Functional near infrared spectroscopy activation during an executive function task differs between healthy older and younger adults

Abstract

Background

Healthy aging can include declines in processing speed and executive function. Further research is needed to characterize the neurobiological underpinnings of these cognitive changes in older adulthood. The current study used functional near infrared spectroscopy (fNIRS), an optical neuroimaging technique, to examine differences in cerebral oxygenation between healthy older adults (OA) and younger adults (YA) during a measure of cognitive interference.

Methods

Thirty-four participants were sampled from two age groups: YA (mean age = 28.1 years, SD = 2.8, F = 9) and OA (mean age = 70.9 years, SD = 5.4, F = 9). Participants completed the Multi-Source Interference Task (MSIT), a measure of executive function with high and low-demand conditions, while undergoing fNIRS recordings using a TechEn CW6 system with 34-source-detector channels, situated over the prefrontal cortex. Functional activation patterns, accuracy, and reaction time were compared between and within groups for each condition.

Results

Behaviourally, during the control condition, OA and YA had comparable accuracy, although OA had significantly slower reaction times than YA. During the interference condition, OA had significantly lower accuracy and slower reaction times than YA. Results demonstrated a significant difference between groups with an age-related increase in HbO for OA in both conditions (p < 0.05). Within groups, OA showed greater activation during the control condition, while YA demonstrated greater activation during the interference condition.

Conclusions

The findings suggest that OA recruit additional neural resources to achieve similar behavioural performance during low-level cognitive interference, but that compensation in OA may be insufficient to support behavioural performance at higher levels of interference.

Introduction

Increases in life expectancy over the last century have led to shifts in global population demographics and a greater proportion of older adults [22]. Studying and defining the progression of healthy aging can present important information in differentiating normal and pathological aging [14]. Consequently, there is a need to better understand the neurobiological aging process and the underlying mechanisms of age-related cognitive changes. Decades of research on normal aging have affirmed that cognitive abilities decline with age [2]. The frontal lobes are particularly susceptible to aging, rendering cognitive skills such as executive functions particularly vulnerable to age-related decline [2]. Typical changes in cognition that occur with age directly relate to the underlying brain structures and functions that subserve them. Neuroimaging techniques have therefore been paramount to understanding the neural underpinnings of cognitive aging (Fig. 3).

Fig. 3.

Approximate probe array positioning. Primarily focused on the prefrontal cortex and the anterior section of the frontal lobe. The probe contained 10 sources and 16 detectors. [18]

To date, many studies have used functional magnetic resonance imaging (fMRI) in combination with cognitive tasks to examine age related changes in brain function- for example, a study by Avelar-Pereira et al. [3] compared healthy young and older participants and their age-related differences in brain connectivity during rest and the multi-source interference task. The multi-source interference task (MSIT) was designed for fMRI studies and is ideally suited to studying age related changes in cognition [5]. The MSIT is an attentional task with an interference component that creates robust and reliable activation in the cingulo-fronto-parietal network, which is associated with executive functioning more broadly [5]. For instance, Salami and colleagues [17] examined age related differences during the MSIT task and reported greater task related fMRI activation in older relative to younger adults, especially in dorsal lateral prefrontal and anterior cingulate cortices during interference conditions. These findings suggest that healthy older adults may over-recruit brain regions during complex cognitive tasks in order to compensate for age related changes, although corroborating research is needed (Fig. 4).

Fig. 4.

