Fear versus anxiety

Adaptation to aversive environmental conditions is crucial for species survival, and at least two distinct behavioral systems have evolved to promote such adaptation. One system is fear (often labeled harm avoidance in the trait literature), which is a specific behavioral system that evolved as a means of escaping unconditioned aversive stimuli that are inherently dangerous to survival, such as tactile pain, injury contexts, snakes, spiders, heights, and sudden sounds. These stimuli are specific, discrete, and explicit, and in turn elicit specific, short-latency, high-magnitude phasic responses of autonomic arousal, subjective feelings of panic, and behavioral escape. Specific, discrete neutral stimuli associated with these unconditioned events elicit conditioned fear.

There are, however, many situations in which specific aversive cues do not exist, and instead, the stimulus conditions are associated with an elevated potential risk of danger or aversive consequences. In such cases, no explicit aversive stimuli are present to inherently activate fear circuitries. Nevertheless, the stimuli can be unconditioned in nature, as in darkness, open spaces, unfamiliarity, strangers, and predator odors, or they can be conditioned contextual cues (general textures, colors, relative spatial locations, sounds) that have been associated with previous exposure to specific aversive stimuli. Conceptually, these stimuli are commonly characterized by unpredictability, uncontrollability, unavoidability, unfamiliarity, or, more simply, uncertainty, which are the same stimulus features that characterize what is referred to as “stress.”

To reduce the risk of danger in such circumstances, a second behavioral system evolved: anxiety. Anxiety is characterized by negative emotions (anxiety, depression, hostility, suspiciousness, distress) that serve the purpose of informing the individual that, although no explicit, specific aversive stimuli are present, conditions are potentially threatening (White & Depue, Reference White and Depue1999). This affective state, and the physiological arousal that accompanies it, continues or reverberates until the uncertainty is resolved. It is this prolonged negative state of anxiety that distinguishes its subjective character from the rapid, brief state of panic and fear associated with the presence of a specific fear stimulus. Associated responses, that may functionally help to resolve the uncertainty, are heightened attentional scanning of the uncertain environment and cognitive worrying and rumination over possible response–outcome scenarios. Thus, anxiety represents a general behavioral system that is activated by a broad class of stimulus (uncertainty), whereas fear represents a more specific behavioral system that evolved to respond to specific stimuli critical for survial. The trait literature supports the independence of anxiety and fear, which as personality traits are completely uncorrelated and are subject to distinct sources of genetic variation (Tellegen et al., Reference Tellegen, Lykken, Bouchard, Wilcox, Segal and Rich1988; White & Depue, Reference White and Depue1999).

Neurobiology of anxiety and stress reactivity

The neurobehavioral system of anxiety is represented in the higher order traits of personality by neuroticism, which is characterized by negative emotions, a heightened and persistent state of stress reactivity, and interpersonal alienation. Animal and human research have identified three neurobiological pathways that are associated with anxiety, stress reactivity, and neuroticism, which is consistent with the systems theory concept of multifinality (Cicchetti & Dawson, Reference Cicchetti and Dawson2002): a genetic polymorphism in the gene that codes for the serotonin transporter (5-HTT), the central corticotropin-releasing hormone (CRH) system, and the peripheral glucocorticoid system. Furthermore, all three are subject to modification by adverse events in early development.

5-HTT and neural circuitries

Following release 5-HT is deactivated by being taken back up into the presynaptic terminal and is then restored in vesicles. This uptake process requires active transport across the presynaptic membrane by the 5-HTT residing in that membrane. As shown in Figure 3, there is a polymorphism in the promotor region of the gene (SLC6A4) that codes for 5-HTT that results in two common alleles, referred to as long (l) and short (s) alleles because their respective promotor regions differ in the number of nucleotides (Lesch & Canli, Reference Lesch, Canli and Canli2006). Because of the facilitatory effects of the promotor region on gene production, the l allele codes for more RNA and therefore produces more transporter protein than the s allele. The s allele is characterized by reduced 5-HTT promotor activity and 5-HTT expression, resulting in a reduced rate of removing 5-HT from the synaptic cleft and increased extracellular levels of 5-HT.

Figure 3.

Illustration of the components of the gene (SLC6A4) that codes for the serotonin transporter (5-HTT). Polymorphism of the promotor region yields two common alleles (S and L) that differ in length of nucleotide number. See text for details. [A color version of this figure can be viewed online at journals.cambridge.org/dpp]

At the behavioral level, 5-HTT knockout mice exhibit increased anxiety and reduced exploration, and primates and humans with the s allele (or the homolog of it in primates) express increased behavioral anxiety, increased scores on neuroticism scales, increased risk of depression, less effective responses to selective serotonin reuptake inhibitor (SSRI) treatment, and reduced social closeness (Hariri, Reference Hariri and Canli2006). At the neural level, the s allele has been associated with both neural function and structure. The s allele has been consistently found to be associated in both males and females with enhanced amygdala reactivity to many forms of emotional stimuli, including fear and angry faces, and negative scenes and words (Canli & Lesch, Reference Canli and Lesch2007; Hariri, Reference Hariri and Canli2006). In addition, the s allele is associated with reduced gray matter volume in the amygdala, suggesting reduced dendritic branching and axonal arborization (Hariri, Reference Hariri and Canli2006). Thus, there is a link between gene variation and basic brain mechanisms involved in processing negative emotion.

The reduced gray matter volume and increased reactivity of the amygdala in s-allele carriers (both S/S and L/S) raises the possibility that the amygdala's neural circuitry interactions may be influenced (i.e., its projections to and efferents from other brain regions). Indeed, recent research on the anterior cingulate cortex (ACC)–amygdala circuitry confirms this possibility. Specifically, the perigenual portion of the ACC, which surrounds the genua of the corpus collosum, is composed of the subgenual, rostral (rACC) and the supragenual, caudal (cACC). As shown in Figure 4, the rACC receives the most dense projections from the amygdala, conveys that information to the cACC, which in turn, sends efferents back to the amygdala. This circuitry is critical to modulating amygdala reactivity via inhibitory feedback from cACC to the amygdala during times of emotional stress.

Figure 4.

Anterior cingulate–amygdala circuitry that provides negative feedback to modulate activity of the amygdala. See text for details.

Pezawas et al. (Reference Pezawas, Meyer-Lindenberg, Drabant, Verchinski, Munoz and Kolachana2005) found psychiatrically normal s-allele carriers to have reduced gray matter volume of both the perigenual ACC and the amygdala, with the most pronounced reduction in the rACC region, and a reduced covariance of perigenual ACC (pACC, especially rACC)–amygdala volumes. Furthermore, during the processing of fear and angry faces, there was greatly reduced positive covariance between rACC–amygdala activation, and reduced negative covariance between cACC–amygdala activation, suggesting a reduced negative feedback of the cACC on amygdala reactivity to stressful stimuli. Covariance between rACC–amygdala activation accounted for ~30% of the vairance in neuroticism scores in the entire sample, whereas no single region was significantly related to neuroticism. Thus, it is the reduced coherence of activity across brain regions associated with the s allele that most markedly affects enhanced reactivity to stress.

