Opinion on monoaminergic contributions to traits and temperament

Abstract

This article critically reviews evidence relating temperamental traits and personality factors to the monoamine neurotransmitters, especially dopamine and serotonin. The genetic evidence is not yet considered to be conclusive and it is argued that basic neuroscience research on the neural basis of behaviour in experimental animals should be taken more into account. While questionnaire and lexical methodology including the ‘Five Factor’ theory has been informative (mostly for the traits relevant to social functioning, i.e. personality), biologically oriented approaches should be employed with more objective, theoretically grounded measures of cognition and behaviour, combined with neuroimaging and psychopharmacology, where appropriate. This strategy will enable specific functions of monoamines and other neuromodulators such as acetylcholine and neuropeptides (such as orexin) to be defined with respect to their roles in modulating activity in specific neural networks—leading to a more realistic definition of their interactive roles in complex, biologically based traits (i.e. temperament).

This article is part of the theme issue ‘Diverse perspectives on diversity: multi-disciplinary approaches to taxonomies of individual differences’.

1. Introduction

There have been several attempts to expand the ancient theory of four ‘temperaments’ by linking them to modern psychometric dimensions of biologically based traits and to neural and neurochemical systems. The latter reduces to chemical neurotransmitter systems such as those of the monoamine (dopamine (DA) and noradrenaline (NA), serotonin, (5-hydroxytryptamine, 5-HT), cholinergic and other neuromodulatory systems (including several neuropeptide neurotransmitters)). While several pioneering investigators realized that these systems undoubtedly modulate relevant defined traits of temperament [1–6], it remains problematic to assert isomorphic, linear relationships with these systems. For example, Cloninger et al. [4] proposed associations between monoamines and temperament traits of Harm Avoidance, Novelty/Sensation seeking/Impulsivity and ‘Reward Dependence’ as measured by his Temperament and Character Inventory (TCI). On the basis of neurochemical and psychopharmacological evidence in humans and animals, and the limited genetic data then available to him, Cloninger assigned Harm Avoidance to variations in the 5-HT system, Novelty Seeking to low activity in the DA system and Reward Dependence to low activity in the NA system. (A fourth dimension, Persistence, was added later). Subsequent work has considered the notion that DA is implicated in Impulsivity or Extraversion [5], and that 5-HT (and possibly NA) is implicated in a ‘Behavioural Inhibition’ system [1]. Such hypotheses have undoubtedly been stimulating and have led to some intriguing findings; however, the human studies have often been based on rather crude and even non-specific pharmacological manipulations. Theories have not always taken account of all of the published evidence available, especially concerning the organizational heterogeneity of the systems, and many of the relevant findings emanating from animal behavioural neuroscience.

First, these monoamine systems are in fact often divided into anatomically segregated subsystems, for example, the mesolimbic, mesostriatal and mesofrontal DA systems; the dorsal and median raphé 5-HT systems, the ascending locus coeruelo-cortical and lateral tegmental NA systems. These systems have different terminal domains (e.g. neocortex and hypothalamus, respectively) and, therefore, contribute to different information processing systems, which may contribute to several behavioural dimensions, rather than being monolithically related just to one. Most of the evidence [7] also recognizes that any relationships with neurochemical activity cannot necessarily be modelled in terms of simple linear or monotonic functions relating behaviour to neurochemical activity. A second major factor is that the neurochemical systems not only interact between or amongst themselves, e.g. DA-5-HT interactions in ‘reward’ functions, but their subsystems also interact, sometimes in an opponent fashion (e.g. subcortical and frontal DA systems). Therefore, because of this opponency, to talk in terms of ‘low DA’ activity is somewhat meaningless unless the overall output of the system can be determined or it can be related precisely to the modulation of particular neural regions or networks. Increasing evidence also suggests that these systems work in different modes, e.g. fast phasic and tonic modes, in the case of the DA and NA systems at least; thus, it is necessary to be precise about what one means in terms of increased or reduced activity in the system. Third, the net activity of a system is also determined by a variety of interacting molecular structures including specialized receptors, which may have quite different (and even opposing) functions, e.g. inhibitory autoreceptors. This is even more likely if they are also located in diverse neural networks and applies to all of the major monoaminergic and cholinergic systems—though particularly for the 5-HT system, which has no fewer than about 16 distinct receptor types [7].

