1.7. Is There an Effect of Adult Testosterone on Attachment and Related Behaviours?
1.7.3. Other Externalizing Behaviours
Although physically aggressive and dominant behaviours as well as more unrestricted sexuality can be conceptualized as “risky behaviour”, and is in fact
conceptualized this way in much of the risk-taking literature, there do remain other types of behaviours that fall into the risk-taking category and which correlate with circulating T levels in adulthood. In men, higher basal T levels have been associated with engagement in risky health behaviours such as cigarette smoking (Booth, Johnson, & Granger, 1999; Knussmann & Christiansen, 1989; Fisher et al., 1997), alcohol use and abuse (La Grange et al., 1995; Mazur & Booth, 2014), and drug use (Booth, Johnson, & Granger, 1999; Mazur & Booth, 2014), as assessed by self-report. Furthermore, within a large-scale study of American Veterans (mean age = 37 years), Booth, Johnson, and Granger (1999) found that high T men (after controlling statistically for age), compared to low T men, had a greater probability of having sexually transmitted diseases and physical injuries. A recent investigation by van Anders and colleagues (2012) found that higher salivary T in young adult men predicted safer sex practices, including higher scores on a safer sex composite measure and safer sex resilience (i.e., more likely to follow through on condom use in the face of barriers to do so). These findings remained even after
controlling (via multiple regression) for variables that may affect associations between T and sexual risk-taking such as sexual frequency and relationship status. Although it would appear that these findings suggest that higher T protects against sexual risk-taking, the authors reason that engaging in safer sex practices may paradoxically represent a
bolder choice for men as it might be more socially risky for their relationships or their reputation, in general, to be sexually responsible.
A number of laboratory studies have found that higher levels of basal T are correlated with more risky financial decision-making in men (e.g., Apicella et al., 2008; Evans & Hampson, 2014; Stanton, Liening, & Schultheiss, 2011a; Reavis & Overman, 2001). Using a task involving certain outcomes and risky gambles, Stanton and
colleagues (2011b) suggested a U-shaped relationship between T and economic risk preferences. Finally, Goudriaan et al. (2010) found that young men who underwent an experimental manipulation that produced high-normal levels of T and low-normal levels of estradiol (the major type of estrogen found in humans) showed a significant increase in risk-taking, compared to baseline. No such increase was found for men randomized to have low-normal T and high-normal estradiol levels.
Although not conducted in an adult sample, a study by Booth et al. (2003) is worth noting. The authors assessed the associations between T, risk-taking, symptoms of depression, and parent-child relationship quality in a sample of boys and girls aged 6-18 years. Results showed that endogenous T levels were independently related to the quality of relationships that sons formed with their mothers, as rated by child reports of intimacy with the parent and parents’ reports of their acceptance of their child, among other factors. In boys only, higher T was related to poorer relationship quality with mothers. In addition, poor parent-son relationship quality moderated the association between T and risk-taking, as assessed by youths’ self-reports of the frequency of engagement in 18 risky activities over the past year (e.g., “skip a day of school”; “stay out all night without parents’ permission). Given the nature of these questions, however, children between the ages of 6-9 years did not complete the risk-taking measure. T-related risk-taking
behaviour was more evident as parent-son relationship quality decreased and was less evident as parent-son relationship quality increased. T was positively correlated with risk-taking only when parent-son relationship quality was rated to be low. This was true regardless of stage of development. These findings support the idea that individual differences in the quality of the relationship a son forms with his parent, which may be
indicative of the nature of the attachment relationship, is associated with T, and that the relationship may have implications for adjustment problems, such as risk-taking.