The implications of amygdala overreactivity are consistent with the phenotypic features of neuroticism. One can expect the persistent increased arousal being conveyed by amygdala efferents to other brain regions to be associated with anxiety and feelings of distress, persistent enhancement of vigilance for threat, increased rumination of negative thoughts, increased acquisition of negative emotional memories, and decreased extinction of anxiety elicited by environmental contexts. The s allele creates a reduced threshold for stimulus-induced arousal in the amygdala that is dependent on less coherent activity within its interactions with perigenual ACC circuitry. This state of persistent overreactivity apparently creates an enhanced risk for anxiety and depressive disorders in s-allele carriers, but only under conditions of stressful life events or early adverse experience such as childhood maltreatment (Caspi, Sugden, Moffitt, Taylor, & Craig, Reference Caspi, Sugden, Moffitt, Taylor and Craig2003).

Serotonin and development of neural connections

The altered coherence in gray matter volume and functional activity in the pACC–amygdala circuitry, implying reduced dendritic branching and axonal arborization, suggests the possibility that the s allele modifies connections between brain regions. Significantly, 5-HT plays a critical role in the development and plasticity of the brain (Gaspar, Cases, & Maroteaux, Reference Gaspar, Cases and Maroteaux2003). 5-HT neurons are among the earliest neurons to be generated prenatally, appearing in the first embryonic month in primates. Knockout of the 5-HT system prenatally results in no gross wiring or structural effects, but 5-HT is used as a later developmental signal, which modulates construction and plasticity of brain circuits. Major pre- and postnatal developmental effects of 5-HT include modulation of dendritic length and number of spines, and axon arborization: that is, the development of selected brain circuits.

Pre- and postnatal developmental effects of 5-HT arise from two sources. First, there is direct release of 5-HT from brain stem raphe (where 5-HT synthesizing neurons are localized) projections to cortical target regions. Second, glutamate neurons, which comprise the main excitatory neurons that interconnect brain regions, in the process of developing thalamic–limbic and corticocortical efferents capture 5-HT via transient axonal expression of the 5-HT transporter. 5-HT is then released at thalamocortical and corticocortical synapses to enhance presynaptic axon arborization and postsynaptic dendritic processes, again enhancing neural circuitries. This represents, then, one of the means by which l- and s-allele variation could make a significant difference in interconnections between brain regions.

The effects of 5-HT during development are dependent on which 5-HT receptors are activated. There are four groupings of these receptors. 5-HT1A autoreceptors, localized to soma and dendrites of raphe nuclei, and 5-HT2A receptors, one of the most widely expressed in the brain, are expressed late in development and enhance dendritic processes. 5-HT2A receptors also modulate the expression of the brain-derived neurotrophic factor (BDNF) gene in the neocortex and hippocampus. BDNF secretion from postsynaptic neurons in response to neurotransmitter activation of postsynaptic receptors sculpts glutamate innervation patterns of dendrites and axons, increases synaptic efficacy of glutamate, and modulates long-term potentiation in the hippocampus. Thus, 5-HT2A receptor activation can in this indirect way modulate dendritic and axonal processes. Conversely, 5-HT1B and 5-HT2C receptors are mainly localized to axons and axon terminals, enhancing synaptic strength via axon arborization of glutamate axonal connections through enhanced release of glutamate.

This 5-HT modulation has its strongest effects during development as opposed to later in life (Ansorge, Zhou, Lira, Hen, & Ginrich, Reference Ansorge, Zhou, Lira, Hen and Gingrich2004; Gaspar et al., Reference Gaspar, Cases and Maroteaux2003; Gross & Hen, Reference Gross and Hen2004; Leonardo & Hen, Reference Leonardo and Hen2006). In rodents, the forebrain has an experience-expectant period of dramatic synaptogenesis and dendritic growth 3–4 weeks postnatally, and hippocampal maturation of dendritic branches occurs during 2–4 weeks postnatally. In primates, this period corresponds to early postnatal months. The 5-HT1A receptor also appears in Week 2 postnatally in the hippocampus, amygdala, and ACC. If knocked out or blocked in the first 4 weeks in the early postnatal period, there is a strong enhancement of adult anxiety that first appears in Week 3, but no anxiety enhancement occurs when the receptor is blocked in adulthood (Lesch & Canli, Reference Lesch, Canli and Canli2006; Tsetsenis, Ma, Iacono, Beck, & Gross, Reference Tsetsenis, Ma, Iacono, Beck and Gross2007). 5-HT1A autoreceptors enhance dendritic maturation, and thus their blockade could affect neural circuitry; whereas 5-HT1A postsynaptic receptors induce hyperpolarization and reduce neuronal excitability in postsynaptic neurons, thus increasing the threshold for firing the postsynaptic neuron. Blockade of this latter effect could contribute to amygdala overreactivity. These findings suggest that 5-HT mediation by 5-HT1A receptors is essential to the establishment of normal anxiety-modulating circuits in the brain during early postnatal development. Knockout of monoamine oxidase A (MAOA), which provides the main degradation of 5-HT after reuptake, inordinately increases 5-HT and has many developmental effects, but only when it is knocked out in the early postnatal period (Gaspar et al., Reference Gaspar, Cases and Maroteaux2003). Consistent with these biochemical findings, adverse rearing environments in monkeys (Champoux et al., Reference Champoux, Bennett, Shannon, Higley, Lesch and Suomi2002) and in humans (Taylor et al., Reference Taylor, Way, Walch, Hilmert, Lehman and Eisenberger2006) interacts with the 5-HTT polymorphism, where the effects of the s allele on anxiety and depression is enhanced by the environment.

In sum, the hypothesis is that the 5-HTT polymorphism affects 5-HT modulated development of dendritic and axonal development and fine-tuning of connectivity patterns in the forebrain. Variation in the connectivity of the pACC–amygdala circuits manifests neurally as amygdala hyperreactivity to stress, and behaviorally as anxiety, stress reactivity, negative emotionality, and an increased risk for anxiety and depressive disorders in the context of accumulating environmental adversity. Risk factors that interact with 5-HT effects may be a polymorphic variation in the BDNF gene (val BDNF) that results in reduced BDNF postsynaptic secretion and hence reduced synaptic connectivity; low levels of MAOA activity, increasing 5-HT presynaptic levels and release; and early environmental diversity.

Central CRH and anxiety

As illustrated in Figure 5, the effects of stress are mediated by a central system of CRH neurons. Stressful contexts can include both explicit, discrete stimuli that generate fear and nonexplicit stimuli associated with uncertainty that generate anxiety. It is now well established that association of discrete, explicit neutral stimuli (fear conditioned stimulus [CS]) with an aversive unconditioned stimulus (UCS) occurs via cortical uni- and polymodal sensory efferents that converge on the basolateral complex of the amygdala (Davis & Shi, Reference Davis and Shi1999; Kalin, Shelton, & Davidson, Reference Kalin, Shelton and Davidson2004; LeDoux, Reference LeDoux1998). Fear CSs elicit a host of behavioral, neuropeptide, and autonomic responses via output from the basolateral amygdala to the central amygdala (Ce), which in turn, sends functionally separable efferents to many hypothalamic and brain stem targets (LeDoux, Reference LeDoux1998).

