Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Arch Gen Psychiatry. Author manuscript; available in PMC 2011 May 2.
Published in final edited form as:
PMCID: PMC3085183

PET Measures of Endogenous Opioid Neurotransmission Predict Impulsiveness Traits in Humans



The endogenous opioid system and μ-opioid receptors are known to interface environmental events, both positive (e.g., relevant emotional stimuli) and negative (e.g., stressors) with pertinent behavioral responses, regulating motivated behavior. Here we examined the degree to which trait impulsiveness, the tendency to act on cravings and urges rather than delaying gratification, is predicted by either baseline μ-opioid receptor availability or the response of this system to a standardized, experientially-matched stressor.


Nineteen (19) young healthy male volunteers completed a personality questionnaire (NEO PI-R) and positron emission tomography scans with the μ-opioid receptor selective radiotracer [11C]carfentanil. Measures of receptor concentrations were obtained at rest and during the receipt of an experimentally maintained pain stressor of matched intensity between subjects. Baseline receptor levels and stress-induced activation of μ-opioid neurotransmission were compared between subjects scoring above and below the population median of the NEO impulsiveness subscale and the orthogonal dimension, deliberation, expected to interact with it.


High impulsiveness and low deliberation scores were associated with significantly higher regional μ-opioid receptor concentrations and greater stress-induced endogenous opioid system activation. Effects were obtained in regions involved in motivated behavior and the effects of drugs of abuse: prefrontal and orbitofrontal cortex, anterior cingulate, thalamus, nucleus accumbens and basolateral amygdala. Mu-opioid receptor availability, and the magnitude of stress-induced endogenous opioid activation in these regions accounted for 21 to 49% of the variance in these personality traits.


Our data demonstrate that individual differences in the function of the endogenous μ-opioid system predicts personality traits that confer vulnerability or resiliency for risky behaviors, such as the predisposition to develop substance use disorders. These personality traits are also implicated in psychopathological states (e.g., personality disorders), where variations in the function of this neurotransmitter system may play a role as well.


There is substantial interest in the mechanistic understanding of traits that may predispose individuals to the development of specific behaviors or psychopathologies.

Trait impulsivity has received substantial attention because of its association with risky behaviors (e.g., experimentation with drugs, sex, problem gambling reckless driving), personality disorders, or even the mortality associated with the mood disorders. Impulsivity, though frequently referred to as a single trait, is better conceptualized as heterogeneous characteristic consisting of multiple dimensions that include sensation seeking, lack of planning, lack of persistence and urgency1, 2. The non-planning dimension in particular appears to be more strongly associated with negative risk taking (i.e. binge eating, problem gambling,2). The same holds true for the urgency dimension which predicts problem behaviors (problem gambling and drinking) whereas factors such as sensation seeking are perhaps more related to risk taking in general2, 3.

Impulsivity as it refers to pathological behavior has been well studied in current and former substance users. In humans, opiate addicts are more impulsive than non-addicts as measured by an increase in the discounting of delayed rewards (i.e. the devaluation of rewards as a function of time)4 and by a reduction in reflection (i.e. the tendency to use information when making a decision)5. In animals, it has been shown that in rodents a preference for immediate reward over larger, delayed reward predicted the development of nicotine self-administration and maintenance of use6, suggesting that this is a trait that predisposes to drug dependence. This form of impulsive choice in rats has also been demonstrated to predict cocaine self-administration7. Other clinical populations are also characterized by or associated with impulsive behavior such as Attention-Deficit Hyperactivity Disorder (ADHD)8, Borderline Personality Disorder9, eating disorders10 and pathological gambling11. Whether impulsivity represents exclusively a predisposing factor or is at least partially the result of prior exposure to drugs or ongoing disease is a matter of current debate.

Impulsive characteristics probably do not affect behavior in isolation but are also likely to interact with other factors such as stress. Stressors have a negative impact on drug initiation, maintenance, craving and relapse12. High impulsive gamblers also show neuroendocrine stress axis and cardiovascular responses to gambling situations relative to their low impulsive counterparts13. In addition, stressors, particularly when combined with substance abuse, are thought to be modulated by individual impulsivity traits to increase the risk of completed suicides14, particularly in younger individuals15.

While it is increasingly clear that impulsivity and stress responses confer vulnerability to substance abuse and other risky behaviors, the neurobiological processes underlying these effects are still poorly understood, particularly in humans. DA neurotransmission appears to be one of the mechanisms involved. Utilizing an animal model of impulsivity that examines anticipatory responses to a food reward as proxy, Dalley and colleagues observed that rats demonstrating greater impulsivity prior to drug exposure exhibited lower dopamine DA D2/3 receptor concentrations in the nucleus accumbens, increased escalation and maintenance of drug self-administration relative to their lower impulsivity counterparts16. Though DA function in the ventral basal ganglia is thought to have an important role1719 it is unlikely to take place in isolation. The nucleus accumbens lies at the interface of sensorimotor and limbic systems, and through its connections with the ventral pallidum and the amygdala, forms part of a circuit involved in the integration of cognitive, affective and motor responses20, 21. This pathway and interconnected regions (e.g., insular and prefrontal cortex, medial thalamus) are heavily modulated by the endogenous opioid system and μ-opioid receptors. For example, this neurotransmitter system is recruited when drug-induced DA release takes place in the context of environmental novelty and stressors2226. Further, the motivated pursuit and positive behavioral responses to rewards27, 28 are enhanced by the selective administration of μ-opioid receptor agonists in the nucleus accumbens and ventral pallidum, nuclei that are central to the regulation of motivated behavior.

The present report examined two orthogonal behavioral traits, impulsiveness (IMP) and deliberation (DLB), as defined by the NEO Personality Inventory Revised (NEO PI-R)29 as a function of in vivo measures of μ-opioid receptor neurotransmission in humans. As defined by the NEO PI-R, the IMP dimension refers to the tendency to act without careful consideration for consequences of delayed gratification, and maps on to urgency, which appears related to problem behaviors such as drug use2. DLB, which corresponds to the lack of planning dimension, is thought to act as a moderating, opposing trait30.

