Chemosensory Perception

, Volume 5, Issue 3, pp 300–307

Olfactory Abilities and Psychopathy: Higher Psychopathy Scores Are Associated with Poorer Odor Discrimination and Identification


    • Department of PsychologyMacquarie University
  • Richard J. Stevenson
    • Department of PsychologyMacquarie University

DOI: 10.1007/s12078-012-9135-7

Cite this article as:
Mahmut, M.K. & Stevenson, R.J. Chem. Percept. (2012) 5: 300. doi:10.1007/s12078-012-9135-7


Olfactory processing is known to involve the orbitofrontal cortex (OFC). The OFC is also believed to function less effectively in individuals scoring higher in psychopathic personality traits. In this study, we examined whether poorer olfactory discrimination and identification—taken as an indicator of OFC integrity—was associated with the degree of presence of psychopathic traits in a community sample. Seventy-nine non-criminal participants completed the Self-Report Psychopathy scale and a standardized measure of olfactory ability, the Sniffin’ Sticks, measuring odor threshold, identification, and discrimination. Consistent with predictions, we found a relationship between psychopathy and olfactory discrimination and identification but not odor threshold, even after controlling for gender, age, empathy, smoking status, and craniofacial surgery/injury. These findings suggest that brain areas subserving higher olfactory processes—identification and discrimination—are somehow less efficient in individuals who score higher on psychopathic traits. In particular, we suggest that this relates to processing within the orbitofrontal cortex.


PsychopathyOrbitofrontal cortexOlfactionNon-criminalSniffin’ Sticks


Psychopathy, as measured by the Psychopathy Checklist Revised (PCL-R; Hare 1991, 2003), extends from a severe personality disorder characterized by callousness, manipulation, sensation-seeking, and antisocial behaviors to the presence of these same traits in otherwise healthy and functional people. Current neurobiological theories of psychopathy aim to provide explanations for these behavioral manifestations. For example, the poverty of emotional experiences evidenced by psychopaths are likely due to deficits associated with the limbic structures, particularly the amygdalae (Blair et al. 2002; Blair 2004, 2006; Birbaumer et al. 2005) and also the anterior and posterior cingulate (Raine 2008). In terms of psychopaths’ failure to inhibit previously punished behaviors, converging research findings indicate that orbitofrontal cortex-associated deficits are responsible (Mitchell et al. 2002, 2006; Blair et al. 2006). The current study was motivated by the observation that the orbitofrontal region also serves as a secondary olfactory cortex (Rolls 2000). We reasoned that if participants scoring highly on psychopathic traits do so in part because the orbitofrontal cortex (OFC) is compromised in some way, then we might also expect to see poorer performance on olfactory tasks that are also subserved by this region. The remainder of this Introduction provides support for these two contentions—OFC dysfunction is associated with psychopathy, in both criminal and non-criminal samples, and olfactory processing draws upon the OFC.

The frontal lobes are largely responsible for what is loosely referred to as “executive functioning,” that is planning, impulse control, and acting in accord with social norms (Lezak 1995). Early research suggested that psychopaths were characterized by poor behavioral controls due to their inability to inhibit previously punished (by electric shock) responses (Lykken 1957) and poor performance on the Porteus Maze task (Schalling and Rosen 1968). Using DSM-III Antisocial Personality Disorder criteria and the socialization scale from the California Psychological Inventory as psychopathy measures, Gorenstein (1982) employed an extensive neuropsychological test battery and also concluded that psychopathy was associated with executive dysfunction. However, subsequent studies using various psychopathy measures (Devonshire et al. 1988; Hoffman et al. 1987; Sutker and Allain 1987), including the PCL (Hare 1984) and PCL-R (Hart et al. 1990; Smith et al. 1992), found no evidence that psychopathy was characterized by generalized frontal lobe dysfunction.