Figure 5.

Components of the central and peripheral corticotrophin-releasing hormone (CRH) systems. Ce, central amygdala nucleus; BNST, bed nucleus of the stria terminalis; LH, lateral hypothalamus; PGi, paragiganticocellularis; PVN, paraventricular nucleus of the hypothalamus; ACTH, adrenocorticotropin hormone from the anterior pituitary.

In contrast, the enduring potentiating effects of nondiscrete, contextually related aversive stimuli on anxiety (but not fear) is dependent on a group of structures collectively referred to as the extended amygdala, which receives massive projections from basolateral and olfactory amygdala complexes, and which represents a macrostructure that is characterized by two divisions, central and medial (Heimer, Reference Heimer2003; Kalin, Shelton, Davidson, & Kelley, Reference Kalin, Shelton, Davidson and Kelley2001). These two divisions originate from the central and medial nuclei of the amygdala, and consist of cell groups that are distributed throughout the sublenticular area, bed nucleus of the stria terminalis (BNST), around striatal and pallidal structures, and back to merge specifically with the caudomedial region of the nucleus accumbens (NAS).

The central division of the extended amygdala appears to associate general contextual features and nonexplicit, nondiscrete conditioned and unconditioned stimuli with reward or punishment (e.g., light conditions, physical features, spatial relations; Davis & Shi, Reference Davis and Shi1999; McDonald, Shammah-Iagnado, Shi, & Davis, Reference McDonald, Shammah-Lagnado, Shi and Davis1999). Similar to the outputs from the Ce, most structures of the central division of the extended amygdala can transmit this motivationally relevant information to some or all hypothalamic and brain stem structures related to emotional expression (Heimer, Reference Heimer2003). This output coordinates activation of (a) the negative affective state of anxiety to inform the individual that the current context is uncertain and potentially dangerous, (b) autonomic arousal to mobilize energy for potential action, (c) selective attention to maximize sensory input at specified locations in the visual field, and (d) cognition to derive a response strategy. Thus, two emotional learning systems may have evolved: the basolateral amygdala to associate reinforcement with explicit, specific characteristics of objects (i.e., for fear), and the BNST to associate reinforcement with nonexplicit spatial and contextual stimulus aspects (i.e., for anxiety).

As indicated in Figure 5 above, the basolateral and central amygdala and the BNST provide integration of behavioral, neuroendocrine, and autonomic responses to stressful circumstances (Leri, Flores, Rodaros, & Stewart, Reference Leri, Flores, Rodaros and Stewart2002). This integration is accomplished by the activity of both the peripheral and central CRH systems. The peripheral system involves CRH neurons located in the paraventricular nucleus (PVN) of the hypothalamus, which when activated initiates the series of events that ends in the release of cortisol from the adrenal cortex. Cortisol then circulates in the bloodstream and increases energy reserves (gluconeogenesis) and, if not excessive in quantity, can enhance neurotransmitters involved in modulating emotion and memory (Kim & Diamond, Reference Kim and Diamond2002).

In contrast, the central CRH system is composed of CRH neurons located in many different subcortical brain regions that modulate emotion, memory, and CNS arousal. Whereas the majority of CRH neurons in the CNS do not mediate the effects of stress, some of the CRH-containing regions that are important in mediating stress effects are illustrated in Figure 5. For instance, the basolateral amygdala detects discrete aversive stimuli associated with stressful circumstances and activates the extensive array of CRH neurons located in the Ce (~1,750 neurons/hemisphere; Merali, Michaud, McIntosh, Kent, & Ainsman, Reference Merali, Michaud, McIntosh, Kent and Anisman2003). These CRH neurons project to many brain regions that modulate emotion, memory, and arousal, including the BNST and the peripheral CRH neurons in the PVN. Stress variables associated with context and uncertainty activate CRH neurons in the BNST, which have similar projection targets as the Ce (Macey, Smith, Nader, & Porrino, Reference Macey, Smith, Nader and Porrino2003). Both the Ce and the BNST can activate CRH neurons in the lateral hypothalamus (LH), a region that integrates CNS arousal as a means of modulating changes in the autonomic nervous system (ANS). In turn, the LH CRH projections modulate ANS activity. As illustrated in Figure 5, all three sources of CRH projections (Ce, BNST, and LH) innervate CRH neurons in the paragiganticocellularis (PGi; Aston-Jones, Rajkowski, Kubiak, Valentino, & Shiptley, Reference Aston-Jones, Rajkowski, Kubiak, Valentino, Shiptley, Holstege, Bandler and Saper1996), which is located in the rostral ventrolateral area of the medulla in the brain stem. The PGi is a massive nucleus that provides major integration of central and autonomic arousal and, in turn, coordinates and triggers arousal responses to urgent stimuli via two main pathways eminating from its own population of CRH neurons, which make up 10% of PGi neurons (Aston-Jones et al., Reference Aston-Jones, Rajkowski, Kubiak, Valentino, Shiptley, Holstege, Bandler and Saper1996). One CRH pathway modulates the ANS via projections to the intermediolateral cell column of the spinal cord, activating sympathetic preganglionic autonomic neurons.

The other CRH pathway modulates central arousal effects via activation of the locus coeruleus (LC), where PGi CRH innervation of the LC in humans and monkeys is dense (Aston-Jones et al., Reference Aston-Jones, Rajkowski, Kubiak, Valentino, Shiptley, Holstege, Bandler and Saper1996). The LC, which is composed of ~20,000 neurons that provide the major source of norepinephrine (NE) in the brain, innervates the entire brain. LC neurons that release NE onto B-adrenergic receptors are responsible for producing a nonspecific emotional activation via broadly collateralizing axons to the CNS (Aston-Jones et al., Reference Aston-Jones, Rajkowski, Kubiak, Valentino, Shiptley, Holstege, Bandler and Saper1996), and this NE release to chronic stress is increased in inbred rat strains with high anxiety (Blizard, Reference Blizard1988) and in posttraumatic stress disorder patients (Charney, Grillon, & Brenner, Reference Charney, Grillon and Bremner1998). This nonspecific emotional activation pattern comprises a global urgent response system that responds to unpredicted events, and hence, facilitates behavioral readiness, alerting (enhanced sensory processing), and attention (enhanced selection of stimuli; Aston-Jones et al., Reference Aston-Jones, Rajkowski, Kubiak, Valentino, Shiptley, Holstege, Bandler and Saper1996), which are all characteristic of states of anxiety in highly stress-reactive rodents and monkeys (Blizard, Reference Blizard1988; Redmond, Reference Redmond and Meltzer1987). This central arousal can be enduring: PGi CRH activation of LC neurons endures and peaks 40 min after stimulation of the PGi (Aston-Jones et al., Reference Aston-Jones, Rajkowski, Kubiak, Valentino, Shiptley, Holstege, Bandler and Saper1996). This persistent activation of spontaneous activity in the LC renders the effects of short-lived, nonsalient stimuli of little effect on the LC, such that salient stimuli have an inordinate effect on LC activity, and hence in turn, on central activation. Thus, taken together, the central CRH neuron system is capable of activating a vast array of behaviorally relevant neural and hormonal processes during stressful conditions, including activation of the peripheral CRH system.

