Higher free testosterone level is associated with faster visual processing and more flanker interference in older men

Summary

The aim of this study was to investigate the relation between free testosterone (FT) level and basic cognitive functions in older men. Three cognitive computer tasks were aimed to measure visual-processing speed, ability to suppress inappropriate responses, and visuospatial cognition, respectively. The tasks employed were a visual backward masking task, the Eriksen flanker task, and a mental rotation task. Participants were 72 older men (mean age = 67.2 years, range 57–79). The influence of FT, age, alcohol consumption, and education on task performances was examined by means of multiple linear regression analysis. FT level was positively associated with accuracy on the short intervals (33 and 50 ms) of the backward masking task and negatively with response latency on congruent flanker-task trials. These results indicate faster visual-processing speed in older men with higher FT levels. FT level was positively associated with interference elicited by irrelevant incongruent flankers, which implies that older men with higher levels of free testosterone exert less inhibitory control. Consistent with previous research, higher FT levels were associated with faster ‘same-different’ responses for unrotated mental rotation stimuli, but not with the slopes of the response time rotation functions. Older age was associated with lower backward masking accuracy at 33 ms and with lower overall mental rotation accuracy. Higher education level was associated with faster processing speed for congruent flankers and with higher overall mental rotation accuracy. Alcohol consumption did not affect the task performances. Our results suggest that higher FT levels in older men may promote sustained visual-processing speed and visuospatial ability, probably at the expense of some inhibitory functioning.

Introduction

With aging, both cognition and testosterone levels decline. Because testosterone has a subtle neuromodulatory role and probably also acts as a neuroprotective agent in brain aging (Janowsky, 2006), testosterone and cognitive aging may be related. With the mean population age rising, research on the association between testosterone and cognition in older people is becoming increasingly important.

In men, the level of bioavailable testosterone steadily declines from age 30 onwards, with a significant proportion of men over 60 years of age having a testosterone deficiency (Harman et al., 2001). Testosterone exerts its neuromodulatory influence through androgen receptors in specific human brain regions, such as the hippocampus, amygdala, and prefrontal cortex (see for review, Janowsky, 2006). In male rats, synaptic density in the hippocampus can be grossly reduced by gonadectomy, and be returned to normal by testosterone (but not estradiol) supplementation (Leranth et al., 2003).

Higher free testosterone (FT) levels are associated with increased hippocampal blood flow in men (Moffat and Resnick, 2007). Consistent with hippocampal androgen activity, studies investigating testosterone decline, deprivation, and supplementation have demonstrated that in men there is a positive association between testosterone levels and memory (e.g., Janowsky et al., 2000, Cherrier et al., 2001, Cherrier et al., 2005, Moffat et al., 2002, Bussiere et al., 2005).

The association of testosterone and cognition is sex-specific, with FT exerting less influence on cognitive functioning in women than in men (Thilers et al., 2006). An exhaustive review of the existing literature is beyond the scope of the present paper, but it is evident that the relationship between testosterone and various cognitive domains is complex and subject of ongoing debate (for reviews, see Moffat, 2005, Janowsky, 2006, Driscoll and Resnick, 2007). For instance, higher testosterone levels were associated with better spatial cognition in some studies (e.g., Janowsky et al., 1994) but not in other (e.g., Yonker et al., 2006, Martin et al., 2008). Likewise, one large population-based aging study found positive associations between testosterone and cognitive functioning (Moffat et al., 2002), whereas another did not (Fonda et al., 2005).

There may be several reasons for the discrepant findings, such as testosterone decline versus supplementation effects, small versus large age ranges, linear versus curvilinear relationships between testosterone and cognitive abilities, and the use of heterogeneous psychometric instruments to represent the same specific cognitive ability.

Curvilinear relationships between testosterone levels and specific cognitive tasks (with the highest testosterone levels going with lower cognitive performances) have been reported by Moffat and Hampson (1996) and Muller et al. (2005), among others. As Thilers et al. (2006) have noted, these studies have included young men in the analyses. In studies with restricted age ranges, older men usually show a linear relationship (e.g., Yaffe et al., 2002, Thilers et al., 2006). In the Muller et al. (2005) study, the oldest age group also showed a linear relationship between testosterone and cognitive performance.