These considerations lead us also to consider genetic factors; the various molecules including amino acids, neuropeptides and structural proteins making up the neurotransmitter system are all genetically determined. This level of analysis may also be most appropriate when considering temperamental or trait-like constructs with biological bases rather than personality per se, which is more obviously related to the additional effects of learning and socio-cultural factors. Some progress has certainly been made to suggest that genetic polymorphisms affect human behavioural traits. For example, (i) harm avoidance in response to aversive stimuli (as measured by the TCI) in the case of ‘long’ and ‘short’ armed alleles of the 5–HT promoter transporter and (ii) novelty/sensation–seeking (as measured by the Tridimensional Personality Questionnaire (TPQ) Novelty Seeking Scale) in the case of a polymorphism of the D4 receptor. However, it is not so clear how (i.e. by which mechanisms) the genetic polymorphism actually affects the activity of the neurochemical system in either case. For example, several hypotheses, including Cloninger's [4], have postulated that ‘low DA’ is related to novelty seeking and impulsivity in humans, and that a D4 polymorphism would reduce the impact of DA in particular neural networks. However, low novelty preference in rats is correlated with low extracellular DA levels in the striatum [12] and low levels of midbrain inhibitory DA autoreceptors— and hence likely enhanced DA neuron activity—are correlated with increased novelty-seeking in humans [13]. A related significant observation is that, although indirect agonists of DA such as methylphenidate or d-amphetamine reduce impulsive behaviour in disorders such as attention deficit hyperactivity disorder (and also in rats with chronically high levels of impulsive behaviour), these drugs actually increase impulsivity in experimental rodents with low baseline levels of this behaviour [14].

Overall, there is rather little evidence from genetic studies to support strong one-on-one relationships of traits with neurotransmitter function [8]. This may not be surprising given the likely polygenic basis of most complex traits, including those related to temperament.

This article will re-appraise what has been concluded with respect to neurotransmitters and traits. Rather than simply relating crude measures of neurotransmitter function in humans to personality, i.e. socially influenced traits (including the lexically derived ‘Big 5’ personality factors of Extraversion, Neuroticism, Openness/Intellect, Conscientiousness and Agreeableness [15]), it will advocate the use of more objective measures of behaviour that may be able to relate more directly to animal studies that employ the sophisticated methods of basic neuroscience to assess chemical neuromodulation (see also [7]). This is partly because it is suspected that the Five Factor Model and subjective questionnaire data, while useful and of interest, are not always strongly related to more objective neurophysiological and behavioural measures. Problems of biases provided by the social use of language and subjective social desirability, as well as measurement stability, may outweigh the convenience provided by self-report questionnaires. It is anticipated that several of the original proposals will eventually require modification in the light of these considerations, and more detailed information on neurotransmitter function.

2. Dopamine in reward sensitivity and extraversion: not the full story

Inspired by original work on DA as an essential modulatory component of the ‘reward system’, and bolstered by the more recent discoveries of Schultz [16] and others on the role of DA cells in mediating plasticity via prediction errors for reinforcement learning, many theorists have attempted to relate this neurotransmitter to one or more of the constructs described above, generally by focusing rather narrowly on the mesolimbic projection to the nucleus accumbens. However, the midbrain DA cells also project to the dorsal striatum and prefrontal cortex as well as the nucleus accumbens, which itself is divided into sub-regions (‘core’ and ‘shell’) with different afferent projections and likely distinct functions that touch on several of the constructs listed above, as well as additional functions that have not been as well characterized as human traits.

There have been several suggestions by several personality theorists to link responses on the Behavioural Approach System (BAS) scale and Extraversion scales to approach or ‘reward’ functions dependent upon mesolimbic DA [1,5,15–19]. Moreover, several theorists [18,20] have noted that tests designed to measure reward approach behaviour and extraversion often also measure impulsivity, a tendency to unduly risky or premature responding. Indeed, in humans, low D2/3 receptor (D2/3–R) availability in the midbrain measured using PET predicts Barratt Impulsiveness Scale (BIS) scores of impulsivity; low numbers of D2/3R there are also associated with elevated DA release in the striatum [21].