Early research demonstrated that CRH in the BNST rather than in the Ce mediates anxiety as opposed to fear, whereas the Ce expresses fear but not anxiety. Thus, injection of CRH into the BNST of rats potentiates startle (an index of anxiety) in a dose-dependent manner for up to 2 hr duration (Davis & Shi, Reference Davis and Shi1999). Subsequent work has confirmed a specific role for CRH in the BNST in the mediation of anxiogenic effects. Intense inescapable stressors induce a persistent anxious state in rats characterized by increased sensory reactivity and enhanced associative learning with aversive UCSs. This is accompanied by elevated levels of CRH in the BNST. Marked anxiety effects lasting longer than 24 hr are produced after administration of three doses of CRH over 1.5 hr, despite no lasting effect on peripheral release of corticosterone (Servatius et al., Reference Servatius, Beck, Moldow, Salameh, Tumminello and Short2005). The source of naturally occurring CRH in the BNST is from the Ce's population of CRH neurons: CRH release from the Ce after a psychogenic stressor causes long-term anxiety effects in rats, which correlates with cerebrospinal fluid CRH concentrations (Merali et al., Reference Merali, Michaud, McIntosh, Kent and Anisman2003).

The anxiogenic effects of CRH are specific to the lateral region of the BNST. Intralateral BNST (but not intra-Ce) administration of CRH in rats elicits dose-dependent anxiogenic effects via CRH-R1 receptors, including reduced exploration of an unfamiliar cage and an elevated T-maze (which are both aversive UCSs); reduced sleep, food intake, and sexual activity; and increased sympathetic and behavioral signs of anxiety (shaking, crouching, cautious approach; Nie et al., Reference Nie, Schweitzer, Roberts, Madamba, Moore and Siggins2004). Furthermore, intralateral BNST (but not intra-Ce) administration of CRH in rats provides a dose-dependent aversion to the CRH-paired environment via CRH-R1 receptors, indicating that the lateral BNST is a critical site for associative conditioning of context and aversive signals mediated by CRH (Sahuque et al., Reference Sahuque, Kullberg, Mcgeehan, Kinder, Hicks and Blanton2006). The CRH-R1 receptors in the forebrain mediate these anxiogenic effects. Transgenic mice with elevated CRH-R1 (but not with elevated R2) receptors in the central forebrain (but with no elevation peripherally in the hypothalamus or pituitary) show extreme indications of anxiety; whereas mice with knockout of the CRH-R1 (but not R2) receptor gene in the central forebrain (but not peripherally) show little anxiety in the elevated T-maze or in the light–dark test (both being aversive UCSs; Deussing & Wurst, Reference Deussing and Wurst2005). Moreover, conditional activation of CRH-R1 gene expression centrally (but not peripherally) produces anxiogenic effects, whereas conditional inhibition of CRH-R1 gene expression centrally (but not peripherally) produces anxiolytic effects. Finally, a CRH-R1 antagonist administered orally to male monkeys impaired development and expression of contextual anxiety conditioning, and blocked social stress-induced (two unfamiliar male monkeys placed in attached plexiglass cages) increases in anxiety behaviors, cortisol, cerebrospinal fluid CRH concentrations, and LC neuronal firing (Habib et al., Reference Habib, Weld, Rice, Pushkas, Champoux and Listwak2000).

These findings, taken together, suggest that anxiety is essentially a stress response system that relies on a central network of CRH neuron populations acting via CRH-R1 receptors that provide integrated responses (hormonal, behavioral, autonomic, and central arousal) to the stressor. CRH especially in the lateral BNST mediates what is a defining characteristic of anxiety as opposed to fear: prolonged anxiogenic effects, which last as long as uncertainty is unresolved, and aversive contextual conditioning. These CRH-induced psychological effects appear to strongly match the content of trait neuroticism measures, where the strongest primary or facet scale in the higher order factor of neuroticism in Tellegen's personality questionnaire is termed stress reactivity because of the persistent negative emotional responses reported to stress-related events in items loading on the scale (Tellegen & Waller, Reference Tellegen, Waller, Briggs and Cheek1997).

Finally, although trait levels of neuroticism are very stable over as many as 20 years, it is less stable up to adolescence. This suggests that neural factors that contribute to neuroticism may be more influenced by experience earlier in life. In support of this notion, a recent study found that early postnatal separations from mother rats daily for 3 hr caused in pups later in life enhanced reactivity to brief stressors in anxiety behaviors and corticosterone secretion, but also in CRH mRNA expression in the Ce (Gillespie & Nemeroff, Reference Gillespie and Nemeroff2007). The latter implies that early adverse experience may result in a persistent Ce-induced enhanced reactivity via its CRH efferents throughout stress-related neural pathways shown in Figure 5, including in the lateral BNST.

Peripheral pathways

As shown on the far right side of Figure 5, the BNST via CRH efferents activates CRH neurons in the PVN under stressful conditions. CRH is then released from the hypothalamic CRH neurons and is transported via a portal system to the anterior pituitary, where adrenocorticotropin hormone (ACTH) is released into the blood stream, activates the adrenal cortex, which then secretes cortisol (mainly corticosterone in rodents). Cortisol travels through the blood stream and has a diverse set of effects on energy metabolism and nerual functioning. In humans and rodents, there are two types of stress-induced cortisol receptors. Type I or mineralocorticoid receptors (MRs) have a high affinity for cortisol and thereby mediate the acute stress-induced effects of cortisol on bodily responses. Type II or glucocorticoid receptors (GRs) have a low affinity for cortisol, and thereby become bound to cortisol only at high levels of cortisol concentration. Such high levels of cortisol in the hippocampus, acting via GRs, activate inhibitory output from the hippocampus to the hypothalamic CRH secretory neurons, thereby reducing the activity of the CRH–cortisol stress response system. This pathway represents the major negative feedback system for controlling stress-induced CRH–cortisol responses.

Peripheral glucocorticoid system and early adversity

Noting that the relations between early adversity and environmental poverty with adulthood health (including stress disorders later in life) appear to be mediated by parental influences, researchers have begun to explore whether these parental influences operate via effects on the development of neural systems that underlie the expression of behavioral and endocrine responses to stress (Gross & Hen, Reference Gross and Hen2004; Meaney, Reference Meaney2001; Sapolsky, Reference Sapolsky2004). Postnatal handling studies in rats provided early support for this notion. Postnatal handling of 3–15 min/day resulted in reduced stress reactivity and fearfulness, and increased exploration; decreased hippocampal aging, and more intact cognitive functioning in old age; increased GRs in the hippocampus; and increased gamma aminobutyric acid (GABA) A receptors (the activation of which reduces anxiety behavior in rats) in the amygdala, LC, and hippocampus. In contrast, maternal separation of rat pups had the opposite effects of postnatal handling. For instance, pups separated from their mothers for 180 min/day for 2–14 days (which simulates a scarce environment in which the mother leaves the nest for days at a time to forage for food) show significant increases in CRH mRNA expression in the Ce and PVN; decreased GR mRNA and reduced 5-HT1A responses in the hippocampus; and reduced GR binding in the hippocampus, hypothalamus, and frontal cortex, all of which blunts the CRH negative feedback system. Of importance, these pups also manifested a lifetime hyperresponsivity to unfamiliar stressors.

