Gonadal steroids and visuo-spatial abilities in adult males: Implications for generalized age-related cognitive decline

Abstract

The relationship between the gonadal steroids, testosterone and estrogen, and individual and group differences in performance on some cognitive tasks remains unclear but sex differences favoring males on some tests of visuo-spatial ability are large and robust. This aim of this review is to assess evidence for both organizational and activational effects of gonadal steroids as the principle cause of sex difference in visuo-spatial ability. Additionally, the implications of this relationship are discussed in the context of decreasing levels of gonadal steroids in aging males and psychological theories of generalized age-related cognitive decline. Based upon human and non-human research gonadal steroids have organizational effects on visuo-spatial ability in adulthood. Activational effects of gonadal steroids on visuo-spatial ability appear most dominant in older men and are necessary for maintaining optimal visuo-spatial ability; randomized clinical trials show that testosterone supplementation improves performance. Additionally, decreasing gonadal steroid levels in aging males may contribute to generalized age-related cognitive decline. Future supplementation studies in men should attempt to control for constituent abilities related to visuo-spatial task performance, and investigate interactions between dosage levels and baseline gonadal status. Further future animal research is required to investigate changes in gonadal steroid levels and their relationship to neurotransmitter systems, neural plasticity, and behavioral correlates.

Keywords:

• Sex hormones
• gonadal steroids
• visuo-spatial ability
• aging
• cognitive decline
• males
• mental rotation
• testosterone
• review

Introduction

Two major domains of cognitive function exhibiting sex differences are verbal and visuo-spatial abilities. Females have consistently been found to outperform males on tests of verbal ability, especially on tests of verbal fluency, vocabulary, and verbal memory, although the effect size for these differences is small. Females also generally outperform males on tests of perceptual speed and motor tasks which require rapid matching of stimuli, or rapid fine motor movements, or both [1]. Perceptual speed tasks (also known as clerical speed tasks) are simple tasks which are not dependent upon knowledge or other cognitive abilities and are assessed by the number of items correctly completed within a specified time limit [2], for example, Digit Symbol from the Wechsler scales [3]. The female advantage on this test has been attributed to superior fine motor ability [4], although this has not been unequivocally established.

Males have been found to outperform females on tasks assessing visuo-spatial abilities [1],[5],[6]. Visuo-spatial abilities have been broadly defined as ‘an individual's abilities in searching the visual field, apprehending the forms, shapes and positions of objects as visually perceived, forming mental representations of those forms, shapes and positions, and manipulating such representations mentally’[7]. Some researchers have reduced the domain to a small number of qualitatively different abilities, for example, tasks which require spatial visualization or spatial perception [1],[5],[7]. Nonetheless, these classifications remain very general and fail to capture the specific aspects of performance which may underlie sex differences in performance. In an analysis of 286 studies, Voyer and colleagues adopted a test-by-test approach and derived an overall mean weighted sex difference in visuo-spatial ability favoring males [8]. The largest effect sizes were reported for the Mental Rotation test (MRT) [9], the Rod-and-Frame test [10], and Spatial Relations from the Primary Mental Abilities (PMA) battery [11]. Accordingly, the MRT showed a sex difference of about one standard deviation, favoring males [5],[8].

Recent studies using virtual environments (VEs), tasks thought to tap ‘real world’ navigational skills, have also demonstrated moderate to large sex differences favoring males [12-20]. Other studies, however, have not found sex differences [21-23]. These discrepant results may be due to a number of possible mediating factors related to VE performance, including: type of VE task employed (e.g., virtual maze or simulated ‘real world’ environment); task performance measures; or individual differences in computer interface proficiency [18], strategy utilization, and prior experience with VEs (e.g., computer game experience). Notwithstanding, one VE task in particular, the Virtual Morris Water Task (VMWT), has demonstrated reliable and robust sex differences favoring males [12-15],[20]. These differences are highly comparable to those observed using naïve rodents with the Morris Water Task (MWT) [24-28], thus providing a basis for cross-species comparisons.

Performance measures on the VMWT are correlated, although only to a limited extent with performance on the Vandenberg and Kuse MRT [14],[15]. The magnitude of these correlations suggests that although the tasks share some common cognitive processes, the two types of visuo-spatial tasks are distinct and hence recruit different skills. Nonetheless, it is the cognitive processes tapped similarly by both these tasks, which presumably form the loci of human sex differences in visuo-spatial abilities. Accordingly data from both these tasks is considered in the following discussion.

This review has two aims: Firstly, to assess evidence for a dominant role of gonadal steroids, principally testosterone (T), in the male advantage in visuo-spatial and navigational behavior; Secondly, to discuss the implications of age-related declines in gonadal steroid levels in males in relation to generalized age-related cognitive decline. Discussion is limited to adult males. Comparable non-human animal research is discussed where evidence from human studies is lacking. The first section focuses on evidence for organizational and the activational effects of both endogenous and exogenous sex hormones on visuo-spatial ability in adults. The second section discusses age-related declines in levels of endogenous gonadal hormones in the context of both psychological theories and physiological changes characterized by generalized age-related cognitive decline. The final section points to areas requiring more research.