Most research that explored the relationship between testosterone and cognition employed standard psychometric measures, that is, conventional neuropsychological paper-and-pencil tests (see Driscoll and Resnick, 2007). Such tests are indispensable for cognitive-aging research, but assess rather broad categories of cognitive functioning. For instance, the Digit Symbol Task is used in testosterone studies to assess visuospatial ability and processing speed, either alone or as part of a composite measure. In this task, the research participant has to fill in blank spaces with a symbol that is paired to a given number as quickly as possible. The Digit Symbol Task is sensitive to mental detoriation (Lezak, 1995) and correlates with bioavailable testosterone levels (Yaffe et al., 2002, Muller et al., 2005). However it is not clear to what extent performance on this task is determined by visual coding, attention, memory, or motor action and which of these processes is associated with testosterone levels. In the present study, we therefore employed well-established computerized tasks that measure basic characteristics of cognitive information processing: visual-processing speed and resistance to interference. Both speed of processing and resistance to interference (i.e., inhibitory control) are major determinants of cognitive aging. The speed of processing account of cognitive decline proposes that there is a generalized slowing of information processing with increasing age (Salthouse, 1996), while the inhibitory control account holds that there is a reduction in the ability to suppress irrelevant information and improper responses (Hasher et al., 1999).

We employed a backward masking task (see Section 2) to measure visual-processing speed and an arrow version of Eriksen's flanker task (Eriksen and Eriksen, 1974) to measure visual-processing speed and inhibitory functioning. Aging studies with backward masking suggest slower processing as age increases (e.g., Di Lollo et al., 1982) while studies employing flanker tasks suggest diminished inhibitory functioning with increasing age (e.g., Shaw, 1991, Zeef and Kok, 1993). To our knowledge, no studies have been published that investigated the relationship of testosterone levels in older men to visual-processing speed and inhibitory functioning as assessed with backward masking or the Eriksen flanker task.

In addition, we employed a computerized version of the mental rotation task. Mental rotation is the most obvious test of spatial ability in which men outperform women, with testosterone playing an important role in both male (Hooven et al., 2004) and female (Aleman et al., 2004) performance. A computerized version of the mental rotation task has the advantage that task performance can be split in independent components such as visual shape comparison and rotation speed. Note that with a paper-and-pencil version of the mental rotation task, these assembled components cannot be distinguished. With a computerized version, Van Strien and Bouma (1990) found sex differences for unrotated stimuli, but equivalent rotation rates for both sexes. This finding suggests that the male advantage for mental rotation is based on overall performances (as with paper-and-pencil tests) and not on rotational ability per se. Computing the regression lines of mental rotation performance on angular disparity, Hooven et al. (2004) found that testosterone level in men was associated with faster reaction times and better accuracy, but that these effects arose from cognitive processes unrelated to the changing of orientation, that is, from the intercepts (=shape comparison) but not the slopes (=rotation angle) of the regression lines.

In the present study, we examined the relationship between metabolically available testosterone as reflected by FT levels (e.g., Moffat et al., 2002, Yaffe et al., 2002) and basic cognitive functioning. We expected that older men with higher FT levels would exhibit better visual speed, that is, better accuracy on the short intervals of the backward masking task. From the earlier literature, there is evidence that, in healthy older men, treatment with methyltestosterone enhances the ability to perceive flickers (Simonson et al., 1944), which indicates that higher testosterone levels are associated with higher visual speed. Higher FT levels will also be associated with higher perceptual speed as reflected by the latencies for congruent trials of the flanker task, since testosterone deprivation results in visuomotor slowing and decreased cognitive processing speed (Salminen et al., 2004). Because testosterone is associated with impulsivity (Bjork et al., 2001) and other indices of lessened frontal functioning (e.g., Van Honk et al., 2004, Martin et al., 2007), higher FT levels will be related to decreased inhibitory functioning as reflected by the interference on incongruent flanker trials. In line with previous research (Van Strien and Bouma, 1990, Hooven et al., 2004), we expected better mental rotation performance in men with higher FT levels, but only for overall performance and unrotated stimuli, and not for rotation speed as expressed by the rotation slopes.