A series of studies of individual differences in rat maternal behavior toward pups has also supported these findings (Champagne et al., Reference Champagne, Bagot, Hasselt, Ramakers, Meaney and de Kloet2008; Leonardo & Hen, Reference Leonardo and Hen2006; Meaney, Reference Meaney2001; Weaver et al., Reference Weaver, Cervoni, Champagne, D'Alessio, Sharma and Seckl2004). Rat mothers naturally differ in their levels of licking/grooming (LG) and arched-back nursing (ABN; which is less stressful for pups than being compressed by the mother's body during nursing), creating two groups of mothers: high versus low LG-ABN. These different maternal patterns are stable over litters, so they appear to be traitlike under normal conditions. Behaviorally, pups of high LG-ABN mothers show reduced startle responses, increased open-field exploration, and reduced latencies to eat food in a novel environment. Neurobiologically, these pups show reduced plasma ACTH and corticosterone responses to acute stress, increased hippocampal GR mRNA expression, and hence, increased glucocorticoid negative feedback sensitivity, decreased CRH mRNA levels in the hypothalamus, increased affinity for GABA at its A receptor in the amygdala and LC, and an increased 5-HT tone in the hippocampus. Moreover, as pups and adults, offspring of high LG-ABN mothers also show less cell death and increased glutamate receptor levels (resulting in enhanced synaptogenesis) in the hippocampus and enhanced memory on hippocampal-dependent tasks.

Thus, the behavior of the mother toward offspring can apparently “program” behavioral and endocrine responses to stress that endure the entire lifespan. Moreover, as illustrated in Figure 6, individual differences in stress reactivity mediated by maternal care are transmitted across generations (Gross & Hen, Reference Gross and Hen2004). Indeed, although adult mothers must overcome initial neophobia to pups, female offspring of high LG mothers are initially less averse to pups. When offspring of low LG mothers are raised from birth by high LG mothers (i.e., cross-fostering), these offspring manifest low stress reactivity and anxiety as pups and adults. Of importance, however, Meaney (Reference Meaney2001) reported that raising offspring of high LG mothers with low LG mothers produced stress reactive offspring, but this has not been replicated: others have found that these offspring show adult stress phenotypes that are more similar to their high LG biological mother (Figure 6; Gross & Hen, Reference Gross and Hen2004). We return to this issue later. What is clear is that variation in maternal care can affect development of neural systems that mediate stress reactivity, which subsequently influences offsprings' maternal behavior.

Figure 6.

Rats raised by mothers that display low licking/grooming behavior exhibit more anxiety-related behavior than rats raised by high licking/grooming mothers. Cross-fostering studies show that the offspring of low licking/grooming mothers raised by high licking/grooming mothers are less prone to anxiety-related behavior as adults. This indicates that the effect is mediated by postnatal maternal environment. However, offspring of high licking/grooming mothers raised by low licking/grooming mothers do not have an increased tendency to develop anxiety-related behavior in adulthood, indicating that specific factors inherited by the high licking/grooming offspring protect them from the effects of being mothered by low licking/grooming females. From “Developmental Origins of Anxiety,” by E. Gross and R. Hen, Reference Gross and Hen2004, Nature Reviews Neuroscience, 5. Copyright 2004 by Nature Publishing Group. Reprinted with permission. [A color version of this figure can be viewed online at journals.cambridge.org/dpp]

It is important to note that there is a convergence between the roles of 5-HT and CRH in these studies that may be significant with respect to the above discussion of the 5-HTT and CRH in anxiety and stress reactivity. Recall that high LG mothers and their offspring showed increased 5-HT tone in the hippocampus, and that maternal separation reduced 5-HT1A responses in hippocampus. Stress-induced CRH release onto CRH-R1 receptors in the dorsal raphe nucleus (a major brain stem site of 5-HT neurons) elevates 5-HT concentrations in the hippocampus by 350%. 5-HT activity in the hippocampus activates GR gene expression, providing increased negative feedback to the PVN and Ce CRH neurons. In turn, GR activation enhances postsynaptic effects of 5-HT1A activation, which increases inhibition of excitation in the postsynaptic neuron. Furthermore, 5-HT activation of postsynaptic glutamate neurons in the hippocampus activates BDNF secretion into the synapse, increasing synaptic processes and decreased aging effects in the hippocampus. Therefore, variation in the functional properties of the 5-HT and/or the CRH-R1 systems could influence the efficacy of the GR-induced negative feedback system for controlling stress-induced cortisol secretion.

Early experience and epigenetic processes

The critical question concerning the early handling, separation, and maternal care findings is how these influences on stress reactivity and anxiety become long lasting. Research thus far has supported the role of two mechanisms: epigenetic effects and DNA methylation, both of which involve an alteration of gene expression by experience. In terms of epigenetic effects, as shown in Figure 7, DNA is wrapped around proteins called histones, with the DNA and histones collectively comprising chromatin (Tsankova, Renthal, Kumar, & Nestler, Reference Tsankova, Renthal, Kumar and Nestler2007). The functional unit of chromatin is the nucleosome, which is composed of 147 base pairs of DNA wrapped around a core histone group, where there are two copies of each of the four histones (Figure 7a). The compactness (openness vs. closedness) of histones in a particular region of chromatin determines how readily transcription factors can access promoters in the DNA (Figure 7b). A wide variety of regulators can alter chromatin structure (i.e., its openness), and hence the access to genes. Neurotransmitter (e.g., 5-HT) or hormone (e.g., cortisol) action at receptors can elicit activity in postsynaptic regulators of chromatin structure, and hence access to genes. For instance, activation of MRs and GRs alters the expression of 70–100 genes, 50% of which are upregulated. As we shall see, experience, via neurotransmitter and hormone effects, can permanently alter chromatin regulators and openness to gene transcription, and hence behavior.

Figure 7.

The general scheme of chromatin remodeling. (a) A picture of a nucleosome showing a DNA strand wrapped around a histone octamer composed of two copies each of the histones. (b) Chromatin can be conceptualized as existing in two primary structural states: as active or open (top left) in which histone acetylation (A) is associated with opening the nucleosome to allow binding of the basal transcriptional complex and other activators of transcription; or as condensed or closed where all gene activity is permanently silenced (bottom left). In reality, chromatin exists in a continuum of several functional states (active; permissive (top right); repressed (bottom right; and inactive). Adapted from “Epigenetic Regulation in Psychiatric Disorders,” by S. Tsankova, D. Renthal, A. Kumar, and E. Nestler, Reference Tsankova, Renthal, Kumar and Nestler2007, Nature Reviews Neuroscience, 8. Copyright 2007 by Nature Publishing Group. Adapted with permission. [A color version of this figure can be viewed online at journals.cambridge.org/dpp]

Alternatively, in terms of DNA methylation, genes can be stably silenced for substantial periods of time (a lifetime in the case of the second X chromosome in females) when a methyl group is attached to nucleotides in a gene's promotor or exon regions, where the methyl group then tightly binds a repressor protein (Figure 7b, inactive and repressed sections). Demethylation of a gene leads to enhanced expression of its protein product.

