Stopping behavior is an important function for an individual in daily activities, such as stopping driving upon seeing an unexpected pedestrian cross the road. Thus, decreasing the adaptive stopping ability may cause severe danger. It has been shown that when people get older, they are more susceptible to distraction (distractibility hypothesis of aging; see Healey et al., 2008 for a review) and disinhibition (inhibition deficit hypothesis of aging; e.g., Hasher and Zacks, 1988; Dempster, 1992; Hasher et al., 1999; Gazzaley and D'Esposto, 2007), resulting in more response intrusions and/or interference from irrelevant stimuli. However, other research indicates that not all older adults are affected by disinhibition, and there might be exceptions either because of individual differences in their brain activity or performance strategies (Hsieh and Fang, 2012; Hsieh and Lin, 2014; Hsieh et al., 2015, 2016). Therefore, research investigating if older adults have generic deficits in inhibition, which to some extent depends upon individual differences in brain activity, is still warranted. In this study we used resting-state functional magnetic resonance imaging (RS-fMRI) in the elderly population. Our main goal was to assess the validity of this resting-state approach, in order to provide a reliable brain-measurement of inhibition in a population for which extensive task-related fMRI testing may not be suitable.

RS-fMRI is a well-known and promising tool to study the relationship between spontaneous brain activity and behavioral performance. Understanding these relationships may have practical clinical applications, especially to infer how well an individual will perform a task if their on-task brain activity is somehow difficult to acquire. Thus, recent studies have investigated the relationship between RS-fMRI and task performance using various cognitive control tasks, including the N-back working memory test (Evers et al., 2012; Sala-Llonch et al., 2012), Stroop task (Evers et al., 2012; Takeuchi et al., 2015), Eriksen flanker task (Mennes et al., 2013), and California verbal learning test (Ystad et al., 2010). In addition, RS-fMRI has also been shown to be correlated to stopping ability (Tian et al., 2012; Hu et al., 2014), which is of main interest in this study. However, these two previous studies were either focused mainly on healthy young adults or biased to sample more healthy young adults across a life-span database, and thus, the association of RS-fMRI with stopping behavior in the middle-aged and elderly is unclear. Therefore, the main purpose of this study is to address this knowledge gap.

Several analytical methods for measuring RS-fMRI have been developed since Biswal et al.'s (1995) pioneering work (for a review, see Zuo and Xing, 2014). These analytical methods can be broadly classified into two categories: one for depicting functional connectivity (FC) between remote brain regions, and one for local FC. The spatial scale for differentiating local FC and remote FC is usually between 10 and 15 mm (e.g., 14 mm used in Sepulcre et al., 2010). The widely used methods in the latter category (local FC) include regional homogeneity (ReHo; Zang et al., 2004), and the amplitude of low-frequency fluctuations (ALFF or fractional ALFF [fALFF]) (Zou et al., 2008). The common rationale for ReHo and f/ALFF methods is that the identification of similar local features of the spontaneous BOLD signal among neighboring voxels within small clusters provides an account of regional functional connectivity. Therefore, ReHo and f/ALFF may be equally useful for exploratory or clinical research because they involve data-driven analyses of the entire brain (whole-brain approach), requiring no a priori selection of brain regions of interest (ROIs), which is required for other RS-fMRI methods such as the seed-based approach (Biswal et al., 1997; Cordes et al., 2000; Jiang et al., 2004). However, ReHo and f/ALFF differ in terms of their clustering algorithms (i.e., definition of similarity), as follows: ReHo measures the temporal synchronization by calculating Kendall's coefficient of concordance (Kendall and Gibbons, 1990) for the time series of a given cluster of neighboring voxels (time-domain analysis), whereas f/ALFF measures the correlation of amplitude/fractional amplitude of spontaneous low-frequency (0.01–0.1 Hz) voxel fluctuations (frequency-domain analysis). Some studies showed strong coupling relationships between ALFF and fALFF (Zou et al., 2008; Zuo et al., 2010), and strong positive correlations between ReHo and ALFF, which suggests that high spontaneous enhanced activity in a given voxel is accompanied by increased synchronization of the surrounding voxels and enhanced amplitude fluctuations of the resting-state blood oxygenation level dependent (BOLD) signals (Yuan et al., 2013; see also Nugent et al., 2015). However, some studies suggested that f/ALFF may be complementary to ReHo, thus suggesting that researchers apply both ReHo and f/ALFF methods to explore which method is more sensitive to local abnormalities and the extent to which they can detect different abnormalities in clinical populations (An et al., 2013; see also Han et al., 2011; Lei et al., 2012; Cui et al., 2014; Premi et al., 2014). Ni et al. (2016) further advocated the advantage of applying these two complementary methods in a study by commenting that: “These two methods are based on different neurophysiology mechanisms…Since the two methods found some changes in common cerebral functional regions, both were adopted to reduce inaccuracies and to provide reliable and comprehensive conclusions” (Ni et al., 2016, p. 1,251).