Recently, premature approach behaviour by rats to stimuli predicting food reward (as measured in the 5-choice serial reaction time task, 5-CSRTT) has been defined as a form of ‘waiting impulsivity’ that has been associated with reduced D2/3 receptors and DA transporter (DAT) function in the shell region of the nucleus accumbens [22]. This behaviour is generally exacerbated by drugs such as amphetamine or methylphenidate that enhance DA levels in the nucleus accumbens and are blocked by lesions of the shell region. In vitro studies of DA function have indicated enhanced DA release in the shell but reductions in the core in ‘high impulsive’ animals [22]. By contrast, another measure of impulsivity, impulsive choice in the delayed discounting paradigm, has been also associated with reductions of DA release in the core region [22]. This is also consistent with the fact that the high impulsive rats also exhibit steeper delayed discounting, i.e. more impulsive choice. Importantly, amphetamine generally reduces impulsive choice (an effect that appears to depend on both DA and 5-HT receptor mechanisms) and so presumably this form of impulsivity is reduced by elevating mesolimbic DA. In rats exhibiting high levels of impulsive premature responding, stimulants such as methylphenidate and cocaine have also been found to alleviate impulsivity while also upregulating D2/3 receptors in the striatum [14,22]. The opposite effect is shown in normal (or low) impulsive rats, indicating that the effects of the drugs (and therefore of DA itself) are ‘baseline-dependent’ ([14], see review [22]). This concept of ‘baseline’- (or rate-) dependency is probably an important functional correlate of behavioural traits; it is upon this prior background of neurotransmitter activity that acute challenges by drugs and environment factors operate. Overall, it is apparent that impulsivity can be related to reductions of DA function in the core region, but also increases in the shell, which makes it difficult overall to argue for a simple monotonic relationship of reduced accumbal DA to ‘impulsivity’ [22].

In terms of personality traits, impulsivity has been associated with Extraversion and a separate factor of cognitive control or constraint [3]. Extraversion itself has been equated with incentive motivation and the sensitivity of the individual to reward and reward-related stimuli by several authors [1,5,17]. However, high impulsive rats are not in fact necessarily most sensitive to reward-related stimuli. Premature or ‘impulsive’ responding in the 5-CSRTT has been shown to be inversely related to responding for conditioned reinforcers such that those animals better able to use reward-related associations to inform responding in the environment are also better able to inhibit inappropriate motor responses [23]. High impulsive rats are also neither most likely to exhibit elevated locomotor activity in novel environments nor enhanced Pavlovian approach conditioning, although these forms of behaviour also undoubtedly both depend on DA-dependent functions of the nucleus accumbens [22]. However, there is a weak, though significant, relationship between impulsivity (measured as premature responding) and novelty preference [22] (relating probably most strongly to the concept of novelty/sensation seeking) that agrees with human data using PET showing correlations between midbrain DA D2 autoreceptor availability and novelty seeking [12] (TPQ-Novelty Seeking Scale) and paralleling the similar findings described above for impulsivity [21] (Barratt Impulsiveness Scale). BAS scores are also not simply equivalent to novelty seeking, as approach and novelty preference have often been dissociated in animal studies; for example, amphetamine can simultaneously enhance the former and reduce the latter [24]. Such data strongly suggest that these overlapping, but distinct, traits are related to variations of DA release in different dorsal or ventral striatal regions.

Segregation of Pavlovian approach responses into those rats that approach the stimuli predicting food (sign trackers) and those that approach the food location itself (goal trackers) has appeared to provide a better match to impulsivity, as sign trackers exhibit greater motor impulsive behaviour (on a timing schedule) [25]. However, sign trackers also exhibit reduced temporal discounting of reward, unlike high impulsive rats [25]. The pattern of behaviour of sign trackers is consistent with the finding in such rats of greater phasic DA activity in the nucleus accumbens core region [25]. Therefore, the equation of impulsivity with incentive motivation per se, as suggested by the proposed link between extraversion and impulsivity, does not appear to be correct. This is more generally consistent with our view that impulsivity and novelty/sensation seeking can also be dissociated, for example, in the context of vulnerability to addiction [26].