Changes in both chromatin remodeling and DNA methylation state are implicated strongly in the effects of handling and maternal style. Acetylation of histones decreases their binding to DNA (increases chromatin openness), thereby increasing access of transcription factors to promoters and exons. Histones surrounding the GR gene were more acetylated in high LG mothers and their offspring. As illustrated in Figure 8, a hypothesized chain of events (Champagne et al., Reference Champagne, Bagot, Hasselt, Ramakers, Meaney and de Kloet2008; Sapolsky, Reference Sapolsky2004; Tsankova et al., Reference Tsankova, Renthal, Kumar and Nestler2007) is that high LG rearing results in increased 5-HT neurotransmission in the hippocampus of pups (Point 1, Figure 8). The resulting increased 5-HT activation of postsynaptic neurons in the hippocampus leads to strong activation of cAMP, protein kinase A, and its cascade of events within the postsynaptic neuron. In turn, an increased acetylation of histones and opening of chromatin occurs (Point 2, Figure 8). Also, increased expression of the transcription factor nerve growth factor-inducible protein A (NGFI-A) occurs, which results in increased NFGI-A binding to its binding site located on the first exon of the GR gene (Point 3, Figure 8). Such binding enhances the expression of GRs in the hippocampus, providing enhanced negative feedback effects on central and peripheral CRH release. 5-HT via its postsynaptic effects activates the BDNF gene, which increases synaptogenesis in the hippocampus when released into the synaptic cleft (Point 5, Figure 8).

Figure 8.

Five effects of high licking/grooming maternal care in the hippocampus on gene expression. See text for details.

In terms of DNA methylation, it has been found that the methylation state of the NGFI-A binding site on the first exon of the GR gene is plastic around the time of birth. The site is then methylated the first day after birth, decreasing transcription of the hippocampal GR gene. Over the next week, exposure to high LG mothering (by biological or cross-fostering females) caused the binding site to be demethylated, increasing transcription of the GR gene, and hence, hippocampal GRs (Point 4, Figure 8). Recent evidence supports a causative role for the above histone actylation and methylation effects. A pharmacological inhibitor of histone deacetylation opens up chromatin and facilitates access of demethylating enzymes, shifting the balance in chromatin toward acetylation. Administration of this inhibitor to adult rats raised in a low LG environment transformed them into a high LG adult in terms of stress reactivity. This behavioral effect was accompanied by an increase in NGFI-A access to the now demethylated promotor and exon 1 of the GR gene and by reduced levels of circulating glucocorticoids.

It is significant that most of the effects of handling and maternal style occur only within a critical developmental period (Champagne et al., Reference Champagne, Bagot, Hasselt, Ramakers, Meaney and de Kloet2008; Gross & Hen, Reference Gross and Hen2004; Leonardo & Hen, Reference Leonardo and Hen2006; Meaney, Reference Meaney2001). Handling, maternal style, and methylation status all must be present in the first week of life to have an effect in rodents. This is consistent with the findings discussed above, where the first 3 weeks in rodents are critical in the developmental effects of 5-HT, the effects of knocking out 5-HT1A receptors and MAOA, and in CRH reactivity. Similarly, SSRIs (which result in enhanced 5-HT availability) administered during Weeks 1–3 (but not during Weeks 6–8) normalize the glucocorticoid response in parentally stressed mice, indicating that this critical developmental period may have a relatively narrow window. Thus, as summarized in Figure 9, experience during early postnatal weeks in rodents, and perhaps early months in primates, appear to interact with genetic polymorphisms in 5-HT, BDNF, and MAOA functioning, creating a critical period for establishing effective synaptic development in the forebrain structures that mediate anxiety in response to experience-dependent signals.

Figure 9.

A summary of the effects of maternal experience and genetic polymorphisms on gene expression of glucocorticoid receptors during an early postnatal period. See text for details.

Dopamine (DA) contributions to extraversion

Through empirical studies and integrative reviews, we have demonstrated that extraversion reflects the activity of a mammalian behavioral approach system based on positive incentive motivation (Depue, Reference Depue and Canli2006; Depue & Collins, Reference Depue and Collins1999; Depue & Morrone-Strupinsky, Reference Depue and Morrone-Strupinsky2005). This system is activated by, and serves to bring an animal in contact with, unconditioned and conditioned rewarding incentive stimuli. Incentives are inherently evaluated as positive in valence, and activate incentive motivation, increased energy through sympathetic nervous system activity, and forward locomotion as a means of bringing individuals into close proximity to rewards. The incentive state is clearly evident in the phenotypic features of extraversion, and is associated with subjective feelings of desire, wanting, excitement, elation, enthusiasm, potency, and self-efficacy that are distinct from, but typically cooccur with, feelings of pleasure and liking.

Animal research demonstrates that the positive incentive motivation and experience of reward that underlies a behavioral system of approach is dependent on the functional properties of the midbrain ventral tegmental area (VTA) DA projection system. DA agonists or antagonists in the VTA or NAS, which is a major terminal site of VTA DA projections, in rats and monkeys facilitate or markedly impair, respectively, a broad array of incentive motivated behaviors. Furthermore, dose-dependent DA receptor activation in the VTA–NAS pathway facilitates the acute rewarding effects of stimulants, and the NAS is a particularly strong site for intracranial self-administration of DA agonists. DA agonists injected in the NAS also modulate behavioral responses to conditioned incentive stimuli in a dose-dependent fashion. In single-unit recording studies, VTA DA neurons are activated preferentially by appetitive incentive stimuli, and DA cells, most numerously in the VTA, respond vigorously to and in proportion to the magnitude of both conditioned and unconditioned incentive stimuli and in anticipation of reward (Schultz, Reference Schultz2007). Similarly, in human neuroimaging studies, increased activation in the VTA and NAS correlates with the amount of reward anticipated, and with ratings of excitement preceding the reward (Canli, Reference Canli and Canli2006; Knutson & Bhanji, Reference Knutson, Bhanji and Canli2006). Morevover, during acute administration the intensity of a subject's subjective euphoria increased in a dose-dependent manner in proportion to cocaine binding to the DA uptake transporter (and hence to DA levels) in the striatum (Drevets et al., Reference Drevets, Gautier, Price, Kupfer, Kinahan and Grace2001; Volkow et al., Reference Volkow, Wang, Fischman, Foltin, Fowler and Abumrad1997). Hence, taken together, the animal and human evidence demonstrates that the VTA DA–NAS pathway is a primary neural circuit for incentive motivation and its accompanying subjective state of reward.