The aim of this study was to investigate the relationship between stopping performance and spontaneous brain activity in the elderly population, and to assess the feasibility of this approach for future use with older adults for whom extensive task-based fMRI testing may not be suitable. As the brain ages, several brain regions and connectivity networks could be altered in terms of dynamics and location, decreasing the accuracy of ROI- and seed-based analyses. For example, Tomasi and Volkow (2012) reported that aging was associated with pronounced decreases in long-range functional connectivity density in the default mode network (DMN) and dorsal attention network (DAN), and that it was also associated with increases in somatosensory and subcortical networks. Evers et al. (2012) further observed that age-related decreases in functional connectivity in DMN begin in middle-age. Moreover, some aging studies have also shown that the local and remote FC may be modulated by aging with a bias from the remote to local FC in the aged population (Tomasi and Volkow, 2012). Under these circumstances, whole-brain data-driven methods such as ReHo and f/ALFF provide a more reliable approach. These two methods provide different, but complementary neurophysiological measures of regional FC (i.e., time vs. frequency domains). In accordance with the view presented in Ni et al. (2016), here we focus mainly on the overlap between ReHo and fALFF, by means of a conservative conjunction analysis, allowing more reliable and comprehensive conclusions regarding regional FC.

The task used in this study to examine stopping ability is the stop-signal paradigm introduced by Logan and Cowan (1984). In behavioral studies, older adults have often been shown to exhibit longer stopping time, suggesting that there is an age-related inhibitory deficit (e.g., Kramer et al., 1994; Bedard et al., 2002; van de Laar et al., 2011; Kleerekooper et al., 2016). However, other studies found either no age-related decline (e.g., Kray et al., 2009) or only a specific deficit (e.g., Anguera and Gazzaley, 2012). This suggests that aging per se may not be the only factor that modulates the inhibitory process. There could be individual differences such as individual brain functions that modulate the inhibitory process. Therefore, in this study, we are particularly interested in investigating whether RS-fMRI is correlated with stop-signal reaction time (SSRT), dependent or independent of aging.

Previous research of on-task fMRI has shown that the right inferior frontal cortex (rIFC) and (pre-)supplementary motor area ([pre-]SMA) are important brain regions that are associated with successful response inhibition (Li et al., 2006; Chevrier et al., 2007; Chao et al., 2009; Hu et al., 2014). Additional brain areas, such as the superior frontal gyrus (SFG), medial frontal gyrus (medial FG), middle temporal gyrus (MTG), precuneus, and insula have also been reported to be involved in inhibition (Aron and Poldrack, 2006; Li et al., 2006; Ramautar et al., 2006; Chevrier et al., 2007; Erika-Florence et al., 2014). However, RS-fMRI studies of response inhibition among healthy young adults have also shown that ReHo and SSRT correlate significantly in the rIFC (Tian et al., 2012), and that fALFF and SSRT correlate significantly in the pre-SMA/SMA (Hu et al., 2014). As described above, when the brain ages, several brain regions and connectivity networks could be altered, and thus, we are interested in investigating if RS-fMRI is still a valid indicator of individual differences that are associated with inhibition performance among the elderly (including the middle-aged) population. Based on previous research, we expect that there will be a significant correlation between RS-fMRI and SSRT in the rIFC and/or pre-SMA/SMA if aging is not the only factor that modulates the inhibitory process.

This study aimed to examine whether spontaneous brain activity evaluated by resting-state fMRI can be used as an indicator of individual differences in association with inhibitory performance even in healthy middle-aged and aged people. We used two local FC analysis methods (ReHo and fALFF) to evaluate the relationship between spontaneous brain connectivity and stopping performance as reflected on SSRT, and we focused on overlapping regions measured across the two methods to provide more conservative results (Ni et al., 2016). The current behavioral results showed that SSRT was not prolonged as a function of age, suggesting that there was no age-related reduction in stopping ability, which was consistent with Kray et al.'s (2009) findings. Yet, the null finding may be simply a result of a reduction in stopping ability that reached a plateau around middle age, because the current study recruited only elderly people. Alternatively, since there seemed to be a “statistical trend” (p = 0.07) and the medium effect size (Cohen's d = 0.45) seemed to suggest that taking a larger sample size might bring this effect into statistical significance. However, this issue is beyond the scope of the current study, because we focused only on elder populations.

We investigated if RS-fMRI is an effective indicator in association with stopping ability in the elderly. The current RS-fMRI results showed that the FC correlation results of the two local FC analysis methods, i.e., ReHo and fALFF with SSRT, overlapped in several brain regions (Tables 7, 8). These results can be summarized into four main features: the activities during the resting state of (1) some parts of the DMN; and (2) the left IFG and bilateral FG were involved; however, (3) the pre-SMA/SMA was not involved in association with stopping performance; and (4) the correlations between the RS-fMRI and stopping performance that we observed were negative rather than positive. The DMN is an interconnected and anatomically defined set of brain regions, consisting of some functional hubs including the posterior cingulate cortex (PCC)/precuneus, medial prefrontal cortex (mPFC), hippocampus, and angular gyrus. The DMN has been shown to deactivate during external goal-oriented tasks such as visual attention or cognitive working memory tasks but it activates during the resting state, thus, leading some researchers to label the network as the task-negative network (e.g., Raichle et al., 2001; Greicius et al., 2003). Its activities have been hypothesized to potentially influence goal-directed behavior and/or mental effort during cognitive tasks (Weissman et al., 2006; Li et al., 2007), self-referential thinking, emotional processing, and recalling memories. The current findings indicate that the medial frontal gyrus and PHG, which are parts of the DMN, were negatively correlated with SSRT, suggesting that DMN can also serve as an effective indicator to associate with stopping performance even in the middle-aged and aged population. Thus, a main contribution of the current study is to provide new evidence showing that the DMN is also involved and associated with stopping behavior.