Probe showing channels (yellow and orange) that demonstrated significant age-related increases in oxygenated hemoglobin during control condition. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Although fMRI yields high resolution for spatial information, it also presents with several limitations, including limited temporal resolution, low threshold for physical movement, and high costs [19]. Therefore, it is increasingly important to investigate alternative neuroimaging methods to mitigate these limitations and to gain novel information regarding the impact of healthy aging on commonly used fMRI tasks, such as the MSIT [19]. An excellent alternative to fMRI is functional near infrared spectroscopy (fNIRS); an optical neuroimaging technique that measures infrared light at different wavelengths to detect the absorption of oxygenated (HbO) and deoxygenated hemoglobin (HbR) [7]. In comparison to techniques such as fMRI, fNIRS has better temporal resolution (with a typical sampling rate of 1–10 Hz) but decreased spatial resolution that is limited to cortical areas [7]. Due to its cost efficiency and ease of transportation, fNIRS is a popular choice for vulnerable populations that cannot easily or safely be in an MRI scanner [7]. It is also non-invasive and minimally obtrusive, which allows for assessment of brain function in a natural environment without negative side effects to participant performance or comfort [23]. However, while fNIRS is increasingly being used as an alternative to fMRI, due to its decreased spatial resolution, it is important to replicate the findings of fMRI with fNIRS.

Thus far, few studies have used the MSIT in conjunction with fNIRS. Harrivel, Weissman, Noll, and Peltier [12] investigated the applicability of fNIRS in distinguishing task engagement in the MSIT across the lifespan. They found that fNIRS could be reliably used to differentiate task and resting states. Halliday and colleagues [9], investigated healthy older adults with a history of falling, given that poor performance in executive function and attentional tasks may be a precursor to falling. Using the MSIT, they found that fallers demonstrated a wider distribution of neural activity relative to non-fallers, while group differences in behavioural performance remained insignificant. A second study by Halliday and colleagues [10], used the MSIT to investigate neural variability in cerebral oxygenation in older adults. Their results suggest that neural variability is adaptive to faster performance during cognitive interference within-persons, and not merely an artifact of individual differences. To date, previous studies that have examined MSIT task performance using fNIRS have not examined the cognitive changes between younger and older healthy adults (Fig. 5).

Fig. 5.

Probe showing channels (yellow and orange) that demonstrated significant age-related increases in oxygenated hemoglobin during interference condition. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Current study

The current study is the first to compare age-related differences during the MSIT using fNIRS. Specifically, the current study used fNIRS to investigate functional and behavioural differences between and within healthy younger and older adults during the MSIT to address the following:

• 1.How do older and younger adults differ in their behavioural performance during cognitive interference? We hypothesized that older adults would demonstrate less accuracy and slower reaction times than younger adults, with the exception of the control condition in which we expected accuracy to be consistent across both groups. We also hypothesized that the interference condition would yield lower accuracy and slower reaction times than the control condition of the MSIT, in both groups.
• 2.How do older and younger adults differ in their functional activation during cognitive interference? We hypothesized that older adults would demonstrate increased activation (as evidenced by cerebral oxygenation levels) during the control condition to support their cognitive performance. We also expected a greater increase in activation during the interference condition in older adults, in order to compensate for the increased task difficulty.
• 3.Are older and younger adults working proportionately hard within task conditions? We hypothesized that older adults would demonstrate consistently more activation than younger adults in both conditions. We expected that older adults would demonstrate greater activation in the interference condition to compensate for increased performance deficiencies, whereas younger adult brains would demonstrate an increase in activation proportionately with the task difficulty.

Methods

Participants

A total of 34 healthy adults participated in the current study, forming a younger adult group (YA, N = 17) and an older adult group (OA, N = 17), both groups had 16 right-handed participants and 1 left-handed participant. Table 1 outlines the demographic characteristics of both groups. A neurological screening measure was not used to assess the community dwelling sample in the current study. Instead, a detailed approach to inclusion and exclusion criteria was taken to confirm that all of the participants were neurologically healthy. Inclusion criteria included: no history of severe psychiatric diagnosis (including schizophrenia or bipolar disorder), no history of neurological diagnosis, including (but not limited to) any type of neurodegenerative disorder, stroke, or moderate-severe TBI. This data set was derived from a larger study which included resting-state fMRI data acquisition; therefore, individuals were also excluded based on contraindications for MRI (e.g., metal implants). The study protocol was approved by the Human Research Ethics Board at the University of Victoria (Fig. 6).