We utilized positron emission tomography (PET) and the μ-opioid selective radiotracer [11C]carfentanil at rest and during the experience of a physical and emotional stressor, moderate levels of sustained pain. Under these experimental conditions, reductions in the availability of μ-opioid receptors during the stress challenge reflect the activation of endogenous opioid neurotransmission and μ-opioid receptors31. It was hypothesized that individual levels of IMP and DBL would be positively and negatively associated, respectively, with the functional response of the μ-opioid system during the stressor. Furthermore, that these effects would take place in motivational circuits modulated by this neurotransmitter system, namely the rostral anterior cingulate and adjacent medial prefrontal cortex, nucleus accumbens/ventral pallidum, amygdala and medial thalamus.

Materials and Methods


Nineteen (19) healthy right-handed, non-smoking men (age range 20–30 years; mean ± SD, 23 ± 3 years) were recruited via advertisement. In addition to completing physical and neurological examinations, subjects were screened using the Structured Clinical Interview for DSM-IV (non-patient version, SCID-NP). Subjects had no current or past history of medical, neurological, or psychiatric illnesses, including substance abuse or dependence and alcohol intake less than 5 drinks/week. Participants reported no current or recent (within 6 months) history of exposure to centrally active prescription or illicit drugs and were asked to restrain from any alcohol intake for 48 hours prior to scanning. Urine drug screens were performed immediately prior to imaging. Subjects reported no family history of psychiatric disease in first-degree relatives. The sample was restricted to males due to the known sex differences in the regional concentration of μ-opioid receptors32, and in the activity of this neurotransmitter system in response to stress33, an effect that is influenced by circulating gonadal steroids34. Furthermore, a link between impulsivity and substance use disorders has been shown most conclusively in males (see Discussion).

Protocols were approved by the Investigational Review Boards of the Universities of Michigan and Maryland and the Radioactive Drug Research Committee at the University of Michigan. Written informed consent was obtained in all subjects.

Personality Inventories

Subjects were administered the NEO Personality Inventory Revised (NEO PI-R)29. The facets “Impulsiveness” (IMP), defined as “a lack of control over cravings or desires”, and “Deliberation” (DLB), or the “tendency to think carefully before acting”, were utilized as the primary scales of interest. These facets have been previously demonstrated to reflect the dimensions of impulsivity that have been associated with negative risk taking2, 3. Individuals endorsing greater behavioral under-control or lack of reflection would display higher IMP and lower DLB scores. The median scores in population samples of comparable age were utilized to separate the study sample into high and low scoring groups (population data, IMP, mean ± SD, 15 ± 4; DLB, mean ± SD, 18 ± 429. Study sample data, IMP, mean ± SD, 15 ± 3; DLB, mean ± SD, 18 ± 3 (Median = 15, High, n = 9 subjects, Low, n = 10; DLB: median = 18, High, n = 9, Low, n = 10).

Stress Challenge

We employed a physical and emotional stressor, moderate levels of sustained pain of experientially adjusted intensity, to activate endogenous opioid-μ-opioid mediated neurotransmission, as previously described31, 35. In short, a steady state of moderate muscle pain was maintained 45–65 min after the radiotracer administration by a computer-controlled delivery system through the infusion of medication-grade hypertonic saline (5%) into the left masseter muscle. In this model of sustained deep somatic pain, the intensity of the painful stimulus is standardized across subjects, as described in detail previously36, 37. Pain intensity was rated every 15 sec from 0 (no pain) to 100 (most intense pain imaginable). During the baseline control condition, no infusions took place and the subject was instructed to lie quietly in the scanner. The pain intensity ratings obtained every 15 sec were recorded in the computer controller and averaged for statistical analyses.

Integrative measures of the pain experience (sensory and pain affect components) were obtained using the McGill pain questionnaire (MPQ), administered upon completion of the challenge38. The Positive and Negative Affectivity Scale (PANAS)39, assessing the internal affective state of the volunteers, was obtained before and after the challenge. The infusion volume required for pain maintenance was also recorded and provided a measure of sustained pain sensitivity for the individual subject.

Scanning Protocols

Two PET scans per subject were acquired with a Siemens HR+ scanner (Knoxville, KN) in 3-D mode (reconstructed FWHM resolution (5.5 mm in-plane and 5.0 mm axially), one at baseline and another using the stress challenge. Radiotracer synthesis and image acquisition, co-registration and reconstruction protocols were identical to those used in previous publications31, 33, 35.

The total activity of [11C]carfentanil administered to each subject in each scan was 14.5 ± 2.7 mCi (535.0 ± 100.9 MBq), with an average mass injected of 0.02 ± 0.01 μg/kg, ensuring that the compound was administered in tracer quantities, i.e., sub-pharmacological doses. Fifty (50) percent of the [11C]carfentanil dose was administered as a bolus with the remainder delivered as a continuous infusion by a computer-controlled automated pump to more rapidly achieve steady-state tracer levels.

Dynamic image data for each of the receptor scans were transformed, on a voxel-by-voxel basis, into two sets of parametric maps, coregistered to each other: (a) a tracer transport measure (K1 ratio), proportional to regional cerebral blood flow, and (b) a receptor-related measure, distribution volume ratio at equilibrium (DVReq). To avoid the need for arterial blood sampling, these parametric images were calculated using a modified Logan graphical analysis40, using the occipital cortex (an area devoid of μ-opioid receptors) as the reference region. The Logan plot became linear by 5–7 minutes after the start of radiotracer administration, with a slope proportional to the (Bmax/Kd)+1 for this receptor site. Bmax/Kd is the “receptor related” measure (μ-opioid receptor availability or μ-opioid receptor “binding potential”, BPND) (Bmax = receptor concentration; Kd = receptor affinity for the radiotracer).

MRI scans were acquired on a 1.5 Tesla scanner (Signa, General Electric, Milwaukee, WI) for anatomical localization and coregistration to standardized stereotactic coordinates. Acquisition sequences were axial spoiled gradient-recalled (SPGR) MR [echo time (TE)= 5.5; repetition time (TR)= 14; inversion time (TI)= 300; flip angle= 20°; number of excitations (NEX)= 1; 124 contiguous images; 1.5 mm-thick; 24 cm field-of-view; image matrix= 256 × 256 pixels, pixel size= 0.94 mm). T1-weighted MR and PET images of each subject were then co-registered to each other using a mutual information algorithm as previously described31, 35.