Sex hormones and visuo-spatial ability

Effects of gonadal steroids during early development are termed ‘organizational’ because they are thought to involve the permanent alteration of patterns of neural organization. Later effects, which are termed ‘activational’ because they involve circulating hormonal levels (e.g., during puberty, or due to diurnal rhythms, or exogenous hormone administration), modify existing steroid-responsive neural circuits [29]. Both organizational and activational effects of gonadal steroids, namely T and estrogen, are thought to contribute to within-sex variability and to between-sex differences in sexually dimorphic behavior [30].

Organizational effects of gonadal steroids

There are three main lines of evidence for an organizational effect of pre- and post-natal gonadal steroid levels on visuo-spatial ability: Animal studies, congenital disorders involving abnormal hormone exposure, and studies in transsexuals.

T exerts its effects directly and also through its metabolites dihydrotestosterone (DHT) and estradiol (E2) [31]. In rats, nuclear receptors for gonadal steroids are located in the hypothalamus, hippocampus, and in parts of the cerebral cortex, regions that support the differentiation of sexual behavior, working memory, and reference memory, respectively [32]. These cortical regions have been found to be structurally different in males and females, differences known to be determined by early exposure to gonadal steriods [33]. Female rodents treated with androgens in the last week of gestation perform like normal males in the MWT; moreover, their hippocampal CA1 and CA3 pyramidal cell field volumes are male-like. Similarly, male rodents treated with flutamide prenatally, an androgen receptor antagonist, and castrated upon delivery perform like females in the MWT and have female-like CA1 and CA3 pyramidal cell field volumes [34]. Postnatal organizational effects of gonadal steroids on rodent's spatial behavior have also been reported. For example, adult male rats castrated at birth showed navigation performance more like normal female rats than uncastrated rats on a radial arm maze test [32]. Conversely, neonatal female rats treated with T for 90 days outperform both female and male controls on the MWT [28]. Together these results implicate both the pre- and post-natal periods as critical for the organizational effects of gonadal steroids on adult spatial behavior and hippocampal morphology in rodents.

The extent to which these observations can be generalized to humans is still unclear, although two lines of research have shed some light on the organizational effects of sex hormones in the human central nervous system and on their effects on cognition. Male humans have elevated levels of T during gestational weeks 8–24 and post-natally until about five months of age [33]. Two clinical disorders resulting in abnormal gonadal hormone exposure during these critical periods are androgen insensitivity syndrome (AI) and congenital adrenal hyperplasia (CAH).

Individuals with AI are genetic males who have normal plasma levels of androgens during the critical developmental periods, but due to defective androgen receptors, they have the external appearance of females and are hence reared as females. Research into the cognitive ability of individuals with AI provides insight into the role that androgen receptor mediated mechanisms play in regulating cognition, although we are aware of only two studies to have done so [35],[36], both utilizing the Wechsler Adult Intelligence Scales (WAIS) [37]. Individuals with AI are significantly impaired relative to both control males and females on the Digit Symbol, Block Design, Picture Arrangement, and Picture Completion WAIS subtests [35]. Similarly, in 15 individuals with AI the lowest mean scores from the WAIS and WISC subtests were Object Assembly, Block Design, Picture Completion, Digit Symbol, and Picture Arrangement [36]. Together these data suggest a role for androgen receptors in the development of both visuo-spatial ability and processing speed, although whether these impairments are caused by a lack of androgen receptor mediated organizational or activational effect remains to be determined.

CAH results in an overproduction of adrenal androgens beginning prenatally. Females with CAH show performance superior to controls on visuo-spatial tasks known to show large sex differences [30]. In a recent study, a relatively large number of patients with CAH (40 females and 29 males) were compared to controls on two mental rotation and two targeting tasks. Females with CAH did not significantly outperform controls on the mental rotation tasks; however, the results were in the predicted direction. Females with CAH also outperformed controls on both targeting tasks. Conversely, males with CAH performed worse than controls on both mental rotation tasks and similarly to controls on both targeting tasks [38]. These data were consistent with the findings of a smaller study which compared the performance of 5 preadolescent boys with CAH to 4 controls on the PMA Spatial Relations test [39]. In contrast, other researchers have not found differences between males with CAH and controls in spatial performance [40],[41]. Due to the fact that for males with CAH, prenatal T levels are generally not elevated [42], Hines and colleagues interpreted these data as suggesting that mental rotation ability develops primarily during the first six months of postnatal life when androgen levels in males are elevated.