Rather than a generalized frontal lobe dysfunction, evidence is starting to suggest a more specific frontal deficit, in particular one involving prefrontal areas such as the OFC. Several studies indicate that OFC damage is associated with the perseveration of behaviors that are no longer associated with reward (Bechara et al. 1994, 1999; Dias et al. 1996; Rolls 2000, 2004; Berlin et al. 2004). Two reviews (Blair 2004, 2006) and the findings from many studies with criminal psychopaths show that they exhibit OFC-associated deficits (LaPierre et al. 1995; Mitchell et al. 2002, 2006; Blair et al. 2006). Using the Iowa Gambling Task, a measure that patients with OFC lesions perform significantly worse compared to controls (Bechara et al. 1994, 1999), found evidence consistent with OFC impairment in psychopaths, but not in controls (Mitchell et al. 2002). A subsequent study (Blair et al. 2006) employed neuropsychological measures that separately tapped OFC functioning (using an object alternation task), the anterior cingulated cortex functioning (using Stroop-like tasks), and the dorsolateral prefrontal cortex functioning (using a spatial attention task). The results of this study showed that PCL-R-defined psychopaths only showed a significant performance deficit on the task sensitive to OFC dysfunction compared to non-psychopaths. Finally, magnetic resonance imaging (MRI) has also revealed that PCL-defined psychopathy is associated with a reduction in the functional connectivity between the OFC and amygdalae and also between the OFC and medial parietal cortex (Motzkin et al. 2011). Given the interconnectedness of the OFC and limbic system, this finding underscores the importance of these neural pathways in the etiology of psychopathy and provides the broader context in which OFC-associated deficits may occur.

Volatile chemicals bind to and depolarize olfactory receptor neurons (ORNs), which are located in the nasal vault. The ORNs send information via the lateral olfactory tract to the bulb and then to primary olfactory cortex (the piriform cortex). From here, information is streamed to the secondary olfactory cortex—the OFC—both directly and indirectly via the mediodorsal nucleus of the thalamus (Zald and Pardo 2000; Hawkes and Doty 2009). This basic anatomical map of olfactory information flow (excluding in course: the amygdalae, hippocampus and other structures) has been built up from primate anatomy and neurophysiological studies and from human neuropsychological and imaging studies (e.g., Rolls et al. 1996; Öngür and Price 2000; Gottfried et al. 2002). Although there is general agreement over the brain structures involved in olfaction, the precise function of these various components is less well understood. For the OFC, damage to this area is associated with impaired discrimination but not with impairments in odor detection (e.g., Hulshoff Pol et al. 2002). As discrimination is a necessary precursor to identification, this too should be compromised by lesions in this location. Neuroimaging data indicate activity in the OFC during passive smelling, as well as for identity, discriminative, familiarity, hedonic, and edibility judgments (Royet et al. 2001; Gottfried et al. 2002; Gottfried 2007). It seems likely then that any complex olfactory task will generate activity within the OFC.

As described above, there is certainly evidence to suggest that OFC function is compromised in psychopathic individuals. If correct, then it would be reasonable to expect poorer olfactory function in these individuals on the basis that OFC areas involved in reward-related judgments overlap with OFC areas involved in olfaction. While dysfunctional olfaction has been established in various neurological disorders such as schizophrenia, Parkinson’s and Alzheimer’s disease (Hawkes and Doty 2009), only two studies have investigated the olfactory abilities of psychopaths; however, their findings fail to clarify our understanding of OFC functioning in psychopathy. The first study (LaPierre et al. 1995) to investigate this link did so in criminal psychopaths and used an abbreviated version (12 odors) of the University of Pennsylvania Smell Identification Test (UPSIT; Doty et al. 1984), among other neuropsychological tests of the prefrontal cortex. Compared to incarcerated non-psychopaths, incarcerated psychopaths correctly identified significantly fewer odors on the odor identification task, made significantly more commission errors on the go/no-go test, and made significantly more qualitative errors on the Porteus Maze Test (LaPierre et al. 1995), all tests that tap OFC-associated functioning. In contrast, there were no performance differences between the psychopaths and non-psychopaths on tasks (e.g., WCST, odor identification) tapping the neural substrates not associated with psychopathy, such as the dorsolateral frontal cortex.