Image Analyses

Differences between groups (high IMP / low IMP; low DLB / high DLB, two tail, unpaired t tests) were mapped into stereotactic space using z maps of statistical significance with a modified version of SPM99 (Welcome Department of Cognitive Neurology, University College, London) and Matlab software (MathWorks, Natick, MA), using a general linear model. No global normalization was applied to the data, and therefore the calculations presented are based on absolute Bmax/Kd estimates. Only regions with specific μ-opioid receptor binding were included in the analyses (voxels with BPND values > 0.2 as calculated with SPM). A priori hypothesized regions were deemed significant at p < 0.0001 uncorrected for multiple comparisons. For other regions, significant differences were detected using a statistical threshold that controls for a Type-I error rate at p = 0.05 for multiple comparisons41. Numerical values for each region were obtained by averaging the values of voxels contained in each significant cluster, up to a p = 0.001. These data were extracted for quantification of regional changes in BP, graphing, determination of correlation coefficients (Pearson correlations at p < 0.05), rule out the presence of outliers, and further statistical analyses with SPSS for Macintosh 11.0.3 (SPSS Inc., Chicago, IL).


Psychophysical Measures

The use of the adaptive, experientially-adjusted stimulus delivery system employed, produced comparable perceptions of the pain stressor among participants by individually titrating the rate of infusion of the algesic substance. No significant group differences were obtained in psychophysical measures of pain or affective state during the stress challenge (high vs. low IMP, high vs. low DLB) (Table 1). As would be expected, IMP and DLB scores were negatively correlated (r = −0.55, p = 0.015).

Table 1
Psychophysical Measures

Baseline μ-Opioid Receptor BP


Significant differences in baseline μ-opioid receptor BPND were observed between high and low IMP groups. Specifically, greater regional μ-opioid receptor BPND was observed in the high, compared to low IMP subjects, in the right anterior cingulate and adjacent medial frontal cortex, right ventral basal ganglia (nucleus accumbens, extending into ventral pallidum), and the basolateral area of the right amygdala (Table 2, Figure 1). No effects were obtained in the opposite direction.

Figure 1
Association between Impulsiveness and Deliberation scores and baseline μ-opioid receptor availability.
Table 2
Differences in Baseline Regional μ-Opioid Receptor BP

Significant positive correlations were obtained between μ-opioid receptor BPND and IMP scores within all these clusters (right dorsal anterior cingulate, r = 0.59, p < 0.01; right ventral basal ganglia, r = 0.50, p = 0.03; right amygdala, r = 0.49, p = 0.03).


Subjects with high DLB scores showed significantly lower baseline regional μ-opioid receptor BPND compared to the low DLB group in the right dorsolateral prefrontal cortex, right dorsal anterior cingulate and medial frontal gyrus, left ventral basal ganglia, right thalamus extending inferiorly into hypothalamus and in the right basolateral amygdala (Table 2, Figure 1). No effects were observed in the opposite direction.

Significant negative correlations between μ-opioid receptor BPND and DLB scores were noted within the right prefrontal cortex (r = − 0.65, p = 0.003), right anterior cingulate (two peaks, x,,y,,z, coordinates in mm, 16, 10, 39, r = − 0.56, p = 0.01; x, y, z, 8, 14, 26, r = − 0.46, p < 0.05), and right amygdala (r = − 0.54, p = 0.02).

Stress – Induced Activation of μ-Opioid Neurotransmission


Subjects with higher IMP scores demonstrated significantly greater stress-induced activation of μ-opioid mediated neurotransmission, compared to subjects in the low IMP group, in the left orbitofrontal cortex, right dorsal anterior cingulate, ventral basal ganglia, bilaterally, extending into the hypothalamus, left anterior thalamus and basolateral amygdala bilaterally. (Table 3, Figure 2). There were no regions where the high IMP group showed significantly lower stress-induced changes in μ-opioid receptor BPND relative to the low IMP group.

Figure 2
Association between Impulsiveness and Deliberation scores and stress-induced activation of μ-opioid receptor mediated neurotransmission.
Table 3
Differences in Stress-Induced Changes in Regional μ-Opioid Receptor BP

Significant positive correlations between μ-opioid system activation (baseline BPND – pain BPND) and IMP scores were noted in the left orbitofrontal cortex (r = 0.63, p < 0.005), right ventral basal ganglia (r = 0.49, p = 0.03), left anterior thalamus (r = 0.61, p = 0.006), and right amygdala (r = 0.50, p < 0.03).


Significant differences in stress-induced activation of endogenous opioid neurotransmission were also detected between the high DLB and low DLB groups. Opposite to the high IMP group and in a direction similar to that observed for baseline binding measures, the high DLB group showed lower stress-induced activation of the endogenous opioid system compared to the low DLB group in a number of brain regions. These included the left dorsolateral prefrontal cortex, right anterior cingulate/medial frontal cortex, orbitofrontal cortex bilaterally, ventral basal ganglia, bilaterally, with extension into the anterior hypothalamus, and basolateral amygdala bilaterally. No effects were obtained in the opposite direction (Table 3, Figure 2).

Significant negative correlations between μ-opioid system activation and DLB scores were observed within the left dorsolateral prefrontal cortex (r = − 0.69, p = 0.001), right anterior cingulate (r = − 0.61, p = 0.005), right and left ventral basal ganglia (r = − 0.54, p < 0.02 and r = − 0.60, p < 0.01, respectively), and right amygdala (r = − 0.70, p = 0.001). The left amygdala cluster followed a similar pattern at trend levels of correlation (r = − 0.41, p = 0.08).

Interactions Among Measures: Conjunction Analyses

The above data suggested the presence of some, but not complete regional overlap for the effects of the related traits IMP and DLB, with neurochemical findings in opposite directions, as would be expected for opposing traits. In an additional analysis, we sought to determine how the individual combination of these two behavioral traits segregated at the levels of anatomical and neurochemical substrates (μ-opioid receptor availability and neurotransmitter responses to stress). Individuals were divided into “behavioral risk” groups based on their IMP and DLB classifications, resulting in three groups with relatively high (High IMP/Low DLB, n=7), low (Low IMP/High DLB, n=7) or intermediate (High IMP/High DLB, Low IMP/Low DLB, n=5) behavioral trait vulnerability. Intermediate groups were not separated because of the small sample sizes in those cells. Then, for both the baseline and activation conditions, we identified brain areas of coincidence, where BPND and stress-induced release were greater in both the high IMP and low DLB groups. For this purpose, the ImCalc function within SPM was utilized to generate a “mask” that contained only those voxels that were significantly different above a p =0.007 (T=1.99) in both of the contrasts (Contrast1 = High vs Low IMP, Contrast 2 = Low vs High IMP; Formula: [(Contrast 1 T score > 1.99)*(Contrast 2 T score > 1.99)]. The resulting area of coincidence contained voxels that were independently significant in each and both of the contrasts, whose joint probability is given by multiplying the probabilities for each contrast: 0.007 × 0.007 = p ≤ 0.000049 (e.g. 42). Measurement values for the regions identified were then extracted for quantification of regional changes in BPND, graphing, and statistical analyses.


