on crude metaphoricity, i.e. on emerging cross-modality, and would have provided clear selective advantages (e.g. the replacement of physical aggression by verbal aggression).¹⁵ This is where human SD crucially differs from that of bonobos and other species—in being intricately entangled with, and accelerated by, the (cultural) emergence of simple language forms, the two engaged in a mutually reinforcing feedback loop [28]. Domestication alone does not yield language, as is clear when one considers other domesticated species, just as culture alone, without the biological foundation, cannot yield languages with tens of thousands of words, and hundreds of grammatical rules and restrictions governing their combinations, as is also evident from failed attempts to teach human languages to primates.¹⁶ As we repeatedly point out, both the propensity for rule-governed communication (enhanced in ASD and TS) and the propensity for creativity and metaphoricity (enhanced in SZ and synaesthesia) are crucial for language, and our proposal suggests that elements of both aspects have been undergoing selection in human populations, with the present-day disorders/conditions only showing the outer edges/poles of the otherwise continuous cline of human variability related to language.

At the point when SD reached its peak, reportedly around 40 ka [192], we can hypothesize that cross-modality, metaphoricity and synaesthetic-like experiences would have reached their peaks, too. From 40 ka onward, SD started to decline. In parallel with signs of decreased SD, such as increasing sexual dimorphism [192], proactive aggression also seems to have enhanced during these more recent times [203]. Proactive aggression is characterized by reduced functional connectivity between selected brain regions, including bilateral precuneus [204], which acts as an important hub of brain organization, mostly related to associative and integrative functions between somatic and visual primary inputs [205], and which has expanded during recent human evolution, seemingly contributing to certain aspects of our cognitive specialization, including language [206]. It is thus plausible that these more recent changes have also contributed to the subsequent refinement and sophistication of metaphoricity and language more generally, via the refinement of the delicate balance between inhibition and disinhibition in these brain networks. As noted in §2, modern metaphorical language tends to be more sophisticated, and less directly linked to the realm of emotions and sensations. Likewise, proactive aggression is more rational than emotional in nature. This reinforces the view that during recent times the nature of human language has shifted from more emotional to less emotional expressions, in part owing to the effects of changes in our SD, impacting upon both aggression management and cross-modality.

7. Conclusion and future prospects

The most surprising, novel element of our proposal concerns the underlying mechanism that unites the three different dimensions in human evolution (cross-modality, aggression and language processing), facilitating an answer to the question of why they cluster together, including in cognitive disorders, and why they seem to have co-evolved. Our proposal implicates the cortico-striatal brain networks, whose enhanced connectivity and plasticity is a result of relatively recent evolution and selection (§5). As seen in the previous sections, an increased neuronal connectivity in these networks is a mechanism for the suppression of reactive aggression, via the enhanced cortical control of subcortical regions, but also a mechanism for cross-modality, and language processing more generally, via the enhancement of the dialogue between distant brain areas. Language needed both: the suppression of reactive aggression, in order to detach from the realm of emotions, which is supported by the global brain connectivity; and controlled cross-modality, in order to enrich its structure and use via metaphorization, which is also supported by global connectivity, as well as a precise degree of (dis)inhibition of such connectivity. An imbalance in this network, in particular an abnormal degree of inhibition of the relevant connectivity, results in increased reactive aggression, but also in diminished cross-modality, i.e. metaphoricity, as attested in, e.g. ASD and TS. On the other hand, an atypical level of disinhibition in these networks can lead to enhanced cross-modality, i.e. enhanced metaphoricity, imageability, creativity and even hallucinations, as attested in synaesthesia, and to some extent also SZ, as well as possibly to altered aggression/anxiety levels.

Given the ‘first in, last out’ principle (§5), the high prevalence of both abnormal aggression and altered cross-modality among disorders implicating language is not surprising, but expected under this view, considering that the delicate balance that regulates these behaviours is a recent evolutionary development. What is surprising and unexpected, on the other hand, is that these disparate symptoms cluster together in these cognitive conditions, and this is the puzzle that we sought to solve in this paper. Ontogeny (even abnormal ontogeny) does not exactly recapitulate phylogeny. Nonetheless, for the reasons provided above (see [207] for ample discussion), these cognitive conditions can be used as useful proxies or windows to previous stages and different aspects of our cognitive and linguistic evolution, especially when considered in conjunction with linguistic proxies of early language evolution (e.g. [28,45]), helping us reconstruct several important aspects of early prehistoric languages.