The organizing and activating effects of sex hormones and performance on visuo-spatial tasks have also been investigated in male and female transsexuals. Nineteen female-to-male (FM) transsexuals, 22 male-to-female (MF) transsexuals, 20 heterosexual male controls (MC) and 23 heterosexual female controls (FC) were administered a battery of five visuo-spatial tasks, including three mental rotation tasks (Rotated Figures–2D [43]; MRT [9]; Rotated Figures-Same Different [44]). Analyses of pre-treatment cognitive functioning of all groups revealed a statistically significant linear increase from FCs to FMs to MFs to MCs on four out of the five visuo-spatial tests [45]. These data suggest an organizing effect of sex hormones on cognitive performance in homosexual transsexuals.

It is possible that a number of psychosocial factors, for example, core sexual identity, may have affected the performance of these individuals on the spatial ability tasks tested. For example, an investigation into the role of gender trait possession and performance on two visuo-spatial tasks, including a 3-D mental rotation task, demonstrated that gender trait possession, in particular, androgyny (i.e., relatively high self rating of both masculinity and femininity), added significantly to the overall explanation of performance on the 3-D mental rotation task [46]. Recently, psychological gender (M-F) was also found to account for 9% of the variance in Vandenberg and Kuse MRT performance in 60 heterosexual males, 60 homosexual males, 60 heterosexual females and 60 homosexual females [47].

Activational effects of endogenous androgens

Studies into the activational effects of endogenous sex hormones on visuo-spatial abilities in males, while numerous, have yielded conflicting results. The inconsistencies plausibly arise because of differing research methodologies, the many factors which co-vary with the hormonal factors of interest, and the difficulties in discriminating cause from effect. The following discussion is limited to studies involving adult males and performance on the most relevant visuo-spatial tasks (i.e., those requiring mental rotation or virtual navigation).

Several studies have reported medium sized positive linear relationships between activational levels of T and performance on visuo-spatial tests requiring mental rotation [48-52]. T was recently found to explain more variance in VMWT navigational performance than either age or sex [15]. Conversely, inverse relationships [53-55], and a complete absence of any relationship have also been reported [56-63]. Methodological issues, including methods for measuring T, lack of control of factors known to affect endogenous T levels, and differing measures of mental rotation, or differences in test administration, or some combination of factors, may be responsible for the inconsistent findings.

Three different measures of T are used: Total Testosterone (TT), Bioavailable Testosterone (BT) and Free Testosterone (FT). Whilst TT, BT, and FT are measured in serum, FT can also be measured in saliva. TT is the sum of both the unbound (free) and bound T (i.e., the fraction bound to either sex hormone binding globulin (SHBG) or albumin). BT is the sum of both FT and the fraction bound to albumin and is suggested to be the best indicator of tissue exposure to androgens [64]. FT measured in serum or saliva is considered to reflect unbound T, that is, the proportion not bound to either SHBG or albumin. The FT measure is typically employed because unlike bound T, it is argued that only FT can pass the blood-brain barrier and hence potentially influence cognitive function [65]. Although the measurement of FT from saliva as opposed from serum offers advantages for researchers, it is important to note that recently methodological factors related to sample collection and storage procedures have been demonstrated to affect FT measurement accuracy [66-68]. These factors may have confounded the results of previous studies and should be considered in future research.

T levels are highest in the early morning and decline throughout the day [33] and it has been shown that in males T variation due to the diurnal cycle influences visuo-spatial ability [54]. In males living in the northern hemisphere, T levels are higher in autumn than in spring [33]. Based upon the analyses of composite scores derived from three visuo-spatial tasks (Hidden Figures [43]; Paper Folding [43]; and Mental Rotations) and samples of salivary T, it has been shown that a male group studied in spring outperformed a male group studied in autumn [69]. Taken together, these studies [54],[69] suggest that high FT levels in young adult males may be associated with poor visuo-spatial performance. It is important to note, however, that this effect was found only in right-handed individuals in one of these studies [54], that both studies utilized a between rather than a within-subjects experimental design and that in neither study other variables known to influence T levels, such as obesity, smoking, alcohol, sexuality, age and presence of chronic disease were either controlled for or reported.

To illustrate the influence of potential confounds, two conflicting studies, which were very similar methodologically, are compared. Silverman and colleagues [52] and Moffat and Hampson [54] both conducted studies using male university students. Both studies used salivary samples to measure FT, both utilized the Vandenberg and Kuse MRT and both used the same laboratory for their saliva sample assays. Additionally, both studies took saliva samples in the early morning and reported highly comparable mean T levels. Despite these methodological similarities, Silverman and colleagues reported a statistically significant positive correlation between T levels and visuo-spatial ability (r = 0.28, p < 0.05) whilst Moffat and Hampson reported a statistically significant negative correlation (r = −0.44, p < 0.01). Silverman and colleagues wrote that the only methodological difference between the two studies in relation to this outcome was the fact that they used different versions of the same visuo-spatial test. Additionally, however, we note that 21 out of the 40 male participants in Moffat and Hampson's study were left-handed; Silverman et al. failed to screen participants for neurological and psychiatric conditions; and both studies failed to control for obesity, smoking, daily alcohol intake and sexuality. Further, methodological differences related to both the transport and storage of the salivary samples could have significantly affected the T measurements [67] and may account for the discrepancies in results.