Three regions showed significant overlap among the two traits: right anterior cingulate, right ventral pallidum, and right amygdala. The baseline BPND for each of the regions analyzed increased in a stepwise progression from the group with the lowest vulnerability traits (Low IMP/High DLB) to the highest (High IMP/Low DLB) (Figure 3, Table 4). This effect was tested statistically with ANOVA, which showed significant effects of “risk” classification for each of the regions ([right anterior cingulate (F(2,16) = 8.009, p = 0.004), right nucleus accumbens/ventral pallidum (F(2,16) = 8.157, p=0.004), right amygdala (F(2,16) = 5.280, p = 0.017)]). Post-hoc tests (Tukey HSD) are shown in Table 4. Of note, both intermediate groups High IMP/High DLB and Low IMP/Low DLB showed similar results for these regions (data not shown).

Figure 3
Conjunction analysis of impulsiveness and deliberation effects on μ-opioid receptor BP and stress-induced endogenous opioid system activity.
Table 4
Conjunction Analyses

Stress –Induced Activation of μ-Opioid Neurotransmission

Five regions showed significant activation overlap among traits: right anterior cingulate, right and left nucleus accumbens/ventral pallidum, and right and left amygdala. Again, we observed a stepwise progression in stress-induced endogenous opioid system activation, with the smallest change in the group with the fewest vulnerable behavioral traits (Low IMP/High DLB) and the greatest from the group with the most (High IMP/Low DLB) (Figure 3, Table 4). A significant main effect of group on stress-induced activation of μ-opioid neurotransmission was present for each of the regions ([right anterior cingulate (F(2,16) = 10.53, p=0.001), right ventral pallidum (F(2,16) = 40.91, p<0.0001), left ventral pallidum (F(2,16) = 5.10, p=0.019), right amygdala (F(2,16) = 23.54, p <0.001), left amygdala (F(2,16) = 5.05, p=0.020)]). Post-hoc tests results are shown in Table 4. As with the baseline data, intermediate groups showed similar results for the overlapping regions (data not shown).


The present study demonstrates that impulsiveness and deliberation are highly predicted by measures of endogenous opioid function in limbic regions. The personality facets studied here refer to the tendency to act rashly and without forethought, and have been associated with various psychopathologies and risky phenotypes (e.g., drug consumption, pathological gambling, personality disorders)2, 3, 9, 11. Our major findings are: First, we find that individuals displaying these risky phenotypes (e.g. high IMP or low DBL) have higher μ-opioid receptor BPND at rest within regions implicated in decision making, reward seeking and emotional responsivity. This higher BPND reflects a greater availability of μ-opioid receptors in a high affinity state (e.g., binding to an agonist radiotracer at low, tracer concentrations)43. Second, following a pain stress challenge we find larger reductions in BPND from baseline in individuals displaying high IMP/low DLB in overlapping regions. These reductions reflect processes related to the release of endogenous opioid interacting with μ-opioid receptors, so these receptors are no longer available for binding to the radioligand43, 44. Third, we demonstrate a cumulative effect of personality traits on in vivo measures of μ-opioid neurotransmission. We found that individuals exhibiting extreme traits (high IMP/low DLB and low IMP/high DLB) display the greatest and smallest, respectively, baseline μ-opioid receptor availability and endogenous opioid system responses to the pain stressor employed.

Personality traits, like impulsiveness, likely manifest as a result of a variety of factors, both biological and genetic. Converging lines of evidence point to the opioid system as one candidate system involved in the expression of the non-planning dimension of impulsiveness. Previous research on a measure related to the non-planning dimension of impulsiveness, delayed discounting, which refers to the devaluation of rewards as a function of time, has indicated prominent roles for several neurotransmitters: serotonin45, dopamine45 and based upon the present results, opioids. Manipulation of the opioid system affects preferences for immediate rewards; for instance, and in animal models, Kieres and colleagues demonstrated that morphine could increase the rate of delayed discounting among rats, an effect blocked by naloxone46. Few human studies have directly addressed this issue, however, multiple studies have shown that several psychiatric groups show steeper discounting of delayed rewards such as pathological gamblers47 and drug addicts (e.g. to opiates4). In addition, opiate addicts show a greater preference to immediate monetary rewards relative to non-addicts4, a preference that is potentiated following mild opiate deprivation48.

In the present work we show that individuals displaying risky personality traits (high IMP, low DLB) showed significantly greater regional μ-opioid receptor availability at baseline and stress-induced regional μ-opioid system activation when compared to individuals endorsing low IMP, high DLB. These effects were observed in multiple brain regions including the orbitofrontal, medial prefrontal and cingulate cortex, nucleus accumbens/ventral pallidum and amygdala. Individually, these regions are known to be involved in impulsive choice, reward seeking, and cognitive-emotional integration and are heavily modulated by μ-opioid receptors4951. Many of these regions, particularly the prefrontal cortex and nucleus accumbens have been implicated in disorders characterized by or associated with impulsive behavior such as ADHD52, substance abuse disorders53, and pathological gambling54. Manipulation of nucleus accumbens activity can directly influence impulsive behavior, i.e, stimulation of the nucleus accumbens core has been shown to decrease impulsive choice55, whereas lesions increase impulsive choice56. Similar roles have been ascribed to the prefrontal cortex, orbitofrontal and amygdala, thought to contribute to decision making by the cognitive and emotional evaluation of future consequences57, 58. Collectively, these regions are thought to be involved in the pursuit and receipt of natural rewards, decision-making and, more generally, motivated behavior25, 5965. Neurobiologically, this regulation of motivated behavior is thought to take place as a result of their extensive reciprocal connections, well-described between the nucleus accumbens, ventral pallidum, mediodorsal nucleus of the thalamus, prefrontal cortex and amygdala66, 67.