Consistent with these findings are results with a polymorphism of the DA D4 receptor gene on chromosome 11 (Munafo et al., Reference Munafo, Clark, Moore, Payne, Walton and Flint2003). The D4 receptor is part of the D2 family of DA receptors, which are generally inhibitory in nature. The D4 polymorphism is characterized by variation in the number of repeats of a 48 base pair unit in exon 3. The most common repeats in the population are 4 and 7 repeats, where the 7-repeat allele is characterized by a twofold reduction in inhibitory effects on postsynaptic second-messenger cAMP relative to the 4-repeat allele. Several small studies have found the 7-repeat allele to be associated with increased extraversion levels relative to the 4-repeat allele, and this was also found using behavioral features of positive emotion in a study of 8- to 12-month-old infants (Lakatos et al., Reference Lakatos, Nemoda, Birkas, Ronai, Kovacs and Ney2003). When this polymorphism is taken in interaction with a polymorphism of the DA transporter (one allele being associated with reduced DA uptake and hence increased time of DA in the synapse to activate postsynaptic receptors), impulsivity was significantly increased in terms of number of comission errors in a go–no-go task.

In the discussion above of a threshold model of behavioral responding, it was argued that higher levels of neurobiological traits underlying personality would enhance the association of neutral cues with emotional state. We found that high extraverts more readily associate specific and general contextual cues to DA-agonist induced subjective reward than low extraverts, who showed no associative conditioning at the dose level of methylphenidate used in the studies (Depue, Reference Depue and Canli2006). These findings are consistent with an enhanced reward encoding of positive cues, and are consistent with the fact that extraversion is correlated with increased amygdala activation to happy versus neutral and negative faces. Moreover, the latter finding implies early detection of the enhanced weighted positive stimuli in the amygdala, which through backprojections to many areas of the brain, including sensory processing regions, can bias perceptual processing and attention to positive cues in the environment. Several studies support just such a bias in extraversion (Canli, Reference Canli and Canli2006).

Early adversity and DA functioning

Early adverse experience in animals has a marked effect on DA functioning, as summarized in Figure 10. Single intense stressors or repeated, prolonged stressors, including maternal separations in neonatal rats and monkeys, lead to a sensitization of NAS DA release to repetition of the same stressor or to novel stressors and DA agonists later in life (Point 6, Figure 10; Depue & Collins, Reference Depue and Collins1999; Meaney, Brake, & Cratton, Reference Meaney, Brake and Cratton2002). There is also an increase in the acquisition of self-administration, and a reduced threshold in terms of effective dosage of psychostimulants, alcohol, and opiates, and an increased intensity of effort (craving?) to obtain these drugs (Meaney et al., Reference Meaney, Brake and Cratton2002). Furthermore, rat pups exposed to maternal separations have a 250% reduction in DA transporters in the NAS as adults, which increases time of DA availability at postsynaptic receptors, whereas rats with a history of handling show reduced DA release in the NAS to stress and DA-agonist drugs (Point 5, Figure 10; Meaney et al., Reference Meaney, Brake and Cratton2002).

Figure 10.

Six effects of early aversive experience on glucocorticoid and dopamine (DA) systems in the ventral tegmental area (VTA) and nucleus accumbens (NAS). See text for details.

These effects appear to be enhanced by an interaction of glucocorticoid activation of GRs with DA. Amphetamine and stress both increase corticosterone secretion and NAS DA release. Administration of dexamethasone (synthetic cortisol) prenatally and during the first 7 postnatal days in rats results in a 50% increase in VTA DA neurons (Point 4, Figure 10), and increased corticosterone and NAS DA release to stress and DA-agonist drugs in adulthood (McArthur, Dalley, Buckingham, & Gillies, Reference McArthur, Dalley, Buckingham and Gillies2005). Even adult humans who report poor maternal care as children have stress-induced increased cortisol levels that correlate (r = .78) with increased NAS DA release (Pruessner, Champgane, Meaney, & Dagher, Reference Pruessner, Champagne, Meaney and Dagher2004). This effect is enhanced by stressor-induced prolonged oversecretion of cortisol (Point 1, Figure 10) that activates GRs located on VTA DA neurons (Oswald et al., Reference Oswald, Wong, McCaul, Zhou, Kuwabara and Choi2005). Such activation has the effect of increasing DA cell responses to input from corticolimbic regions (Point 2, Figure 10), possibly from an enhancement of glutamate release mediated by GRs on glutamate terminals that synapse on VTA DA neurons. In addition, cortisol activation of GRs on VTA DA neurons also enhances the frequency of DA cell firing (Point 3, Figure 10).

Thus, prolonged stressors, such as poor, conflictual, violent, or inconsistent familial environments or physical and/or sexual abuse early in life, can sensitize the VTA–NAS DA system in an enduring manner. This may have significance for personality disturbance. High levels of extraversion are often associated with histrionic–impulsive behavioral features that can lead to both occupational and interpersonal impairment. In addition, if extreme levels of extraversion reflect a sensitized DA system because of early adverse conditions, the risk for increased acquisition of addiction to even low doses of DA-agonist drugs (psychostimulants, alcohol, and opiates) later in life may be significantly increased. The increased associative conditioning of context to reward in high extraverts that we found may contribute to rapid conditioning of drug-associated contexts that enhance drug craving and maintenance of addiction. All of these effects would aggravate personality disturbance that is created in combination with other subsets of critical traits, such as neuroticism.

The nature of constraint

Constraint represents a broad trait of impulsivity that modulates the expression of many domains of behavior, including emotional, motor, cognitive, and sensory reactivity (Carver & Miller, Reference Carver and Miller2006; Depue & Collins, Reference Depue and Collins1999; Depue & Lenzenweger, Reference Depue, Lenzenweger, Lenzenweger and Clarkin2005; Depue & Spoont, Reference Depue and Spoont1986; Spoont, Reference Spoont1992; Zald & Depue, Reference Zald and Depue2001). Whereas the four traits described above all reflect the activity of neurobehavioral systems, constraint lacks ties to any specific motivational-emotional system (Carver & Miller, Reference Carver and Miller2006; Depue & Collins, Reference Depue and Collins1999; Spoont, Reference Spoont1992). Thus, I have proposed that constraint is the personality trait that reflects the greatest CNS weight on the construct of a minimum response threshold (Depue & Collins, Reference Depue and Collins1999; Depue & Lenzenweger, Reference Depue, Lenzenweger, Lenzenweger and Clarkin2005; Depue & Spoont, Reference Depue and Spoont1986). This can be visualized in Figure 2 above, where the trait of constraint would be the major modulator of the threshold for behavioral manifestation. In this model, emotional higher order personality traits reflect the influence of neurobiological variables that strongly contribute to the threshold for responding, such as DA in the facilitation of incentive motivated behavior, μ-opiates in the experience of affiliative reward, and CRH in the potentiation of anxiety. Constraint modulates the probability of elicitation of all of those variables.

Significantly, as shown in Figure 11, constraint in its modulatory interaction with an emotional personality trait creates a diagonal of stability of emotional behavior. This diagonal extends from lability where the emotional trait's level is high and constraint is low (i.e., increasingly easy stimulus elicitation of behavior because of a low threshold created by the interactions in Figure 2) to rigidity when the emotional trait's level is low and constraint is high (i.e., increasingly difficult stimulus elicitation of behavior because of a high threshold created by the interaction).

Figure 11.

The interaction of an emotional trait with the higher order trait of constraint, forming a diagonal of behavioral stability extending from labile to rigid. See text for details.