A proposed quadratic (i.e., inverted-U shaped) relationship between endogenous gonadal steroids and visuo-spatial performance [70] has been used to explain contradictory results between various studies. The essence of this theory, based upon data from both males and females, is that there is an optimal level at which gonadal steroids induce peak visuo-spatial performance. Deviations from these, that is, levels either higher or lower than this optimal range, result in decreased performance. Despite the popularity of this theory, the empirical evidence typically cited in its favor is rather weak [53-55],[65]. Individually, none of these studies has actually obtained the proposed quadratic function in males [52]. Additionally, the contrary results of the two studies just described seem inconsistent with the theory on logical grounds: Whilst both studies reported similar mean FT levels, one reported a negative and the other a positive relationship. Other weaknesses in the theory include the lack of discrimination between organizational and activational effects of gonadal steroids and the lack of consensus as to whether it is T itself, or its derivatives [71], which are responsible for the proposed relationship [54]. Notwithstanding, it is interesting to note that non-linear relationships have been reported between endogenous T levels and performance on non-spatial cognitive tests in males [72]. Additionally, several studies have reported significant impairments in visuo-spatial performance following the exogenous administration of supraphysiological doses of T, both in adult men [73] and in male rodents [28],[74],[75].

Activational effects of exogenously administered androgens

The reported effects of T supplementation on visuo-spatial ability are contradictory and the effects appear dependent on the methodology used and the age groups studied. Four studies that investigated the effects of T supplementation on visuo-spatial abilities in young hypogonadal adult males reported no effect [73],[76-78]. Three of these studies [73],[77],[78] had small sample sizes (six, eight, and seven hypogonadal men, respectively) and failed to include a mental rotation task. Alexander et al. also failed to observe an effect despite employing a much larger sample of 33 young hypogonadal adult males and utilizing four visuo-spatial tasks, including a mental rotation task [76]. Nonetheless, the potential for longer-term effects of T supplementation cannot be ruled out on the basis of these studies.

The effects of supraphysiological doses of T on visuo-spatial ability have also been investigated in normal healthy eugonadal young males. Four studies reported that, in comparison to control groups, supraphysiological doses of T had no effect on visuo-spatial performance [76],[78-80], whilst a fifth study reported negative effects [73]. After six weeks of supplementation with high-dosage T injections no effect was observed in eight young eugonadal adult males [78] on the Revised Minnesota Paper Form Board Test-Form AA [81]. Similarly, after six weeks of supplementation with supraphysiological doses of T, no statistically significant effects, beyond practice effects, were observed in 10 young eugonadal adult males on a battery of tests of cognitive ability including four visuo-spatial tasks (i.e., Mental Rotations, Surface Development, Paper Folding, and Hidden Patterns) [76]. Consistent with these data two studies reported that supraphysiological doses of T had no statistically significant effects on spatial memory [79],[80]. Interestingly, O'Connor and colleagues observed that performance declined significantly on the WAIS Block Design task in 14 young eugonadal adult males who received four weeks of supraphysiological doses but performance returned to baseline levels after eight weeks of treatment [73].

Due to the relatively small sample sizes employed in these studies and their heterogeneous visuo-spatial measures, it is difficult to draw definitive conclusions from the data. Nevertheless, it appears that supraphysiological T doses in young adult males neither improve nor significantly impair spatial performance, at least over the intermediate term.

There is considerable interest in the effects of T supplementation in older adult males due to the hypothesized benefits on aging and cognition. The majority of studies have reported a beneficial effect of T supplementation on visuo-spatial ability in aging men [82-89], although four studies have failed to detect any effect [90-93]. Four studies have reported positive effects following T supplementation in older adult males on the Block Design test, whilst only one study has reported no effect (seeTable I). The Block Design test is a timed test of visuo-spatial ability that measures a person's ability to analyse and construct abstract figures from their component parts [37]. Interestingly, it appears that the ratio of serum T to E2 levels following T supplementation may play an influential role in these observed improvements in test performance. Cherrier and colleagues, for example, reported that E2 followed TT as a significant predictor of Block Design performance, explaining an additional 26% of the variance in a regression analysis [83]. Janowsky and colleagues also reported a statistically significant improvement in Block Design performance but only after correcting for baseline E2 levels. Similarly, a multiple regression analysis of the treatment group's performance revealed E2 but not T as a significant contributor to the model [86]. Two recent failures to replicate this effect on the Block Design test [90],[91] may be because the effect of E2 levels were not accounted for. For the Self-Ordered Pointing Task (SOPT) [94] a measure of executive function, T supplementation improved older adult males' performance to levels approximately equivalent to a younger adult male control group [85]. Consistent with the pattern of results outlined, better performance was related to higher FT to E2 ratios. Interestingly, in aged male rodents a similar pattern of results pertaining to improvements in spatial working memory performance following T supplementation has also been observed [95].