We also observed greater stress-induced activation of this neurotransmitter system in subjects scoring above the population average of NEO IMP scores, compared with subjects scoring below, in regions at least partially overlapping with those where baseline differences were observed. Opposite effects (lower stress-induced opioid system activity in high scoring subjects) were observed for the orthogonal domain, DLB. These data then supports the contention that there are interactions between neurobiological processes related to stress responsiveness and impulsivity. Physiological stress responses seem greater in more impulsive individuals even among risky populations (e.g., pathological gamblers13) and therefore point to factors that may contribute to interindividual variations in risky behavior in various pathological states. Outbred rats exposed to the mild stress of a novel environment may show high (HR) or low (LR) rates of exploratory locomotion, and HR rats learned to self-administer psychostimulants faster than LRs6871. It has been proposed that activation of DA neurotransmission and stress responses during risky behavior are the critical variable underlying the reinforcement of this behavior in the more impulsive individuals72, an effect that may be mediated by the increase in corticosterone induced by the stressor7376. Relevant to the results presented here, HR rats, more prone to acquire drug-self administration also show increased nucleus accumbens proenkephalin gene expression77.

A conjunction analysis more formally determined the overlap in the processes and brain regions where IMP and DLB effects were obtained. It demonstrated a cumulative effect of personality risk factors on measures of μ-opioid neurotransmission. Extreme traits (high IMP/low DLB, low IMP/high DLB) demonstrated greatest and smallest, respectively, endogenous opioid system responses to a standardized stressor and μ-opioid receptor availability at baseline. “Intermediate”, compounded traits (high IMP/high DLB, low IMP/low DLB) showed intermediate effects for both measures. This is consistent with the observation that the accumulation of risky traits is associated with a greater probability of problem behaviors and substance use problems78. The coalescence of IMP and DLB effects were observed in the dorsal anterior cingulate, nucleus accumbens/ventral pallidum and amygdala, centrally implicated in decision-making and motivated behavior, as noted above20, 21, 79.

Regional μ-opioid receptor availability and μ-opioid system activation during the stressor accounted for 24 to 40% of the variance in IMP scores, and 17 to 49% of the variance in DBL scores. In contrast, no significant relationships have been reported between NEO impulsiveness and dopamine D2/3 receptor binding in the basal ganglia as measured with [11C]raclopride80 or with dopamine turnover as measured with [18F]fluorodopa81. Amphetamine-induced dopamine release in the ventral basal ganglia accounted for 9–20% of the variance in NEO impulsiveness scores in a healthy sample similar to the one studied in the present report80.

Because the study sample was restricted to males to reduce experimental complexity, additional questions remain that will need to be addressed in subsequent work. Effects of gender, gonadal steroids and age by gender interactions have been described for μ-opioid receptors and stress-induced μ-opioid system activation3234. These effects may or may not be related to IMP and DBL traits and will require specific studies addressing their effects. From a different perspective, impulsive behavior has been suggested to be a result of prefrontal cortex dysfunction. For instance, Bechara and others have described problems with decision-making, specifically insensitivity to future consequences, following damage to the ventromedial prefrontal cortex57, 58. The relationship between ventral prefrontal cortex function and endogenous opioid system activity measures are presently unexplored.

It is also unlikely that complex personality domains are solely related to a single neurotransmitter system. Indeed, DA D2/3 receptor concentration within the ventral basal ganglia has been demonstrated to predict impulsive anticipatory responses to food reward in an animal model of impulsivity16. The involvement of DA mechanisms, however, is not exclusionary of effects by other systems, such as the endogenous opioid. Dopamine-opioid interactions have been described in the striatopallidal pathway and interconnected regions, where acute and chronic DA receptor stimulation induce opposite effects on the functional capacity of the μ-opioid system in animal models8287 and in humans35, 88. These and other, not yet described neurotransmitter systems may underlie the psychophysical differences, such as heart rate and pupilary responses, classically noted between otherwise healthy shy and uninhibited children89.

The present study provides the first evidence in humans that IMP and DBL, behavioral facets relevant to motivated behavior, the pursuit of reward and risk taking, including the development of substance use disorders, are related to the individual function of the endogenous opioid system. Baseline measures of μ-opioid receptor availability and the capacity to activate this neurotransmitter system in limbic and paralimbic regions in response to stress accounted for up to half of the variance in trait IMP and DBL scores in a healthy sample.


Grant Support: Supported by the National Institute of Drug Abuse grant R01 DA 016423


Disclosures: No known potential conflicts.