The second study to investigate the relationship between olfactory functioning and psychopathy, again using an abbreviated version of the UPSIT, utilized a juvenile sample of incarcerated psychopaths (Roussy and Toupin 2000). Psychopaths in this study did not perform worse on the odor identification task compared to non-psychopaths. In an attempt to explain these different findings—apart of course from the different ages of the two samples—Roussy and Toupin (2000) suggested that the brevity of the odor threshold test used by LaPierre et al. (1995; six trials compared to their 20) may have resulted in the inclusion of participants with poor olfactory acuity. This argument is plausible and so we cannot be sure whether poor acuity might account for the findings of LaPierre et al. (1995).

Data collected from convicted or incarcerated psychopaths have several interpretive difficulties. Many of these individuals may have abused drugs, smoked, sustained head injuries, and have engaged in lifestyles that risk infectious disease; all of which may compromise brain function and/or the integrity of the olfactory system. One way to surmount these type of confounds is to study psychopathy in the community. Unfortunately, relatively few studies of OFC function (imaging/neuropsychological) have been conducted with such non-incarcerated, PCL-R-defined psychopaths, and no olfactory studies have as yet been published. Evidence for OFC dysfunction in community-dwelling psychopaths is mixed but generally supportive. Performance on the Iowa Gambling Task, a measure sensitive to OFC dysfunction, has provided some favorable findings, with riskier decisions and poorer capacity to learn from errors in individuals scoring higher in psychopathic traits (Belmore and Quinsey 1994; Mahmut et al. 2008; van Honk et al. 2002). Similar findings have also been observed on other neuropsychological tests sensitive to OFC function (Dinn and Harris 2000). Two out of the three structural MRI studies by Raine and colleagues with community-dwelling (PCL-R-defined) psychopaths have found evidence of prefrontal cortex (i.e., including the OFC) dysfunction within this population (Raine et al. 2000; Yang et al. 2005). These findings then would lead one to suggest that olfactory deficits should be present in individuals who score higher in psychopathic traits. Specifically, we hypothesized that such individuals should demonstrate poorer odor discrimination ability, poorer odor identification ability, but with an intact capacity to detect odors. As females have demonstrated superior olfactory abilities compared to males (e.g., Stevenson and Mahmut 2011) and higher degrees of empathy are associated with increased olfactory abilities (e.g., Spinella 2002), we predicted that the relationship between psychopathy and olfactory performance would persist after controlling for these other variables.

Materials and Methods


A total of 79 participants (40 females, mean age 19 years; 39 males, mean age 21 years) completed the study for course credit. The ethical aspects of this study were approved by Macquarie University’s Ethics Committee and informed consent was obtained from each participant. One male participant was excluded from all analyses due to having a criminal conviction, and two participants (one male, one female) were excluded due to an inconsistent pattern of performance on the odor threshold task. Prior to the commencement of the study, a brief, standardized interview was conducted to enquire about (for example) previous craniofacial surgery/injury, current/previous smoking status, and current health. All participants were deemed eligible to complete the current study.


Sniffin’ Sticks

The Sniffin’ Sticks (Hummel et al. 1997) comprised three tests each designed to measure a distinct olfactory ability: odor threshold, odor identification, and odor discrimination. Each odor is contained within an individual pen and each of the three tests has a unique set of pens. The experimenter presented all pens (i.e., odors) bilaterally approximately 2 cm beneath the nose. To reduce the chance of adaptation and fatigue, one of the personality measures was completed between each of the three Sniffin’ Sticks’ tests. Each olfactory test is described in more detail below.

Olfactory Threshold Test

The olfactory threshold test was used to measure olfactory acuity and consists of 16 levels that differ in concentration of n-butanol (1 = strongest, 16 = weakest). Each level consists of a triplet of pens (one target containing n-butanol and two odorless distracter pens) presented in a randomized order. Before the test begins, the participants are familiarized with the n-butanol odor. On each trial, the participants are asked to select the target pen from the distracter pens while blindfolded. There was an interval of 5 s between the presentation of each odor within a trial and an interval of 30 s between trials. A staircase procedure was used to present the odors to the participants with progressively higher concentrations of n-butanol. This was continued until the target odor pen was correctly identified, which then triggered a reversal of the staircase downwards (i.e., to lower concentrations). Incorrect guesses then also triggered a reversal of the staircase upwards, until the target was correctly selected. The threshold score was the mean of the last three of five reversals in the staircase.