Table I.  Summary of effects of T supplementation on tests of visuo-spatial ability in older adult males.

Visuo-spatial ability test	Improvement	No effect
Block Design	Cherrier et al. [84]	Haren et al. [90]
	Cherrier et al. [83]	Lu et al. [91]
	Cherrier et al. [82]
	Janowsky et al. [86]
Self-Ordered Pointing Task	Janowsky et al. [85]	Cherrier et al. [87]
3-D Spatial Memory	Cherrier et al. [84]	Cherrier et al. [96]
	Cherrier et al. [87]
	Cherrier et al. [82]
Visuo-Spatial Memory	Gray et al. [88]
The Clock Drawing Test	Tan & Pu [89]	Kenny et al. [92]
Visual Reproduction		Janowsky et al. [86]
JOLO		Lu et al. [91]
SALT		Cherrier et al. [82]
RVDLT		Sih et al. [93]
RVDTR		Sih et al. [93]

The Clock Drawing Test [157]; Visual Reproduction [158]; JOLO is Judgment of Line Orientation [159]; SALT is Spatial Array Learning Tests [82]; RDVLT is Rey Visual Design Learning Test [160]; RVDTR is Rey Visual Design Test-Recall [160].

Table II.  Summary of the effects of endogenous testosterone levels on processing speed in older adult males.

Authors	Tests	Testosterone measure	Age group Or mean (years ± SD)	Effect
Hogervorst et al. [111]	Pattern and Letter Comparison Test	TT	61–72	Positive
			72–87	Negative
Moffat et al. [51]	Trails A	FTI	64.1 ± 9.4	Positive
Yaffe et al. [112]	Digit Symbol	BT	73.0 ± 7.1	Positive
Muller et al. [113]	Composite: (Digit Span Forward Digit Span Backward Digit Symbol and Trails A)	TT and FT	71–80	Positive
Fonda et al. [56]	Digit Symbol	TT and FT	62.7 ± 8.2	No Effect
Lessov-Schlaggar et al. [114]	Trails A	TT	63.1 ± 2.1	No Effect
	Digit Symbol	TT	63.1 ± 2.1	No Effect

TT refers to total testosterone; FTI, free testosterone index; BT, bioavailable testosterone; FT, free testosterone; Pattern and Letter Comparison Test [161]; Trails A [162]; Digit Symbol [37]; Digit Span Forward [158]; Digit Span Backward [158].

A recent study investigated the effect of aromatization of T to E2 following T supplementation by using anastrozole, an aromatase inhibitor [87]. Sixty healthy, community-dwelling males aged from 50 to 90 years were randomly assigned to one of three groups: A T only group (T); a T and anastrozole group (AT); and a placebo group (P). Participants were tested on a battery of cognitive tests, including a 3-D spatial memory route test [82] and the SOPT, at baseline and then at weeks three and six of treatment. A statistically significant improvement from baseline performance was observed on the 3-D spatial memory route task for the AT group but not for either the T or P groups. No treatment effect was observed with the SOPT, although these researchers used a short version of the task. These data suggest that T but not E2 was responsible for the observed improvement in 3-D spatial memory performance. Consistent with these results, Cherrier and colleagues also showed that supplementing older hypogonadal men with DHT improved their performance both on their 3-D spatial memory task and the Spatial Array Learning Tests (SALT) [82] when compared to baseline levels [96]. Together these data implicate the role of non-aromatizable androgens in the improvement of visuo-spatial performance with androgen supplementation in older adult males.

It is also interesting to note that two of these studies which found no effect on tests of visuo-spatial ability [i.e., [90],[92], specifically employed older men with lower T levels. These results are consistent with studies in young hypogonadal adult males, which similarly reported no effect [73],[76-78]. This pattern of results suggests that baseline gonadal status is an important factor which mediates the effect of exogenous T supplementation on visuo-spatial ability in adult males. Further research utilizing groups of older males with differing baseline gonadal status and/or dosage regimes may elucidate this issue.

Endogenous sex hormones, aging, and cognition

Plasma T levels decrease progressively but variably as men age [97],[98]. A recent large scale cross-sectional study of 400 community dwelling men aged between 40 and 80 years reported that endogenous levels of TT, BT, and dehydroepiandrosterone sulfate decline with age at a rate of 0.2, 0.7, and 1.2% a year, respectively [99]. It has been estimated that approximately 7% of 40–60-year-old males, 20% of 60–80-year-old males, and 35% of males older than 80 years have TT levels below the normal lower limit of 12 nmol/L (350 ng/dl) [100]. Levels of E2 do not appear to change with age [99].