1. Whiteside SP, Lynam DR. The Five Factor Model and impulsivity: using a structural model of personality to understand impulsivity. Personality and Individual Differences. 2001 Dec 21;30:669–689.
2. Smith GT, Fischer S, Cyders MA, Annus AM, Spillane NS, McCarthy DM. On the validity and utility of discriminating among impulsivity-like traits. Assessment. 2007 Jun 1;14(2):155–170. [PubMed]
3. Fischer S, Anderson KG, Smith GT. Coping with distress by eating or drinking: role of trait urgency and expectancies. Psychology of addictive behaviors: journal of the Society of Psychologists in Addictive Behaviors. 2004 Sep 1;18(3):269–274. [PubMed]
4. Madden GJ, Petry NM, Badger GJ, Bickel WK. Impulsive and self-control choices in opioid-dependent patients and non-drug-using control participants: drug and monetary rewards. Experimental and clinical psychopharmacology. 1997 Aug 1;5(3):256–262. [PubMed]
5. Clark L, Robbins TW, Ersche KD, Sahakian BJ. Reflection impulsivity in current and former substance users. Biological psychiatry. 2006 Sep 1;60(5):515–522. [PubMed]
6. Diergaarde L, Pattij T, Poortvliet I, et al. Impulsive choice and impulsive action predict vulnerability to distinct stages of nicotine seeking in rats. Biological psychiatry. 2008 Feb 1;63(3):301–308. [PubMed]
7. Perry JL, Nelson SE, Carroll ME. Impulsive choice as a predictor of acquisition of IV cocaine self- administration and reinstatement of cocaine-seeking behavior in male and female rats. Experimental and clinical psychopharmacology. 2008 Apr 1;16(2):165–177. [PubMed]
8. Swanson JM, Sergeant JA, Taylor E, Sonuga-Barke EJ, Jensen PS, Cantwell DP. Attention-deficit hyperactivity disorder and hyperkinetic disorder. Lancet. 1998 Feb 7;351(9100):429–433. [PubMed]
9. American Psychiatric Association . Diagnostic and statistical manual of mental disorders: DSM-IV-TR. 4th ed. American Psychiatric Association; Washington, DC: 2000. American Psychiatric Association. Task Force on DSM-IV.
10. Fahy T, Eisler I. Impulsivity and eating disorders. The British journal of psychiatry: the journal of mental science. 1993 Feb 1;162:193–197. [PubMed]
11. Alessi SM, Petry NM. Pathological gambling severity is associated with impulsivity in a delay discounting procedure. Behav Processes. 2003 Oct 31;64(3):345–354. [PubMed]
12. Sinha R. How does stress increase risk of drug abuse and relapse? Psychopharmacology (Berl) 2001 Dec 1;158(4):343–359. [PubMed]
13. Krueger TH, Schedlowski M, Meyer G. Cortisol and heart rate measures during casino gambling in relation to impulsivity. Neuropsychobiology. 2005 Jan 1;52(4):206–211. [PubMed]
14. Mann JJ, Waternaux C, Haas GL, Malone KM. Toward a clinical model of suicidal behavior in psychiatric patients. Am J Psychiatry. 1999 Feb 1;156(2):181–189. [PubMed]
15. Zouk H, Tousignant M, Seguin M, Lesage A, Turecki G. Characterization of impulsivity in suicide completers: clinical, behavioral and psychosocial dimensions. J Affect Disord. 2006 Jun 1;92(2–3):195–204. [PubMed]
16. Dalley JW, Fryer TD, Brichard L, et al. Nucleus accumbens D2/3 receptors predict trait impulsivity and cocaine reinforcement. Science. 2007 Mar 2;315(5816):1267–1270. [PMC free article] [PubMed]
17. Rouge-Pont F, Deroche V, Le Moal M, Piazza PV. Individual differences in stress-induced dopamine release in the nucleus accumbens are influenced by corticosterone. Eur J Neurosci. 1998 Dec 1;10(12):3903–3907. [PubMed]
18. Rouge-Pont F, Piazza PV, Kharouby M, Le Moal M, Simon H. Higher and longer stress-induced increase in dopamine concentrations in the nucleus accumbens of animals predisposed to amphetamine self-administration. A microdialysis study. Brain Res. 1993 Jan 29;602(1):169–174. [PubMed]
19. Thierry AM, Tassin JP, Blanc G, Glowinski J. Selective activation of mesocortical DA system by stress. Nature. 1976 Sep 16;263(5574):242–244. [PubMed]
20. Kalivas PW, Churchill L, Romanides A. Involvement of the pallidal-thalamocortical circuit in adaptive behavior. Ann N Y Acad Sci. 1999 Jun 29;877:64–70. [PubMed]
21. Mogenson GJ, Yang CR. The contribution of basal forebrain to limbic-motor integration and the mediation of motivation to action. Adv Exp Med Biol. 1991;295:267–290. [PubMed]
22. Badiani A, Oates MM, Day HE, Watson SJ, Akil H, Robinson TE. Amphetamine-induced behavior, dopamine release, and c-fos mRNA expression: modulation by environmental novelty. J Neurosci. 1998 Dec 15;18(24):10579–10593. [PubMed]
23. Badiani A, Oates MM, Day HE, Watson SJ, Akil H, Robinson TE. Environmental modulation of amphetamine-induced c-fos expression in D1 versus D2 striatal neurons. Behav Brain Res. 1999 Sep 1;103(2):203–209. [PubMed]
24. Day HE, Badiani A, Uslaner JM, et al. Environmental novelty differentially affects c-fos mRNA expression induced by amphetamine or cocaine in subregions of the bed nucleus of the stria terminalis and amygdala. J Neurosci. 2001 Jan 15;21(2):732–740. [PubMed]
25. Napier TC, Mitrovic I. Opioid modulation of ventral pallidal inputs. Ann N Y Acad Sci. 1999 Jun 29;877:176–201. [PubMed]
26. Uslaner J, Badiani A, Day HE, Watson SJ, Akil H, Robinson TE. Environmental context modulates the ability of cocaine and amphetamine to induce c-fos mRNA expression in the neocortex, caudate nucleus, and nucleus accumbens. Brain Res. 2001 Nov 30;920(1–2):106–116. [PubMed]
27. Pecina S, Berridge KC. Hedonic hot spot in nucleus accumbens shell: where do muopioids cause increased hedonic impact of sweetness? J Neurosci. 2005 Dec 14;25(50):11777–11786. [PubMed]
28. Smith KS, Berridge KC. Opioid limbic circuit for reward: interaction between hedonic hotspots of nucleus accumbens and ventral pallidum. J Neurosci. 2007 Feb 14;27(7):1594–1605. [PubMed]
29. Costa P, McRae R, Anonymous Normal personality assessment in clinical practice: The NEO Personality Inventory. Psychological Assessment. 1992;4:5–13.
30. Fisher S, Smith GT. Deliberation aff ects risk taking beyond sensation seeking. Personality and Individual Differences. 2004 Dec 19;36:527–537.