A further complication for the simple equation of DA with reward sensitivity, and hence extraversion, is that striatal DA also has important functions in yet another, motor, form of impulsivity as gauged by individual differences in stop signal reaction time performance [22]. This behaviour has been shown to be mediated by the dorsal rather than the ventral striatum (nucleus accumbens) [22,27], thus representing another form of impulsive behaviour that undoubtedly contributes to questionnaire scores or ratings of impulsivity. Data from human PET studies employing DA D1 and D2/3 ligands show that there is a clear relationship between D2/D3 binding and measures of behavioural inhibition, as distinct from behavioural activation and approach, in dorsal striatal circuits [28]. Pharmacological data in rats indicate that D2 receptor blockade in the dorsal striatum, but not the nucleus accumbens core, impair stop signal performance, confirming a general role for DA D2 receptors in impulsive behaviour [27], but also showing that impulsivity may be fractionated into different forms (‘waiting’ and ‘stopping’ impulsivity), mediated by distinct neural circuitries [29].

A review of PET studies of D2/3 receptor availability in relation to personality traits in humans in this issue [30] surveys several relevant studies pertinent to this discussion, but also reports a significant (and replicated) correlation with striatal D2 receptor availability with the personality trait of ‘social detachment’. This association illustrates possible associations of DA with other temperamental dimensions, besides novelty seeking and impulsivity, that may plausibly be related to psychoticism and social withdrawal, made plausible by the therapeutic actions of anti-psychotic D2 receptor drugs [30]. These effects are not clearly related to the dimensions of extraversion and impulsivity.

In summary, detailed studies of DA and temperament indicate a complex involvement in impulsivity, novelty seeking and incentive motivation, probably dependent on the precise neural networks engaged by these traits, including their corticolimbic components modulated by DA, as well as other factors, including the involvement of other modulatory neurotransmitters [22]. An earlier theoretical position pointed out that these overlapping traits all necessarily involved some degree of active behavioural output, labelled as ‘activation’, which captures much of the contribution of DA to behaviour, together with the plasticity associated with reinforcement learning [31]. Further progress might depend on theoretical developments, possibly including neurocomputational modelling approaches.

3. Serotonin in anxiety, neuroticism and behavioural inhibition: not the full story

Serotonin, or 5-HT, has traditionally been associated with anxiety or aversive motivation, neuroticism and associated states of behavioural inhibition, as shown, for example, by tests of punished responding in rats and humans; 5-HT depletion generally disinhibits such behaviour [32–34]. However, Deakin & Graeff [35] found that 5-HT could have opposing effects on reflexive aversive responses in rats such as freezing and escape, and instrumental avoidance responding motivated by anxiety, respectively, at 5-HT2 receptors in the brain stem and forebrain, thus apparently paradoxically enhancing anticipatory anxiety but inhibiting panic. Second, recent work has clearly shown that 5-HT has a role in positive reinforcement as well as punishment [36,37]. These examples make it clear that 5-HT does not have a unitary effect on a single dimension of behaviour. Serotonin has frequently been considered as an ‘opponent’ system to that of DA [38], and there is certainly some evidence of this; for example, in tests of impulsivity in rats dependent on premature responding in the 5-CSRTT, forebrain 5-HT depletion enhances impulsive responding [39].