Odor Identification Test

The identification test measured a participant’s ability to identify common odors (e.g., orange, coffee, and leather) and consisted of 16 individual odor pens presented in the same order. The participants are asked to select the correct odor name from a list of four (i.e., three names are distracters) and each trial is separated by a 30-s interval. The odor identification score was calculated by summing the correct responses with the highest possible score being 16.

Odor Discrimination Test

The discrimination test measured a participant’s ability to discriminate between odors. The test consists of 16 trials, each with a triplet of pens: two containing the same odor, and a third containing a different odor. There was an interval of 5 s between the presentation of each odor within a trial and an interval of 30 s between trials. The odor discrimination score was calculated by summing all correct response trials, with the highest possible score being 16.

Self-Report Psychopathy Scale IV

The Self-Report Psychopathy scale IV (SRP-IV; Paulhus et al. 2012) is a self-report analogue of the PCL-R with the same four-factor structure as the PCL-R (Williams et al. 2007; Mahmut et al. 2011; Neal and Sellbom 2012). The SRP-IV consists of 64 items and loads onto one of the four facets: interpersonal manipulation (IPM), callous affect (CA), erratic lifestyle (ELS), and criminal tendencies (CT). Response options on a five-point category scale (1 = disagree strongly, 5 = agree strongly) and sample items from each facet include “I purposely flatter people to get them on my side” (IPM), “People sometimes say that I’m cold-hearted” (CA), “I enjoy doing wild things” (ELS), and “I have broken into a building or vehicle in order to steal something or vandalize” (CT). Six different scores were calculated for the SRP-IV: a total SRP score which was the sum of all items after reverse scoring (SRP Total), a total SRP score without items from the CT facet (SRP-no-CT), and four total scores for each of the individual facets (CA, IPM, ELS, and CT) calculated by summing items for each respective facet. Theoretically, SRP Total scores can range from 64 to 320, SRP-no-CT scores from 48 to 240, and each facet’s scores can range from 16 to 80. Higher scores on the SRP-IV indicate higher degrees of trait psychopathy. See Fig. 1 for the distribution of SRP Total scores by gender.
Fig. 1

Histogram of SRP Total scores by gender

Emotional Empathy Questionnaire

The Emotional Empathy Questionnaire (EEQ; Mehrabian and Epstein 1972) is a 33-item self-report measure of emotional empathy with responses answered on a 9-point category scale (1 = disagree very strongly; 9 = agree very strongly). Two sample items from the EEQ are “It upsets me to see helpless old people” and “It is hard for me to see how some things upset people so much.” The total EEQ scores are calculated by summing the responses on all items (after reverse scoring) and can theoretically range from 33 to 297. Higher EEQ scores indicate higher levels of emotional empathy.


Relationship Between Measures

The initial analyses involved determining the association between our various measures by examining the Pearson correlation coefficient between each continuous variable (see Table 1) and examining the association between gender and olfactory performance. All correlations between odor discrimination and the psychopathy total and facets scores were significant, the strongest was with SRP Total followed by the SRP total score without the CT facet (see Table 1). There was no significant difference in mean odor discrimination scores between males (11.8) and females (12.5; t(77) = 1.40, p = .17). Similarly, all correlations between odor identification and the psychopathy total and facets scores were significant; except the IPM and CT facets (these being in the expected direction); the strongest was with callous affect facet (see Table 1). As expected, based upon previous findings, females evidenced significantly higher mean odor identification scores than males (12.0 vs. 11.1, respectively; t(77) = 2.47, p = .016). Finally, none of the correlations between odor threshold score and psychopathy scores were significant; however, females evidenced significantly higher mean threshold scores than males (9.8 vs. 8.4, respectively; t(77) = 2.64, p = .010). Based on normative data (Hummel et al. 2007), males and females scored within the normal range on all olfactory measures of the Sniffin’ Sticks.
Table 1

Correlation matrix with psychopathy, EEQ, and Sniffin’ Sticks






SRP Total










































SRP-no-CT SRP total score without the criminal tendencies facet, IPM interpersonal manipulation facet, CA callous affect facet, ELS erratic lifestyle facet, CT criminal tendencies facet, EEQ emotional empathy questionnaire