Constraint and serotonin functioning

Functional levels of neurotransmitters that provide a strong, relatively generalized tonic inhibitory influence on behavioral responding would be good candidates as significant modulators of a response elicitation threshold, and hence, would likely account for a large proportion of the variance in the trait of constraint. Numerous researchers previously (Carver & Miller, Reference Carver and Miller2006; Coccaro & Siever, Reference Coccaro and Siever1991; Depue, Reference Depue1995; Depue & Spoont, Reference Depue and Spoont1986; Lesch & Canli, Reference Lesch, Canli and Canli2006; Panksepp, Reference Panksepp1998; Spoont, Reference Spoont1992; Zald & Depue, Reference Zald and Depue2001; Zuckerman, Reference Zuckerman, Halverson, Kohnstamm and Marten1994) suggested that 5-HT, acting at multiple receptor sites in most brain regions, is such a modulator. 5-HT modulates a diverse set of functions, including emotion, motivation, motor, affiliation, cognition, food intake, sleep, sexual activity, and sensory reactivity (Carver & Miller, Reference Carver and Miller2006; Depue & Spoont, Reference Depue and Spoont1986; Lesch & Canli, Reference Lesch, Canli and Canli2006; Spoont, Reference Spoont1992). Furthermore, reduced 5-HT functioning in animals (Depue & Spoont, Reference Depue and Spoont1986; Spoont, Reference Spoont1992) and humans (Coccaro & Siever, Reference Coccaro and Siever1991) is also accompanied by irritability and hypersensitivity to stimulation in most sensory modalities (Spoont, Reference Spoont1992). Thus, 5-HT plays a substantial modulatory role in general neurobiological functioning that affects many forms of motivated behavior. In this sense, constraint might be viewed as reflecting the influence of the CNS variable of 5-HT, which we refer to as neural constraint in Figure 1. In support of this notion, we found that 5-HT agonist-induced increases in serum prolactin secretion were correlated significantly and specifically only with the Constraint/Impulsivity (control) Scale (−0.44, p < .01) from Tellegen's personality questionnaire of 13 primary scales (Depue, Reference Depue1995).

Constraint, instability, and personality disturbance

Along with neuroticism and low social closeness, constraint is well correlated with personality disturbance, particularly of the type related to unstable behavior (as in the classical unstable group of histrionic, antisocial, and boderline PDs). A vast body of animal and human evidence consistently associates reduced functioning of 5-HT neurotransmission with behavioral instability. In addition, recent research on two polymorphisms of the gene for tryptophan hydroxylase (TPH2, which is localized in brain rather than the periphery), the rate limiting enzyme in the synthesis of 5-HT, indicates marked behavioral instability (particularly in affective aggression and impulsive suicide) and increased amygdala reactivity to positive and negative emotional slides in one of the alleles that reduces brain 5-HT by 50% (Hennig, Reuter, Netter, & Buck, Reference Hennig, Reuter, Netter and Burk2005; Manuck et al., Reference Manuck, Flory, Ferrell, Dent, Mann and Muldoon1999; Rujescu et al., Reference Rujescu, Giegling, Bondy, Gietl, Zill and Moller2002). Furthermore, reduced levels of 5-HT are associated with many clinical conditions, including impulsive, violent suicide across several types of disorder, obsessive–compulsive disorder, disorders of impulse control, aggression and irritability, depression, anxiety and enhanced stress reactivity, arson, unconstrained sexual behavior, and substance abuse (Carver & Miller, Reference Carver and Miller2006; Coccaro & Siever, Reference Coccaro and Siever1991; Depue & Spoont, Reference Depue and Spoont1986; Winstanley, Theobald, Dalley, & Robbins, Reference Winstanley, Theobald, Dalley and Robbins2005).

The nature of this instability depends on the neurobehavioral system being elicited. An example of the interaction of neurobiological variables associated with constraint (5-HT) and extraversion (DA) helps to illustrate this point (for a discussion of the behavioral consequences of an interaction of each of the higher order traits with constraint, see Depue & Lenzenweger, Reference Depue, Lenzenweger, Lenzenweger and Clarkin2005). 5-HT is an inhibitory modulator of a host of DA-facilitated behaviors, including the reinforcing properties of psychostimulants, novelty-induced locomotor activity, the acquisition of self-administration of cocaine, and DA utilization in the NAS (Depue & Collins, Reference Depue and Collins1999; Depue & Spoont, Reference Depue and Spoont1986; Spoont, Reference Spoont1992). This modulatory influence arises in large part from the dense dorsal raphe 5-HT efferents to the VTA and NAS, connections that are known to modulate DA activity (Spoont, Reference Spoont1992). A 5-HT-related reduction in the threshold of DA facilitation of behavior results in an exaggerated response to incentive stimuli that is most apparent in reward–punishment conflict situations. In such situations, exaggerated responding to incentives results in (a) a greater weighting of immediate versus delayed future rewards, (b) increased reactivity to the reward of safety or relief associated with active avoidance (e.g., suicidal behavior), (c) impulsive behavior, that is, a propensity to respond to reward when withholding or delaying a response may produce a more favorable long-term outcome, and (d) various attempts to experience the increased magnitude and frequency of incentive reward, for example, self-administration of DA-active substances (Coccaro & Siever, Reference Coccaro and Siever1991; Depue & Iacono, Reference Depue and Iacono1989; Lesch & Canli, Reference Lesch, Canli and Canli2006; Winstanley et al., Reference Winstanley, Theobald, Dalley and Robbins2005). All of these effects impair the ability to sustain long-term goal-directed behavior programs (such as obtaining a college degree) by mental representations of expected rewards that can be repeatedly accessed, held on-line in the prefrontal cortex, and thereby symbolically motivate behavior (Depue & Collins, Reference Depue and Collins1999).

This discussion suggests an important role for 5-HT in the model of personality disturbance shown in Figure 1. Reduced levels of 5-HT create a highly reactive individual in terms of hyperactivity of the ANS to stress, hyperirritability to sensory overload and life frustrations, enhanced affective aggression to even minor provocations, unplanned suicidal attempts when stressors become unbearable, instability of attachment to others (vacillating between approach and avoidance), and labile positive and anxious moods. This behavioral instability may reflect the effect of 5-HT's reduced neural constraint over many CNS variables (Spoont, Reference Spoont1992), which might explain how alterations of 5-HT functioning is found in so many different clinical disorders. Such weak constraint over CNS variables might render the effects of stress as markedly increased on biology and hence behavior. In other words, 5-HT constraint may serve as a general modulator of expression on the reaction surface, increasing expression of personality disturbance during times of life stress and interpersonal chaos. At the same time, stress may increase the probability that 5-HT constraint will disinhibit the elicitation of the neurobehavioral systems of other emotional traits: animal research has shown that stressors of many types, but especially perceived or actual social aggression and social isolation (both of which are common in the life of individuals with personality disturbance), further decrease 5-HT functioning. Thus, 5-HT constraint may serve as a strong modulator of the extremity of an individual's position on a trait and, in turn, the course of intensity of personality disturbance over the life span.