31. Zubieta JK, Smith YR, Bueller JA, et al. Regional mu opioid receptor regulation of sensory and affective dimensions of pain. Science. 2001 Jul 13;293(5528):311–315. [PubMed]
32. Zubieta JK, Dannals RF, Frost JJ. Gender and age influences on human brain mu-opioid receptor binding measured by PET. Am J Psychiatry. 1999 Jun 1;156(6):842–848. [PubMed]
33. Zubieta JK, Smith YR, Bueller JA, et al. mu-opioid receptor-mediated antinociceptive responses differ in men and women. J Neurosci. 2002 Jun 15;22(12):5100–5107. [PubMed]
34. Smith YR, Stohler CS, Nichols TE, Bueller JA, Koeppe RA, Zubieta JK. Pronociceptive and antinociceptive effects of estradiol through endogenous opioid neurotransmission in women. J Neurosci. 2006 May 24;26(21):5777–5785. [PMC free article] [PubMed]
35. Zubieta JK, Heitzeg MM, Smith YR, et al. COMT val158met genotype affects mu-opioid neurotransmitter responses to a pain stressor. Science. 2003 Feb 21;299(5610):1240–1243. [PubMed]
36. Stohler CS, Kowalski CJ. Spatial and temporal summation of sensory and affective dimensions of deep somatic pain. Pain. 1999 Feb 1;79(2–3):165–173. [PubMed]
37. Zhang X, Ashton-Miller JA, Stohler CS, Anonymous A closed-loop system for maintaining constant experimental muscle pain in man. IEEE transactions on biomedical engineering. 1993 Apr 1;40(4):344–352. [PubMed]
38. Melzack R, Torgerson W. On the language of pain. Anesthesiology. 1971;34:50–59. [PubMed]
39. Watson D, Clark L, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. J Personal Soc Psychol. 1988;54:1063–1070. [PubMed]
40. Logan J, Fowler JS, Volkow ND, et al. Distribution volume ratios without blood sampling from graphical analysis of PET data. Journal of cerebral blood flow and metabolism: official journal of the International Society of Cerebral Blood Flow and Metabolism. 1996 Sep 1;16(5):834–840. [PubMed]
41. Friston KJ, Worsley KJ, Frackowiak RS, Mazziotta JC, Evans AC. Assessing the significance of focal activations using their spatial extent. Human Brain Mapping. 1994;1:210–220. [PubMed]
42. Dolcos F, LaBar KS, Cabeza R. Dissociable effects of arousal and valence on prefrontal activity indexing emotional evaluation and subsequent memory: an event-related fMRI study. Neuroimage. 2004 Sep 1;23(1):64–74. [PubMed]
43. Narendran R, Hwang DR, Slifstein M, et al. In vivo vulnerability to competition by endogenous dopamine: comparison of the D2 receptor agonist radiotracer (−)-N-[11C]propyl-norapomorphine ([11C]NPA) with the D2 receptor antagonist radiotracer [11C]-raclopride. Synapse. 2004 Jun 1;52(3):188–208. [PubMed]
44. Innis RB, Malison RT, al-Tikriti M, et al. Amphetamine-stimulated dopamine release competes in vivo for [123I]IBZM binding to the D2 receptor in nonhuman primates. Synapse. 1992 Mar 1;10(3):177–184. [PubMed]
45. Winstanley CA, Theobald DE, Dalley JW, Robbins TW. Interactions between serotonin and dopamine in the control of impulsive choice in rats: therapeutic implications for impulse control disorders. Neuropsychopharmacology. 2005 Apr 1;30(4):669–682. [PubMed]
46. Kieres AK, Hausknecht KA, Farrar AM, Acheson A, de Wit H, Richards JB. Effects of morphine and naltrexone on impulsive decision making in rats. Psychopharmacology (Berl) 2004 Apr 1;173(1–2):167–174. [PubMed]
47. Dixon MR, Marley J, Jacobs EA. Delay discounting by pathological gamblers. Journal of applied behavior analysis. 2004 Feb 11;36(4):449–458. [PMC free article] [PubMed]
48. Giordano LA, Bickel WK, Loewenstein G, Jacobs EA, Marsch L, Badger GJ. Mild opioid deprivation increases the degree that opioid-dependent outpatients discount delayed heroin and money. Psychopharmacology (Berl) 2002 Sep 1;163(2):174–182. [PubMed]
49. Khachaturian H, Watson SJ. Some perspectives on monoamine-opioid peptide interaction in rat central nervous system. Brain Res Bull. 1982 Jul 1;9(1–6):441–462. [PubMed]
50. Mansour A, Meador-Woodruff JH, Camp DM, et al. The effects of nigrostriatal 6-hydroxydopamine lesions on dopamine D2 receptor mRNA and opioid systems. Prog Clin Biol Res. 1990;328:227–230. [PubMed]
51. Vogt LJ, Sim-Selley LJ, Childers SR, Wiley RG, Vogt BA. Colocalization of mu-opioid receptors and activated G-proteins in rat cingulate cortex. J Pharmacol Exp Ther. 2001 Dec 1;299(3):840–848. [PubMed]
52. Spencer TJ, Biederman J, Wilens TE, Faraone SV. Overview and neurobiology of attention-deficit/hyperactivity disorder. The Journal of clinical psychiatry. 2002 Jan 1;63(Suppl 12):3–9. [PubMed]
53. Volkow ND, Fowler JS, Wang GJ. The addicted human brain viewed in the light of imaging studies: brain circuits and treatment strategies. Neuropharmacology. 2004 Jan 1;47(Suppl 1):3–13. [PubMed]
54. Reuter J, Raedler T, Rose M, Hand I, Gläscher J, Büchel C. Pathological gambling is linked to reduced activation of the mesolimbic reward system. Nat Neurosci. 2005 Feb 1;8(2):147–148. [PubMed]
55. Sesia T, Temel Y, Lim LW, Blokland A, Steinbusch HW, Visser-Vandewalle V. Deep brain stimulation of the nucleus accumbens core and shell: Opposite effects on impulsive action. Exp Neurol. 2008 Jul 29; [PubMed]
56. Cardinal RN, Pennicott DR, Sugathapala CL, Robbins TW, Everitt BJ. Impulsive choice induced in rats by lesions of the nucleus accumbens core. Science. 2001 Jun 29;292(5526):2499–2501. [PubMed]
57. Bechara A, Damasio AR, Damasio H, Anderson SW. Insensitivity to future consequences following damage to human prefrontal cortex. Cognition. 1994 Jan 1;50(1–3):7–15. [PubMed]
58. Bechara A, Tranel D, Damasio H, Damasio AR. Failure to respond autonomically to anticipated future outcomes following damage to prefrontal cortex. Cereb Cortex. 1996 Jan 1;6(2):215–225. [PubMed]
59. Austin MC, Kalivas PW. Dopaminergic involvement in locomotion elicited from the ventral pallidum/substantia innominata. Brain Res. 1991 Feb 22;542(1):123–131. [PubMed]
60. Berridge KC. Food reward: brain substrates of wanting and liking. Neurosci Biobehav Rev. 1996;20(1):1–25. [PubMed]