**p < .001; *p < .05, based one-tailed distributions (n = 79)

Regression Analyses

Two measures from the Sniffin’ Sticks, odor discrimination, and odor identification formed the two dependent variables (DV) for this study, and these were analyzed separately using the following procedure. In order to determine the best predictor of each olfactory DV, regression with backward elimination was used, with odor threshold score, gender, previous craniofacial surgery/injury, current smoking status, age, and EEQ scores included as predictors alongside the main independent variable (IV), psychopathy (SRP Total score). Note that the same regression results were obtained with the SRP Total score without the criminal tendencies facet (SRP-no-CT) as the main predictor.

For odor discrimination, following regression with backwards elimination, the final model was significant [F(2, 78) = 5.69, p = .005] and explained 10.7 % of the variance (Radj.2). This model contained two predictors, with both making significant independent contributions to explaining variance in odor discrimination score. These predictor variables were odor threshold (β = .25; sr2 = 6.1 %) and SRP Total (β = −.23; sr2 = 5.2 %). Figure 2 illustrates the partial regression scatter plot of odor discrimination score with SRP Total.
Fig. 2

Partial regression scatter plot (with fitted linear slope) with standardized SRP Total and standardized odor discrimination score

To determine whether the four facets (i.e., CA, IPM, ELS, and CT) differed in their ability to predict odor discrimination score, we conducted four hierarchical regression analyses. Each analysis involved two steps. First, and in each one, odor threshold was added alongside three of the four facet scores. Then, in the second step, the final facet score was added, and Finc was determined to see if there had been a significant increase in variance accounted for by the addition of this extra facet. In no case did the addition of the extra facet significantly increase variance accounted for, indicating that no single facet—as also suggested by the correlations in Table 1—was uniquely driving the association between SRP total and odor discrimination score.

For odor identification, a backward regression analysis revealed that the final model was significant [F(2, 78) = 7.00, p = .001] and explained 13.3 % of the variance (Radj.2). This model contained two predictors: threshold, which explained a significant proportion of variance in odor identification score (β = .31; sr2 = 9.6 %), and SRP Total, which approached significance (β = −.20; sr2 = 3.9 %, p = .064). Figure 3 is a partial regression scatter plot of identification score with the main IV, SRP Total.
Fig. 3

Partial regression scatter plot (with fitted linear slope) with standardized SRP Total and standardized odor identification score

Once again, to determine whether the four facets (i.e., CA, IPM, ELS, and CT) differed in their ability to predict odor identification score, four hierarchical regression analyses were conducted. Each analysis initially involved the inclusion of the odor threshold score as a predictor along with three of the four facet scores. Next, the final facet score was added to determine if it resulted in a significant increase in variance accounted for in odor identification. Only the addition of the CA facet resulted in an increase in variance explained that bordered on significance, F(1, 73) = 3.02, p = .086 (β = −.23; sr2 = 3.5 %).


The aim of this study was to determine the relationship between psychopathic traits and olfactory ability in a non-criminal sample. As hypothesized, psychopathy was a significant predictor of performance on the odor discrimination task of the Sniffin’ Sticks and approached significance on the odor identification task but was not related to odor threshold. The former relationships held even after controlling for relationships between olfactory performance, gender, and empathy. The current study’s findings are unique in demonstrating a relationship between psychopathic traits and olfactory performance in a non-criminal sample. In addition, they are consistent with a previous study that utilized a criminal (adult) sample (LaPierre et al. 1995), but not with another employing a juvenile criminal sample (Roussy and Toupin 2000). More generally, these findings suggest that the presence of high levels of psychopathic traits is associated in some manner with a less functional OFC, which is then reflected in somewhat poorer olfactory performance.