Theories of generalized age-related cognitive decline and endogenous gonadal steroids

The declines in endogenous sex hormones that occur with aging in males may in turn be part of the reason for, or at least associated with, the age-related declines in several pertinent domains of cognitive function. Whilst some cognitive abilities, such as vocabulary (i.e., verbal ability) improve with age until the mid 60s, or later, before declining, others, such as mental speed, reasoning ability, and memory, decline nearly linearly with age [101]. Visuo-spatial ability, as measured by traditional psychometric tests, also declines with age at a rate similar to these other abilities [102]. Significant age-related deficits in human place navigation measured by the VMWT have also been reported [15],[20],[23]. Despite age-related declines in visuo-spatial abilities across the lifespan, however, the male advantage remains robust [1],[102].

There are two main psychological theories proposing causal factors responsible for generalized cognitive decline in humans: (i) processing speed theory; and (ii) the theory of pre-frontal decline.

Processing speed theory [3] accounts for age-related declines in fluid abilities in terms of general cognitive slowing. Fluid abilities are considered to be biologically based and are thought of as fundamental to reasoning and novel problem-solving and are not dependent upon culturally attained knowledge [7],[103]. These fluid abilities include varieties of memory, reasoning, and visuo-spatial ability.

It has been argued that a major contributor to the age-related decline of fluid abilities is the general slowing of the speed of execution of cognitive operations, or speed of information processing. Critics of this theory have questioned the validity of the concept of ‘processing speed’ and the measures used to assess it [104]. This criticism is justified by the fact that there are at least six different types of variables (i.e., decision speed, perceptual speed, psychomotor speed, reaction time, psychophysical speed and time course of internal responses) used to assess processing speed [2]. Nevertheless, the variables used to gauge processing speed have consistently demonstrated a moderate to strong relationship with age [105-107]. Furthermore, when processing speed is statistically controlled it has been found to account for about 75% of the age-related variance in a wide variety of memory and cognitive measures, including visuo-spatial abilities [2],[108]. Age-related decline in processing speed has also been implicated in age-related deficits in mental rotation performance [109],[110].

The mechanisms by which changes in processing speed occur in the aging brain remain unclear. Changes in myelination, neurotransmitter activity and weakened inhibitory circuits have all been hypothesized to contribute to decline in processing speed [104]. Whether any of these changes are associated with age-related declines in the levels of endogenous gonadal steroids is yet to be determined, although some recent research has provided support for a relationship between endogenous sex hormone levels and processing speed. For example, it was recently reported that men younger than 72 years with higher TT levels performed better on a composite processing speed measure, whilst for older age groups higher TT levels were associated with poorer performance [111]. This trend remained in the analyses even after adjusting for potential confounds such as SHBG levels, years of education, depression, BMI, current smoking and alcohol use. Consistent with these data, Moffat and colleagues [51] reported that in 254 men, faster Trails A performance was associated with a higher free testosterone index (FTI) after adjusting for health and disease-related variables. Similarly, in 310 community dwelling older men aged over 50 years, those with higher levels of BT performed better on the Digit Symbol test from the Wechsler scales, after adjusting for age and education [112]. Further, in 84 men aged between 71 and 80 years, those with both higher TT and BT levels scored better on a composite processing speed measure, which included the Digit Symbol test. Quintiles analyses revealed differences between the highest and lowest extremes of about 1 SD of the Z-score distribution. No associations were found, however, in the younger aged groups [113].

In contrast to these findings, a recent large-scale epidemiological study of 981 middle- to old-aged men found that logged FT and TT scores were not predictive of Digit Symbol performance in adjusted models [56]. Similarly, Lessov-Schlaggar and colleagues [114], failed to find a relationship between TT and neither Digit Symbol nor Trails A performance in 349 twins; however, in this study the cognitive tests were administered 10 to 16 years after the blood samples were collected and analyses were not adjusted for BMI.

Due to differences in the hormone and outcome measures and the statistical methods utilized, it is difficult to draw conclusions from these data. Nevertheless, both an effect of age and of endogenous gonadal hormone levels on processing speed is apparent in men, and further research to determine definitive relationships is required.

The theory of prefrontal decline proposes that atrophic changes in the brain's frontal lobes accompanying aging are responsible for a disproportionate decline in executive functions [115],[116]. Executive functions are best conceptualized as having four key components: volition; planning; purposive action; and self-regulation [117]. All of these executive mental capacities are intrinsic to the performance of complex behaviors and, based upon neuropsychological evidence, are generally considered functions of the brain's frontal lobes [117]. Executive functions are intrinsic to the performance of visuo-spatial tasks [118-120]. For example, a subregion within the frontal lobes, the dorsolateral pre-frontal cortex (DLPFC), has been reported to be particularly susceptible to age-related decline in function [121],[122]. Consistent with the theory, Raz and colleagues found significant relationships between the volume of the DLPFC and performance on the visuo-spatial mental imagery tasks, including several involving mental rotation; this relationship, however, did not hold for the other recruited cortical regions investigated [123]. Apart from neuronal volume, pronounced age-related neurobiological changes of the frontal lobes potentially related to impaired executive function with aging include: decreased synaptic density; the concentration, synthesis, and number of receptor sites for some neurotransmitters, specifically dopamine, and increases in senile plaques [115].