Secondly, the current study also shows that the left IFG can serve as an indicator that is associated with stopping performance. The right IFG is well-known to be important for successful response inhibition (for reviews, see Aron et al., 2004; Verbruggen and Logan, 2009). Although the current study observed that the left IFG, rather than the right IFG, was associated with the forthcoming successful inhibition, some studies have also reported bilateral IFC activations (Bunge et al., 2002; Li et al., 2006; Tian et al., 2012) or left IFG activation (Swick et al., 2008), as we showed in this study. Therefore, the current results suggest that the left IFG may also play an important role in response inhibition.

Thirdly, in the current study, pre-SMA was not found to be associated with SSRT performance. The current results seem to be inconsistent with those of other studies, such as those by Chao et al. (2009), Chevrier et al. (2007), Li et al. (2006), and Hu et al. (2014). However, these previous studies mostly investigated the relationship between the pre-SMA and inhibition performance during the on-task period, rather than the pre-task resting period, which we investigated in this study. Therefore, the discrepancy may be attributed to the role of pre-SMA in reactive motor inhibition, rather than the attentional processing of the stop-signal that was modulated by the IFG. Thus, the current results provide indirect evidence showing that the pre-SMA is involved in reactive inhibition, whereas the IFG is involved in the attentional processing of the stop-signal (see also Duann et al., 2009).

Finally, the current findings appear to be consistent with the reports by Tian et al. (2012) showing significant correlations between SSRT and the ReHo of the ITG, STG, IFG, and medial frontal gyrus, but there is one main difference. This difference is that, while Tian et al. (2012) observed positive correlations between SSRT and the IFG, and between SSRT and the DMN (MPFC, IPL, precuneus), our results showed negative correlations. A major difference between Tian et al.'s (2012) results and the current results is in participants' age. This suggests that for younger adults, the correlation between the IFG and SSRT is positive, as shown by Tian et al. (2012), but for middle-aged and aged adults, the correlation becomes negative, as shown in the current study. Another study by Hu et al. (2014) also addressed the correlation between resting-state fMRI and SSRT, but they showed a different pattern with a negative correlation between SSRT and the fALFF in the pre-SMA/SMA. In Hu et al.'s (2014) study, participants' age varied widely, from 18 to 72 years (with 63 out of 111 healthy participants in the age range of 20–29 years, and 18 out of 111 healthy participants in the age range of 30–39 years). A potential hypothesis to explain the discrepancies between the previous findings of positive correlations and the current findings of negative correlations is a shift from remote to local connectivity during aging, which results in a polarity reversal for the correlations between RS-fMRI and SSRT.

To clarify if the current findings on the correlation between RS-fMRI and SSRT were modulated by age, we used age as a covariate while computing the correlations between ReHo/fALFF and SSRT, to partial-out the effect of age. The results showed that only the medial frontal gyrus was excluded and the left STG was additionally included, whereas the bilateral cerebellum, pons, bilateral ITG, right MTG, bilateral FG, bilateral PHG, and left IFG maintained their significantly negative correlation with SSRT. Therefore, the current findings regarding the relationship between resting state activities of these brain regions and stop processing cannot solely be attributed to the effect of age. Our age range included participants in the age range of 40–77 years (mainly middle-aged and aged healthy participants), because this study mainly focused on the elder population. Whether or not this might therefore underestimate the effect of age is beyond the current research scope and further research is required to address this issue. Finally, these results suggest that RS-fMRI may be a viable method to assess inhibition in an older population, in which it is often not feasible perform extensive task-based fMRI testing.

The results of this study showed that although aging may alter brain networks, the spontaneous activity of the age-related brain networks can still serve as an effective indicator of individual differences in association with inhibitory performance in healthy middle-aged and elderly people. The current findings have two major contributions: (1) clinical application: one may infer the stopping efficacy based on the resting-state neuroimaging for individuals who have difficulties completing the task; and (2) methodological conservativeness: this is the first study to use both ReHo and fALFF on the same dataset for conjunction analyses to reduce inferring inaccuracies and provide reliable and comprehensive conclusions regarding regional functional connectivity.

HL collected the data, analyzed the data, and drafted some parts of the manuscript. SH initiated the research idea, applied for the funding, designed the task, supervised data analyses, drafted and revised the manuscript.

This work was supported by the Ministry of Science Technology (MOST) of the Republic of China, Taiwan for financially supporting this research (Contract No. 104-2410-H-006-021-MY2).