However, there is considerable and mounting evidence to show that unitary conceptions of the functioning of the 5-HT system, while parsimonious, are far too simple. First, considering impulsivity, 5-HT2C receptor antagonism produced by SB242084 infused into the nucleus accumbens enhanced impulsive responding, but a 5-HT2A receptor antagonist (M100907) infused into either the nucleus accumbens [40] or the medial prefrontal cortex [41] had the opposite effect. Such findings have been corroborated and extended by using molecular genetics to track the expression of cortical 5-HT2A and 2C receptors [42]. Moreover, direct measures of 5-HT by in vivo microdialysis in the rat medial prefrontal cortex suggest enhanced release of the neurotransmitter in the mPFC to be associated with impulsive responding [43]. Like the mesostriatal and mesofrontal DA systems, there may be some reciprocity of function between cortical and subcortical systems (see also Dellu-Haggedorn et al. [44]). Second, 5-HT2C and 5-HT2A receptors also exert opposite effects on behaviour in tests of impulsivity and compulsivity, the detrimental tendency to repeat responding (perseveration) [45]. The 5-HT2C antagonist SB242084, while exacerbating impulsivity in the 5-CSRTT [39], also reduced perseverative responses in reversal learning tasks, whether injected systemically or infused into the orbitofrontal cortex [46]. Thus, the 5-HT2A receptor antagonist had an opposite profile of effects, inhibiting impulsive responding but enhancing compulsive responding [45].

These opposite effects indicate that a simple unitary account of these results in terms, for example, of behavioural inhibition is not parsimonious. However, it is possible that different circuits (especially fronto-striatal circuits) may be differentially implicated in impulsivity and compulsivity. There is mounting evidence that 5-HT depletion in the orbitofrontal cortex rather specifically impairs reversal learning, by enhancing perseverative responding in both the marmoset [47,48] and the rat (J. Alsio, O Lehmann, C McKenzie, JW Dalley, TW Robbins, unpublished observations). Whether this perseverative tendency can be considered ‘trait-like’ cannot yet be determined with certainty. However, it is significant that Barlow et al. [49] found that, in a population of Lister hooded rats, those with a tendency to perseverate in extinction had reductions in 5-HT, its metabolite, 5-hydroxyindoleacetic acid (5-HIAA) and in the density of 5-HT2A receptors in the lateral orbitofrontal cortex. This tendency to perseveration was ameliorated by systemic administration of citalopram, the selective serotonin reuptake inhibitor (SSRI). By contrast, there were no effects of the NA reuptake inhibitor atomoxetine. Clearly, 5-HT plays a role in moderating compulsive behaviour and promoting behavioural flexibility, possibly relevant to its implication in obsessive–compulsive disorder and ruminations in depression. It may further be relevant to a hypothetical trait of ‘compulsivity’ [50,51].

It is becoming clear from the examples provided above that any conclusions concerning the roles of the monoamines in temperament or personality have to make reference to the precise neural circuit implicated in relevant behavioural processes. Thus, it could be postulated that the neuromodulatory effects of 5-HT on measures of impulsivity and compulsivity depend on distinct fronto-striatal circuits; respectively, the medial prefrontal cortex and nucleus accumbens and the orbitofrontal cortex and dorsomedial striatum.

Somewhat analogous conclusions have been reached by Hariri [52] in considering the neurobiology and genetics of individual differences of complex behavioural traits in humans within a programme of functional brain imaging. For example, he describes the mapping of individual differences in trait anxiety onto threat-related amygdala reactivity, which in turn was related to serotonin signalling and thence to the influence of a common functional polymorphism (HTR1A-1019G allele). This allele, which one would expect to lead to reduced terminal 5-HT release as a consequence of increased density of auto-inhibitory receptors on dorsal raphé 5-HT neurons, is associated with significantly decreased threat-related amygdala reactivity, as measured using the functional brain imaging and the BOLD response. This general conclusion was suggested to be consistent with effects of other polymorphisms presumably similarly affecting 5-HT, the 5HTTLPR short allele and the MAOA low-activity alleles, if one assumes that these genotypes are indeed associated with increased 5-HT signalling. There is, however, some dissent with that view when a pharmacological rather than a genetic approach is used. We found that dietary tryptophan depletion (i.e. of the essential amino acid precursor of 5-HT) enhanced the responsiveness of an amygdala network for fearful face processing, though only in individuals showing self-reported vulnerability to threat (high BIS scores) [53]. These findings are consistent with those of a study examining effects of an acute low dose of citalopram (a SSRI), which can be expected to reduce 5-HT activity [54] via actions on raphé inhibitory autoreceptors. Thus, it is not entirely certain whether enhanced amygdala reactivity to faces in anxious subjects is related to enhanced or diminished 5-HT function (although it is certainly related to 5-HT function!) Resolving these basic issues will be necessary before dimensions of temperament or personality can be related definitively to neurotransmitter function; there are many such paradoxes still to resolve.