Our findings, along with those from the study of LaPierre et al. (1995) are at odds with the failure of the Roussy and Toupin (2000) study to find a psychopathy–odor identification association. There are several possible reasons why this different outcome was obtained. For example, while both LaPierre et al. (1995) and Roussy and Toupin (2000) employed the same odor identification (from the UPSIT) and psychopathy measures, the former recruited an adult criminal sample whereas the latter a juvenile criminal sample. Thus the difference in results may in part be attributable to age-related changes. A further possibility concerns the non-psychopaths in each study having significantly different odor identification means (11.5 and 10.0, respectively), while the psychopaths have very similar means (10.5 and 10.0, respectively). Thus, the inconsistent findings by the LaPierre et al. (1995) and Roussy and Toupin (2000) studies could be due to differences in the choice of controls. Indeed, as we noted in the “Introduction,” the choice of controls in such studies is problematic as criminal psychopaths are likely to engage in violent behavior (Hare 2003; Harris et al. 1991) and drug abuse (Taylor and Lang 2006). Chronic drug and cigarette use (Sugiyama et al. 2006) and facial/cranial injuries (Jimenez et al. 1997) are known causes of impoverished olfactory abilities and are more likely to occur in a criminal than non-criminal sample. To find the predicted relationship in populations, demonstrating psychopathic traits but without these risk factors suggests that these olfactory–psychopathy relationships probably also occur in criminal samples, even though they may be masked by confounding factors.

The current study’s finding that higher degrees of psychopathy are associated with poorer odor discrimination and odor identification ability provides further support for the premise that OFC-associated deficits may be a characteristic of the non-criminal psychopath (e.g., Mahmut et al. 2008) as they are for the criminal psychopath (Mitchell et al. 2002; Blair et al. 2006; Mitchell et al. 2006). However, no published studies have employed olfactory measures to investigate the relationship between non-criminal psychopathy and OFC-associated functioning (i.e., neuropsychological measures that purportedly tap the intactness of processing in this neural substrate), so the current findings are preliminary and come with caveats that are detailed next.

One limitation of the current study is the relatively small percentage of variance in odor discrimination (~5 %) and odor identification (~4 %) scores explained by psychopathy. Threshold score was also a signification predictor of odor discrimination (~6 %) and identification scores (~10 %), but a large proportion of variance remained unexplained. While other known correlates of olfactory abilities such as empathy (Spinella 2002) as measured by the EEQ and gender (Stevenson and Mahmut 2011) were included in the regression analyses, these predictors were not significant and were not present in the final models. However, the findings suggest the callous affect facet may be a better predictor of olfactory performance, particularly odor identification, as it correlated more strongly with odor identification (r = .27) than did SRP Total score (r = .24). Moreover, compared to the SRP Total score, the CA facet’s beta-weight was slightly larger (β = −.23 vs. β = −.20) and explained a similar proportion of variation in odor identification (~4 %). Finally, the CA facet had the strongest (inverse) correlation with the EEQ (r = −.70) compared to all other psychopathy scores, including SRP Total score.

A callous lack of empathy has been clearly linked to amygdale-associated dysfunction (Raine et al. 2004; Blair 2006) so the inclusion of variables that measure amygdalae function in future research may reveal other correlates of odor identification and discrimination scores. Evidence for this assertion comes from recent fMRI findings which indicate that degraded functional connectivity between OFC and amygdalae pathways are present in criminal psychopaths (Motzkin et al. 2011). Flattened psychophysiological arousal (e.g., skin conductance response, startle blink) is evidence of an amygdalae dysfunction and a characteristic of the psychopath (Lorber (2004) for review), so the use of psychophysiological measures in future studies may account for some of the unexplained variance in olfactory performance. Moreover, the inclusion of an intelligence measure in future research is important as it is a possible correlate of olfactory ability (cf. Danthiir et al. 2001). The current study did not control for the possible influence of intelligence on olfactory performance and may limit the generalizability of the findings. Finally, we did not screen the participants for any previous acute nasal infections which can compromise olfactory performance and so this should be considered a caveat to our findings.

The convergence of research findings with both criminal and non-criminal samples suggests that psychopathy is a significant predictor of olfactory ability. We conclude, tentatively, that the poorer olfactory performance observed in individuals who score higher on psychopathic traits may manifest through some subtle impairment of the OFC. Finally, we suggest that olfactory measures represent a potentially interesting marker of psychopathic traits because performance expectancies are unclear and may therefore be less susceptible attempts to “faking good/bad”.

Copyright information

© Springer Science + Business Media, LLC 2012