There is some data to support a mediating role of endogenous gonadal steriods on the rate of pre-frontal decline. For example, in 547 community-dwelling men aged between 59–89 years men with higher levels of BT and lower levels of E2 performed better on the Blessed Information Memory Concentration (BIMC) test [124] (a measure of mental control) after adjusting for age and education [72]. Perry and colleagues [125] also reported that in 81 community-dwelling males aged over 55, declining BT was associated with impairment in executive function, as assessed by the Executive Interview measure (EXIT) [126]. Recently, in 395 men aged between 40 and 80 years, there was also a significant linear relationship reported between TT and composite executive function scores after adjusted multivariate analyses [113]. In accordance with these data, T supplementation in older males improved performance on the SOPT to levels almost equivalent to those of a young adult male control group [85]. The SOPT is widely used as a test of executive function and has proven an effective measure of frontal lobe dysfunction [127]. Consistent with the findings of Barrett-Connor and colleagues, better performance in this study was related to higher FT to E2 ratios [72].

Animal research into age-related cognitive decline, predominantely using rodents, has demonstrated that damage to the hippocampus produces cognitive deficits highly comparable to those observed with aging, specifically in relation to spatial navigational behavior [128],[129]. Spatial navigational performance measured with the VMWT is similarly dependent on hippocampal function in humans [130]. Age-related structural and physiological changes in the rodent hippocampus associated with impaired MWT performance include: alteration in Ca 2+ conductance across CA1 pyramidal cell membranes; decreases in the NMDA receptor-mediated responses at performant path synapses onto dentate gyrus granule cells; and loss of functional synapses in the CA1 and actual synapses of the dentate gyrus [129]. In aged humans, increased neuronal loss and/or density in the hippocampus, implicated by decreases in the ratio of the metabolites N- acetylaspartate and creatine and volume is related to impaired VMWT performance [20]. Data from studies in animals suggests that gonadal steriods may be involved in some of these morphological changes within the hippocampus. For example, there is some indication that E2 is neuroprotective. Intracerebral administration of fadrozole, an aromatase inhibitor, into intact male rodents enhanced kainic acid induced neurodegeneration in the hippocampus. E2 administration counterbalanced this effect and fadrozole blocked any potential neuroprotective effects of T in castrated male rodents. Together these results suggested that it was E2 aromatized from T which had neuroprotective effects [131]. Further evidence of estrogen's neuroprotective effects are reviewed elsewhere [132].

Gonadal steroids also have beneficial neuromodulatory effects acting via traditional steroid receptors within the hippocampus. For example, in adult male rodents a small intrahippocampal T dose enhanced MWT performance on the first day of acquisition; however, larger doses were observed to impair performance on the final day of acquisition. Subsequent intrahippocampal injections of different doses of flutamide were also observed to impair MWT performance on the final two days of acquisition [74]. Consistent with these data, large intrahippocampal T doses or doses of anisomycin, a protein synthesis inhibitor, were demonstrated to impair performance on the MWT in male rodents, however, the group which received both T and anisomycin together performed similar to controls [75]. Together these data suggest that both genomic, androgen receptor mediated effects, and non-genomic, non-androgen receptor mediated effects, mediate the activational effects of T on spatial performance in adult male rodents.

Non-genomic neuromodulatory effects of gonadal steriods have also been associated with spatial performance in rodents. For example, androgen insensitive male rodents expressing the tfm mutation of the androgen receptor gene, but with near-normal androgen levels in adulthood perform intermediately between normal males and females on the MWT [25]. Whether this effect is organizational or activational, however, is indeterminate. Administration of T or DHT but not E2 to castrated male rodents increased CA1 spine density to normal levels, which reversed the 50% decline in spine density caused by castration [133]. Moreover, fimbria-fornix transections in male rodents, which sever the axons projecting between the basal forebrain and the hippocampus, block increases in CA1 spine density induced by T [134]. Scopolamine, a muscarinic receptor antagonist, was observed to cause nearly a three-fold increase in working memory errors in castrated male rodents relative to those gonadally intact in an eight-arm radial arm maze [135].

The age-related loss in the male advantage on the MWT corresponds to a decline in the size of basal forebrain cholinergic neurons (BFCN) in male rodents relative to females [136]. Estrogen has trophic effects on BFCNs in female rodents [137],[138]. Intrahippocampal injections of E2 in male rodents have been demonstrated to enhance spatial memory performance in the MWT 24 hours after initial training; an effect subsequently blocked by a subeffective dose of scopolamine [139].