The literature on adult T and impulsivity has generally considered studies investigating the relationship between T and overt behaviours (e.g., criminality and aggression, sensation seeking) as evidence of a T-impulsivity association. Among the few studies that investigated impulsivity specifically, or have employed measures of trait impulsiveness, the findings are mixed. O’Connor et al. (2002) found that hypogonadal young men reported significantly lower levels of impulsiveness (i.e., a lacking of “futuring” or forethought), as assessed by the Barratt Impulsivity Scale-11 (BIS-11), compared to healthy young men. No significant differences were found for the cognitive or motor components of impulsivity. Similarly, Schmidt and colleagues (2004) found that healthy young men at baseline had significantly higher levels of sensory
impulsiveness (but not motor or interpersonal impulsivity), assessed using an earlier version of the BIS, compared to their scores following pharmacologically-induced hypogonadism. Two studies employing a behavioural measure of impulsivity, the delay- discounting task, have, however, failed to produce consistent associations between T concentrations and impulsivity in young adult men (Ortner et al., 2013; Takahashi et al., 2006). Some work in non-human animals has demonstrated that gonadectomy in adult male rats enhances behavioural inhibition to footshock, indicating improved impulse control (Svensson, Söderpalm, & Engel, 2000), while testosterone treatment induces behavioural disinhibition, reflecting impulsive-like behaviour (Svensson et al., 2003), lending support to an activational effect of T on impulsiveness.
1.7.4. Internalizing Behaviours
Higher levels of T appear to be associated with higher levels of externalizing behaviours, and conversely may also be associated with lower levels of internalizing symptoms. Studies of young men who are hypogonadal have found that self-reported levels of depressive symptoms are significantly higher among the hypogonadal group compared to healthy controls, and that exogenous T administration over a set treatment period results in a significant reduction in depressive symptoms from baseline (Aydogan
et al., 2012; O’Connor et al., 2002). Similarly, Schmidt and colleagues (2004) found that pharmacologically-induced hypogonadism (low T levels) precipitated depressive
symptoms in a subset (~10%) of healthy young men with no previous history of psychiatric illness, supporting the possibility that reduced T may predict the onset of depressive symptoms. Studies of non-human animals in which T levels have been manipulated in adulthood provide further support for a causal relationship (e.g., Bernardi et al., 1989; Buddenberg et al., 2009; Carrier & Kabbaj, 2012). Assessment of basal T in a sample of undergraduate men also found that T was negatively related to depression, and specifically was associated with self-reported sleep symptoms of depression (Sankar & Hampson, 2012; but see van Honk et al., 1999 who found no significant association between T and depression in healthy young men). In contrast, administration of
exogenous T to men without a psychiatric history does not appear to have any effect on self-reported levels of affective symptoms (O’Connor et al., 2004).
Higher T concentrations have also been suggested to have an anxiolytic effect. In young men, a hypogonadal state has been associated with higher levels of anxiety in several studies (e.g., Ayodogan et al., 2012; O’Connor et al., 2002; cf., Schmidt et al., 2004). Studies measuring basal T in healthy males have similarly pointed to a negative relationship between T and self-reported anxiety symptoms (e.g., Berglund et al., 2011; Granger et al., 2003; cf., Maner et al., 2008; van Honk et al., 1999). Again, experimental T-manipulation studies conducted in non-human animals support the possibility of a causal effect of T on anxiety symptom reduction, although the mechanism behind this association is not yet known (e.g., Aikey et al., 2002; Carrier & Kabbaj, 2012; for further references see review by McHenry et al., 2014).
Among the limited work available on the topic of androgenic effects on self- esteem, the data point towards a positive relationship between levels of self-esteem and T concentrations in adulthood. Johnson and colleagues (2007) demonstrated that T levels were a positive predictor of scores on a global self-esteem scale in a sample of
undergraduate men. Vermeersch et al. (2010) found similar results, albeit in an adolescent male sample (mean age = 14.4 years). In a placebo-controlled study, a positive effect of exogenous T treatment on self-esteem was found in a group of
hypogonadal young men (O’Connor et al., 2002). Significant increases in men’s levels of self-esteem were seen between the second and third assessment timepoints in the study (i.e., week 4 and week 8). Consequently, it cannot be ruled out that the observed changes in self-esteem may have been due to associated physical changes secondary to T
administration, including concomitant restoration of sexual function. In another study, pharmacologically-induced hypogonadism in healthy men for a period of two months failed to significantly reduce self-esteem scores compared to baseline (Schmidt et al., 2004). Because these men were in a hypogonadal state only transiently, it is possible that the physiological and sexual changes associated with acute hypogonadism may not have been substantive enough to cause negative self-interpretations in participants.