Recent work has attempted to link variations in 5-HT function in aversive situations to social dimensions. Thus, for example, an acute dose of citalopram enhanced the tendency to make ‘personal’ moral judgements and also to reduce ‘rejection’ responses in the Ultimatum Game—both of which could be construed as facilitating harm avoidance in the social context [55]. These effects were greater in subjects with high empathy scores, supporting views that 5-HT has functional interactions with hormones such as oxytocin, which have been linked to sociability. There is again the question of whether acute doses of citalopram actually reduce rather than enhance 5-HT function; however, in the Ultimatum Game, low tryptophan had the opposite effect of enhancing rejections [56], suggesting this interpretation to be valid.

Overall, attempts to link distinctive personality factors to serotonin function have shown that no single factor is related uniquely to parameters of 5-HT function (as inferred by relationship, for example, to genetic factors or PET receptor binding) [57]. Thus, for example, Frokjaer et al. [58] found in 83 healthy volunteers that frontolimbic serotonin 2A receptor binding was positively associated with the personality dimension of neuroticism, with still stronger relationships being found for the constituent traits of anxiety and vulnerability. A second study [59] also found a correlation of 5-HT transporter binding with neuroticism, but only in the thalamus. A rather different study found that an index of ‘Openness to Experience’, especially aspects relating to flexibility (Openness to Actions and Values), was negatively related to 5-HT transporter binding in the midbrain, putamen and thalamus [60]. This work thus intriguingly links well to studies suggesting a role for 5-HT in individual differences in perseverative behaviour [49], which however, specifically implicates the orbitofrontal cortex (see above).

By contrast, Allen & DeYoung [15] have postulated that serotonin is implicated in ‘Stability’ ‘representing shared variance of Conscientiousness, Agreeableness and low Neuroticism’, but this hypothesis appears to run against the evidence from the animal studies described above. It underlines the fact that it is difficult to come to any conclusion about a unitary function for serotonin, however general, that can account for all of the data, and agrees with the likely diverse influence of this neurotransmitter in different forebrain circuits.

4. Noradrenaline, acetylcholine and other neuromodulatory influences

Less evidence about relationships with temperament and personality exists concerning these systems. Much of the relevant animal and human evidence links NA to roles in selective attention [61], including the ‘exploit’/explore dichotomy [61], ‘alerting’ [6,62,63] or working memory [64], with some additional evidence suggesting that NA also mediates aspects of stress and is linked to anxiety and aversive memory consolidation/reconsolidation (see reviews [62,65]). These different functions probably depend, once again, on the precise neural network with which the ascending noradrenergic projections interact. The phasic/tonic distinction of locus coeruleus NA function posits that focused attention with engagement on a task associated with reward is associated with variations in phasic locus coeruleal activity, whereas tonic NA activity is associated with distractibility and exploration of other aspects of the environment [61]. Whether such a phasic/tonic distinction (or alternatively an exploit/explore dichotomy) can be related to consistent individual differences in trait-like behaviour is not yet known.

Not very much of this neurobiological evidence concerning functions of the central NA systems can readily be interpreted in terms of Cloninger's ‘Reward Dependence’ scheme or the proposal from Five Factor theory that NA mediates neuroticism. Van Gestel & Van Broeckhoven [66] summarized the sparse genetic evidence available in 2003 for associating polymorphisms of the noradrenaline transporter (NET) with personality dimensions, including Cloninger's postulated Reward Dependence. A study of 115 individuals [67] reported a relationship between the NET transporter polymorphism T-182C and reward dependence, as defined by the TCI. However, subsequently there has been a paucity of confirmatory evidence for this hypothesis using modern techniques. In particular, the role of NA in temperament and personality has been hindered by the relative lack of noradrenergic PET ligands [30]. Much of the animal research evidence relevant to this construct has focused instead on the role of DA in addiction.