61. Gong W, Neill D, Justice JB. Conditioned place preference and locomotor activation produced by injection of psychostimulants into ventral pallidum. Brain Res. 1996 Jan 22;707(1):64–74. [PubMed]
62. Hiroi N, White NM. The ventral pallidum area is involved in the acquisition but not expression of the amphetamine conditioned place preference. Neurosci Lett. 1993 Jun 25;156(1–2):9–12. [PubMed]
63. Johnson PI, Stellar JR, Paul AD. Regional reward differences within the ventral pallidum are revealed by microinjections of a mu opiate receptor agonist. Neuropharmacology. 1993 Dec 1;32(12):1305–1314. [PubMed]
64. Napier TC. Dopamine receptors in the ventral pallidum regulate circling induced by opioids injected into the ventral pallidum. Neuropharmacology. 1992 Nov 1;31(11):1127–1136. [PubMed]
65. Tom SM, Fox CR, Trepel C, Poldrack RA. The neural basis of loss aversion in decision-making under risk. Science. 2007 Jan 26;315(5811):515–518. [PubMed]
66. Haber SN, Groenewegen HJ, Grove EA, Nauta WJ. Efferent connections of the ventral pallidum: evidence of a dual striato pallidofugal pathway. J Comp Neurol. 1985 May 15;235(3):322–335. [PubMed]
67. Kelley AE, Baldo BA, Pratt WE, Will MJ. Corticostriatal-hypothalamic circuitry and food motivation: integration of energy, action and reward. Physiol Behav. 2005 Dec 15;86(5):773–795. [PubMed]
68. Kabbaj M, Norton CS, Kollack-Walker S, Watson SJ, Robinson TE, Akil H. Social defeat alters the acquisition of cocaine self-administration in rats: role of individual differences in cocaine-taking behavior. Psychopharmacology (Berl) 2001 Dec 1;158(4):382–387. [PubMed]
69. Marinelli M, White FJ. Enhanced vulnerability to cocaine self-administration is associated with elevated impulse activity of midbrain dopamine neurons. J Neurosci. 2000 Dec 1;20(23):8876–8885. [PubMed]
70. Piazza PV, Deminiere JM, Le Moal M, Simon H, Anonymous Factors that predict individual vulnerability to amphetamine self-administration. Science. 1989 Sep 29;245(4925):1511–1513. [PubMed]
71. Piazza PV, Deroche-Gamonent V, Rouge-Pont F, Le Moal M. Vertical shifts in self-administration dose-response functions predict a drug-vulnerable phenotype predisposed to addiction. J Neurosci. 2000 Jun 1;20(11):4226–4232. [PubMed]
72. Dellu F, Piazza PV, Mayo W, Le Moal M, Simon H. Novelty-seeking in rats--biobehavioral characteristics and possible relationship with the sensation-seeking trait in man. Neuropsychobiology. 1996;34(3):136–145. [PubMed]
73. Deroche V, Piazza PV, Casolini P, Maccari S, Le Moal M, Simon H. Stress-induced sensitization to amphetamine and morphine psychomotor effects depend on stress-induced corticosterone secretion. Brain Res. 1992 Dec 11;598(1–2):343–348. [PubMed]
74. Deroche V, Piazza PV, Le Moal M, Simon H. Individual differences in the psychomotor effects of morphine are predicted by reactivity to novelty and influenced by corticosterone secretion. Brain Res. 1993 Oct 1;623(2):341–344. [PubMed]
75. Deroche V, Piazza PV, Le Moal M, Simon H. Social isolation-induced enhancement of the psychomotor effects of morphine depends on corticosterone secretion. Brain Res. 1994 Mar 21;640(1–2):136–139. [PubMed]
76. Rouge-Pont F, Marinelli M, Le Moal M, Simon H, Piazza PV. Stress-induced sensitization and glucocorticoids. II. Sensitization of the increase in extracellular dopamine induced by cocaine depends on stress-induced corticosterone secretion. J Neurosci. 1995 Nov 1;15(11):7189–7195. [PubMed]
77. Lucas LR, Angulo JA, Le Moal M, McEwen BS, Piazza PV. Neurochemical characterization of individual vulnerability to addictive drugs in rats. Eur J Neurosci. 1998 Oct 1;10(10):3153–3163. [PubMed]
78. Bry BH. Reducing the incidence of adolescent problems through preventive intervention: one- and five-year follow-up. Am J Community Psychol. 1982 Jun 1;10(3):265–276. [PubMed]
79. Bush G, Vogt BA, Holmes J, et al. Dorsal anterior cingulate cortex: a role in reward-based decision making. Proc Natl Acad Sci USA. 2002 Jan 8;99(1):523–528. [PubMed]
80. Oswald LM, Wong DF, Zhou Y, et al. Impulsivity and chronic stress are associated with amphetamine-induced striatal dopamine release. Neuroimage. 2007 May 15;36(1):153–166. [PubMed]
81. Laakso A, Wallius E, Kajander J, et al. Personality traits and striatal dopamine synthesis capacity in healthy subjects. Am J Psychiatry. 2003 May 1;160(5):904–910. [PubMed]
82. Unterwald EM, Horne-King J, Kreek MJ. Chronic cocaine alters brain mu opioid receptors. Brain Res. 1992;584(1–2):314–318. [PubMed]
83. Unterwald EM, Tempel A, Koob GF, Zukin RS. Characterization of opioid receptors in rat nucleus accumbens following mesolimbic dopaminergic lesions. Brain Res. 1989;505(1):111–118. [PubMed]
84. Unterwald EM, Rubenfeld JM, Kreek MJ. Repeated cocaine administration upregulates kappa and mu, but not delta, opioid receptors. Neuroreport. 1994 Aug 15;5(13):1613–1616. [PubMed]
85. Chen JF, Aloyo VJ, Weiss B. Continuous treatment with the D2 dopamine receptor agonist quinpirole decreases D2 dopamine receptors, D2 dopamine receptor messenger RNA and proenkephalin messenger RNA, and increases mu opioid receptors in mouse striatum. Neuroscience. 1993 Jun 1;54(3):669–680. [PubMed]
86. Steiner H, Gerfen CR. Role of dynorphin and enkephalin in the regulation of striatal output pathways and behavior. Exp Brain Res. 1998 Nov 1;123(1–2):60–76. [PubMed]
87. George SR, Kertesz M. Met-enkephalin concentrations in striatum respond reciprocally to alterations in dopamine neurotransmission. Peptides. 1987 May 1;8(3):487–492. [PubMed]
88. Zubieta JK, Gorelick DA, Stauffer R, Ravert HT, Dannals RF, Frost JJ. Increased mu opioid receptor binding detected by PET in cocaine-dependent men is associated with cocaine craving. Nat Med. 1996 Nov 1;2(11):1225–1229. [PubMed]
89. Kagan J, Reznick JS, Snidman N. Biological bases of childhood shyness. Science. 1988 Apr 8;240(4849):167–171. [PubMed]