These data suggest an interaction between gonadal steroids and the cholinergic system underlying hippocampal based changes in spatial performance in rodents. These activational effects appear to be modulated primarily by T, which has been demonstrated to modulate choline acetyltransferase activity in the hippocampus and other regions associated with spatial processing in adult male rodents [140]. Nonetheless, nongenomic effects on spatial behavior may also be associated with interactions between gonadal steroids and other neurotransmitter systems, such as the dopaminergic [141-144], serotonergic [141],[142],[145], or the noradrenergic and adrenergic systems [146]. Feedback mechanisms acting between these neurotransmitter systems following the exogenous supplementation of gonadal steroids may also underlie changes in spatial performance. Alternatively, gonadal steroid modulations of hippocampal CA1 neural plasticity (i.e., E2; [147],[148]; and T; [149-151]) may account for both the genomic and non-genomic effects on spatial performance.

Conclusions and future research

Taken together the data reviewed here suggest a relationship between gonadal steroids and visuo-spatial ability in males. Whilst organizational effects appear to be evident throughout the lifespan, activational effects are most apparent in older males. This may relate to the modifying effect of gonadal steroids on changes in cognitive function, specifically, processing speed, prefrontal and hippocampal function. At present the exact nature of these relationships and the mechanisms by which they occur remain unresolved. Non-human animal research into gonadal steroid effects on the hippocampus and related brain structures suggests that gonadal steroids have neuroprotective effects, and that both genomic and non-genomic activational mechanisms are critical to maintaining spatial performance in adult males. These conclusions may similarly transpose to humans. The question of whether activational effects contribute to generalized age-related cognitive decline requires further research. Supplementation studies, in particular, should attempt to control for constituent abilities related to visuo-spatial task performance, such as processing speed, executive and hippocampal function, and attempt to define mediating relationships. Furthermore, specific attention should be paid to the issue of baseline gonadal status and/or the effects of differing dosage regimes.

Based upon non-human data, a potentially promising line of research would be to investigate the effects of gonadal steroids and their interaction with the basal forebrain cholinergic system in humans. According to the so-called ‘cholinergic hypothesis’, loss of cholinergic function within the central nervous system significantly contributes to age-related cognitive decline and Alzheimer's disease (AD) [152],[153]. A recent review focusing on the behavioral influences of the basal forebrain cholinergic system argued for its principal involvement in attentional processes: Cholinergic transmission simultaneously amplifies the detection of attention demanding signals and suppresses the retrieval of internal associations [154]. It is possible, therefore, to speculate that some of the age-related cognitive changes outlined in this review, specifically in relation to processing speed, executive and hippocampal function may be related to deficits in attentional processes mediated by changes in gonadal steroid levels. Consistent with this hypothesis, an investigation into attention and aging found that TT levels were inversely correlated with alert, working memory and divided attention reaction times (RTs), in addition to the number of errors committed. FT levels were also inversely correlated with alert RTs and the number of errors committed [155]. Similarly, a recent investigation into the cognitive effects of 12 months of androgen deprivation therapy (ADT) in prostate cancer patients found that changes in T levels were associated primarily with impairments in processing speed (i.e., Digit Symbol and Recognition speed) and sustained attention [156]. Future research using ADT patients may elucidate this issue.

Additionally, further non-human research is needed to investigate how changes in gonadal steroid levels interact with pertinent neurotransmitter systems, neural plasticity and behavioral correlates. Such research potentially may elucidate the neural mechanisms underlying specific cognitive processes and factors affecting both generalized and pathological age-related cognitive decline. The clinical implications of such research include the potential to counter cognitive decline in older males and enhance independence and quality of life.

Glossary of terms

Table

Vandenberg and Kuse Mental Rotation Test (MRT)	This visuo-spatial test requires discriminating between mirror images and rotated 2D representations of 3D block objects.
Morris Water Task (MWT)	This visuo-spatial task was designed to study learning in animals. The task consists of a round pool of water which contains a submerged platform. Animals are placed into the pool and timed to reach the platform.
Processing Speed	Refers to the speed of execution of cognitive operations or speed of information processing. Measures are not dependent upon culturally attained knowledge and are generally easily completed in the absence of time constraints.
CA1 Pyramidal Cells	Are multipolar neurons with a triangular shaped cell body located in the CA1 subfield of the hippocampus. The CA1 subfield is the first region within the hippocampus which contains a major output pathway.
NMDA(N-methyl-D-aspartic acid)	An NMDA receptor agonist and an excitotoxin involved in pathological processes within the brain.
Fimbria-fornix transactions	A surgical procedure involving lesions to septo-hippocampal axons within the dorsal fornix and fimbria. The procedure causes cholinergic dennervation of the hippocampus.