Normal aging is associated with declines in several cognitive domains, most notably episodic memory and executive functions (for reviews, see [1–4]). These cognitive deficits are associated with myriad brain changes, including structural and functional deterioration of prefrontal, basal ganglia, and medial temporal areas and their interconnections. However, establishing a link between these changes and cognitive decline in normal aging has proven surprisingly difficult [2, 5].
Alterations in two classic neurotransmitter systems have drawn considerable attention in cognitive aging: dopamine  and acetylcholine. For decades, acetylcholine (ACh) was thought of primarily as a memory-related neurotransmitter, but this view has recently been revised, with ACh now thought to play an equally if not more crucial role in executive functions (for reviews, see [7–9]). The integrity of cortical cholinergic inputs appears to be critical for modulating attention, by enhancing responsiveness to sensory inputs to facilitate cue detection and orienting  (for a review, see ). Cholinergic neuromodulation may also play an important role in executive functions by selectively enhancing task-relevant inputs via bottom-up thalamic processes, while suppressing irrelevant stimuli via top-down prefrontal modulation  (for other perspectives, see [12, 13]). This cholinergic-dependent interaction between bottom-up and top-down processes appears to be affected by aging, leading to difficulty in task-switching, handling competition among several possible responses, and suppressing unwanted responses . In memory, optimal levels of ACh may facilitate encoding by increasing the influence of inputs into the hippocampus through enhanced potentiation [9, 14], and/or by providing the attentional “glue” to bind together disparate elements of an episode into a unified memory trace [15, 16].
Experimental and correlational animal studies, as well as computational modelling, have yielded much information on the role of the cholinergic system in cognition. However, the extent to which age-related changes in cholinergic neuromodulation contribute to cognitive decline in normal human aging remains unclear. There are at least three reasons for this: First, making inferences from animal and computational models to humans has sometimes proven surprisingly difficult (e.g., [17, 18]). Second, much of what we infer about the role of ACh in cognitive aging comes from studies in which Alzheimer’s patients are treated with cholinesterase inhibitors, including donepezil, galantamine, and rivastigmine (e.g., ). Unfortunately, these patients can be difficult to test and experience other confounding factors including significant structural and functional brain changes. Third, manipulation of ACh via agonist and antagonist drugs (e.g., scopolamine) has produced a vast amount of data, but strictly speaking this line of research tells us more about acute effects than it does about the long term decline in cholinergic activity seen in normal aging. There is thus a need to further examine the in vivo contribution of age-related alterations in central cholinergic function to declines in human cognition.
Recent advances in the field of non invasive brain stimulation have yielded new opportunities to examine the neurophysiological correlates of aging using markers of cortical excitability that can be linked with relative confidence to specific neurotransmitter systems . One such marker involves pairing afferent nerve stimulation with transcranial magnetic stimulation (TMS) of the motor cortex to modulate motor responses evoked in contralateral hand muscles . When applied at short intervals (e.g., 18–20 milliseconds [ms]) before TMS pulses, afferent nerve stimulation typically leads to a period of inhibition of the motor evoked potentials (MEPs). This short-interval afferent inhibition (SAI) is mediated at the cortical level through cholinergic-dependent GABAA receptor activation . The implication of cholinergic action in mediating SAI is supported by in vivo observations of its reduction or even abolition by administration of a selective muscarinic cholinergic receptor blocker (scopolamine) in healthy participants . Further, SAI is lower than expected in Alzheimer’s patients but restored by cholinesterase inhibitors . SAI is also reduced in other disorders characterized by cholinergic dysfunction, including Lewy body dementia , multiple sclerosis , and Wernicke–Korsakoff syndrome , but it is normal in frontotemporal dementia, a non-cholinergically mediated form of dementia . Together, these observations provide strong evidence that SAI is a cholinergic-dependent marker of motor intra-cortical excitability.
Given the clear decline in cholinergic modulation with age [28, 29], one would predict that SAI would be altered in healthy older adults. Yet, very few studies have examined this issue. Oliviero et al.  compared SAI levels in healthy young and older adults and found no age differences. More recently, Degardin et al.  performed a similar study and reached a similar conclusion. However, as we and others  have argued previously, the use of varying test intensities to obtain a constant MEP size across participants might have contributed to masking any age effects in the two studies above. In line with this, we recently found a large and selective decrease in SAI in healthy seniors when we used a constant TMS test intensity approach . Further, we found that age-related variations in SAI explained a substantial proportion of the variance in timed motor tasks assessing processing speed.
This study constitutes an extension of our previous findings; data were derived from the same sample of participants as already described . In the present study, we examined possible relationships between SAI, as a putative marker of cholinergic-dependent cortical inhibition, and cognition in young and older healthy adults. Because mean differences between young and older adult groups are often small, especially relative to the extensive variability that can be seen among healthy older adults (e.g., some perform much more poorly than young people, whereas others are indistinguishable from the young ), we capitalized on the individual-differences approach used by Glisky and colleagues [35, 36]. This approach allows the characterization of each participant’s long-term memory and executive functions using neuropsychological testing to construct aggregate scores reflecting performance across several tasks in each domain (for details, see Method). We hypothesized that age-related differences in SAI levels would be associated with age-related differences in memory and executive functions. For memory, several investigators have emphasized ACh’s putative role in binding information in memory , which we assessed using a canonical measure of paired associate learning (Verbal Paired Associates from the Wechsler Memory Scale-III; WMS-III ). We also examined face recognition from the WMS-III because recent studies have also described cholinergic modulation of face-memory-related activity in the fusiform gyrus . Given the emphasis in the recent literature on the crucial role of ACh in modulating executive functions [19, 39, 40], we also expected correlations between SAI and our aggregate executive function measure.