In this respect, it is reasonable to project that the prehistoric phenotype that saw the earliest stages of language/grammar in our species (roughly dated to around 200 ka or earlier, for the reasons offered in [28]) was characterized by a brain with limited cortico-subcortical connectivity, correlating with high levels of reactive aggression, and low levels of cross-modality, as well as by high levels of sexual dimorphism, which, in these respects, approximates the ASD phenotype, and the TS type to some extent.¹⁷ This phenotype started to use the expressions such as the ones identified as linguistic proxies/'fossils’ in §2. Once the global connectivity was enhanced, as a result of SD (entailing more control by cortical regions), but also as a result of the pressure to process this type of early metaphorical language, we hypothesize that a mental architecture characterized by enhanced cross-modal thinking would have emerged, and that mental conditions characterized by exuberant connectivity, such as synaesthesia and (to some extent) SZ, would have significantly increased in prevalence. The latest step of this process would have entailed a prolonged struggle to fine-tune a delicate balance between inhibition and disinhibition in these brain circuits, the kind of balance that we encounter in typical cases in modern humans, with their greater detachment of language from emotions and their increased attachment of language to propositional content, this resulting in more complex grammars and less iconicity. At the same time, extreme phenotypes within this continuum, including ASD/TS and SZ/synaesthesia, which would have once been much more prevalent, are now reduced to marginal rarities (e.g. according to [117], the prevalence of ASD is about 1–2%, and that of synesthaesia, 4%). Nonetheless, as can be seen from the persistence of these features among the modern populations, this process of fine-tuning has not been completed, and probably never will be, as individual variability in these realms provides some beneficial trade-offs, as well as a possible source of future adaptations and innovations.¹⁸

In other words, we contend that language evolution has been characterized by a fine-tuning of inhibition/disinhibition in the cortico-subcortical circuits, in particular cortico-striatal networks, through a continued feedback loop between biological forces, such as SD and selection, and cultural processes, such as the emergence, as well as complexification, of (the uses of) rudimentary language/grammar, illustrated with the reconstructed linguistic proxies/'fossils’ in §2, resulting in the typical delicate balance attested in modern humans. As pointed out in the previous section, just SD alone does not yield language, and just culture alone, without the biological foundation, cannot yield languages with their present-day enormous richness and complexity. It takes a complex and continuous feedback loop, an intricate back and forth, between our genes and our grammars, and the uses to which they are put, and our goal in this paper was to identify just the cultural and genetic developments that would have fruitfully engaged in such a loop, and thus significantly ratcheted up the process of language evolution. What is of particular interest in our proposal is the identification of a common underlying cause, i.e. atypical (dis)inhibition of connectivity in the cortico-striatal networks, which begins to explain why phenomena as disparate as aggression and cross-modality/metaphoricity nonetheless cluster together in cognitive disorders affecting language. We have ascribed this clustering to the common evolutionary cause, related to human SD, and the specific way in which it interacts with the emergence of simple forms of language.

The specific details that our proposal provides make it possible to seek answers to many questions that surely remain, offering fertile ground for hypothesis generation and experimental testing, including genetic studies, neuroimaging experiments and psycholinguistic tasks. For example, one can look for candidate genes for the disorders highlighted in the paper among the genes affected in recent evolutionary processes. This should enable us to identify candidate genes for important steps leading to human cognitive and behavioural evolution. We agree with the conclusion in van Leeuwen and colleagues [117, p. 4] that there should be serious, large-scale research that directly compares autism and synaesthesia, and we would like to add that such research should take into account the relevant linguistic considerations, as well as evolutionary considerations, which have, as we have argued, played a key role in the evolution and retention of variability in humans, including the poles such as autism and synaesthesia. In this respect, one can examine how individuals with these conditions process linguistic items of interest, particularly derogatory expressions and ideophones. For example, one can continue to probe the question of whether synaesthetes are better at appreciating ideophones even in languages they do not speak, in contrast to those with other disorders, as well as controls, who are predicted to score much lower than synaesthetes on this task. Likewise, one can study the effects of coprolalia (or coprolalic-like language) on the cortisol levels of individuals with conditions such as TS, in hope of getting a better glimpse into how physical tension might have been channelled into verbal aggression, and verbal behaviour more generally, in prehistory. Given the cross-relatedness of these conditions and the overlap of symptoms, such experiments should target all the affected groups, i.e. the processing of both derogatory language and ideophones should be tested in all four conditions considered in this paper, against the backdrop of controls. Neither of these linguistic phenomena has received any significant scientific attention in clinical conditions. Experiments like these will not only lead to a better understanding of language disorders, but will also shed further, more specific light on human prehistory and language evolution.

Data accessibility

This article has no additional data.

Authors' contributions

A.B.-B.: conceptualization, data curation, investigation, writing—original draft, writing—review and editing. L.P.: conceptualization, data curation, investigation, writing—original draft, writing—review and editing.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.