For Gray's original ‘behavioural inhibition system’ [68], the locus coeruleal NA projection was held to be a distinct part of a regulatory influence for aspects of ‘arousal’ (hypothetically by affecting the hypothalamus) and selective attention to important events (by affecting the septo-hippocampal system) [68]. These roles were distinguished from that produced by manipulations of 5-HT, which was postulated to have a crucial role in ‘behavioural inhibition’ in the septo-hippocampal system [68]. Both neurotransmitter systems were thus assumed to contribute to anxiety, although in distinct ways. Subsequent manifestations of the theory [69] have had relatively little to say about the respective roles of these modulatory systems, while partially embracing the hypothesis of Deakin & Graeff [35] described above, and actually rejecting the idea that hypothalamic NA is responsible for mediating behavioural ‘arousal’. In fact, several neurochemical influences on septo-hippocampal function have subsequently been postulated, including cholinergic as well as monoaminergic influences, leading to a suggestion that there may be several distinct neurochemical forms of ‘arousal’, [70] a position previously arrived at, though based on somewhat different evidence, by Robbins & Everitt [63]. This position is compatible with hypotheses that neurotransmitters such as NA and acetylcholine modulate more basic processes of alerting, orienting and selective attention, which might be related to novel trait dimensions [62].

A similar conclusion about the current lack of evidence for linear dimensions of temperament or personality would hold for acetylcholine, which has been linked most clearly to processes of selective attention and memory by basic neuroscience studies [63,71]. However, there has been relatively little information on its role in traits or in studies of personality. A recent study showed that availability of the α4β2 nicotinic receptor was associated with elevated Harm Avoidance scores on the TCI test [72]. This finding incidentally underlines the point that single neurotransmitters are unlikely to be uniquely related to these temperamental dimensions. Another investigation has revealed how another polymorphism limiting cholinergic capacity (a variant of the acetylcholine precursor choline transporter gene SLC5A7) influences distractibility [73] (possibly complementing the locus coeruleus NA systems). These examples show how more emphasis has to be placed on the interactions between neurotransmitter systems (including neuropeptides such as encephalin and orexin and also modulatory neuroendocrine influences such as oxytocin and testosterone) in defined neural networks, in order best to define the biological basis of temperament.

Trofimova [6] and Trofimova and Robbins [65] have recognized the need for greater complexity of neurotransmitter interactions based on her Functional Ensemble of Temperament (FET) model, which derives from the Russian tradition of experimental and clinical observations in differential neurophysiology (Luria and Rusalov).The FET corresponds to the Structure of Temperament Questionnaire (STQ)-77 self-report questionnaire used to study temperament profiles in clinical [74] and healthy samples [75]. This model is unusual in that it considers a large number of temperamental traits that result not from one-to-one correspondence between temperament traits and neurotransmitters but rather from the interactions among several neurotransmitters. The FET derives traits from specified (sometimes hypothetical) interactions among the neurotransmitter systems, including quite innovatively, other systems beyond the monoamines such as acetylcholine, as well as the opioids and neuropeptides, that map onto interactions between cognitive, motor and social modalities and specific processes such as behavioural orientation, plasticity and ‘energetic aspects’ [6,65,74,75]. While the model undoubtedly needs further testing and further formulation that would ideally arise from utilizing objective behavioural measures to supplement and validate the FET, it does represent one of the first attempts to begin to grapple with the complexity of providing a neurochemical basis of temperamental factors.

Further analysis of how the ascending monoamines contribute to temperamental factors should recognize not only that the genetic basis of such complex traits is likely to be polygenic (which in itself means that it is likely that each system will contribute to each trait) but also that these systems likely mediate different and functionally highly specific components of a general arousal-type process. As we pointed out earlier [63,65], several arousal systems have been identified in the human brain regulating very specific functional aspects of behaviour. This generic process is linked to different functional contexts and aspects of performance mediated by distinct neural networks [63] shaped by experience and learning and operating on a baseline background of activity, which may be in part inherited. The goal for future studies should be to use all of the available information to produce a more detailed account of mechanisms underlying these factors, with implications for understanding their contribution to biologically based traits in healthy humans and to extremes in their expression in mental disorders.

Data accessibility

This article has no additional data.

Competing interests

I declare I have no competing interests.

Funding

The author's research is supported by the Wellcome Trust.