Table 1.

Descriptive statistics for demographic data. Mean age and education level in years is displayed (with standard deviation in brackets), with numbers of males and females and handedness of participants in each group.

	Older Adults	Younger Adults	P-Value
Age	70.9 (5.6)	28.1 (2.8)	0.0001
Education Mean	17.2 (3.0)	17.7 (2.2)	0.621
Females: Males	9: 8	9:8
RH: LH	16:1	16:1

Fig. 6.

Probe showing channels that demonstrated significant differences in control (orange), interference (yellow), or both (blue) conditions in older adults. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Measure

The MSIT was administered on a Dell computer screen that presented the task using E-Prime2. Participants were approximately 37 cm away from the screen. The response box had three buttons to correspond to the triggers that participants were reacting to. Participants used their dominant hand and their first three fingers (index, middle, and ring) to complete the task. Participants had one round of practice prior to the beginning of the experiment. In this study, there were 60 control trials and 60 interference trials, with each stimulus presented for 1750 ms. Trials were presented in four blocks with 15 control or 15 interference trials in each block, which were each interspersed with blocks of rest. Participants always started with the control block, and blocks alternated between conditions. After each block, there was a break controlled by the participant to continue. Not including participant-controlled breaks, the task took approximately 7 min to complete (Fig. 7).

Fig. 7.

Probe showing channels that demonstrated significant differences in control (orange), interference (yellow), or both (blue) conditions in younger adults. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Functional Near-Infrared spectroscopy data acquisition

A multichannel continuous-wave TechEn CW6 system (TechEn Inc., Milford, Massachusettes), with a sampling frequency of 50 Hz (corresponding to TR ¼ 20 ms) was used to collect the data. The fNIRS cap comprised 34-source-detector channels and was placed over prefrontal cortex (PFC) as a region of interest specific to the MSIT task. An array of 16 detectors and 10 sources was used to measure wavelengths of 690 nm and 830 nm, corresponding to the measurement of deoxygenated hemoglobin and oxygenated hemoglobin, respectively. The array was arranged in a lattice pattern with a distance of 3 cm between each set of detectors and sources. In order to standardize placement of the apparatus among all participants, the International 10–20 System of Electrode Placement was used [20]. This system includes four main placement indicators: nasion to inion, preauricular points, FPz to Oz, and F7 to F8. The positioning for the probe and region of interest was influenced by previous fNIRS studies that used the same task [9], [10].

Data analyses

Behavioural performance data included accuracy and reaction time (milliseconds). The raw data were processed by E-Prime 2 and exported to Microsoft Excel and SPSS 27 for further analyses. Accuracy and reaction time results were compared both between groups for both conditions, and within groups across conditions. Accuracy is defined as the number of correct responses to the task, correct responses were coded as 1 and incorrect responses were coded as 0. Reaction time referred to the amount of time it took for individuals to response to the stimulus in milliseconds. These comparisons were performed using a 2x2 analysis of variance for reaction time and accuracy, with follow up t-tests employed, as appropriate.

The fNIRS data were preprocessed using Homer 3; an open-source analysis software that performs basic processing of the data [13]. The data were separated by condition and by OA or YA group. Optical density values were converted to hemoglobin concentrations for both oxygenated (HbO) and deoxygenated (HbR) hemoglobin using the modified Beer-Lambert law. A pruning filter was applied to eliminate channels that had excess noise. Due to the pruning filter, some of the subsequent statistical tests excluded participants who had motion artifacts that exceeded an amplitude threshold of 5.0. A wavelet motion corrector was applied using an interquartile range of 1.5. Physiological and electronic noise were corrected by a low pass filter of 0.5 Hz. The data were then block averaged for two seconds before the onset of the first stimulus of a given block, and thirty seconds after, for each condition across both groups of participants. The preprocessed data were then used for image reconstruction [1]. The mean subject hemodynamic response function (HRF) values were exported to Microsoft Excel and SPSS for both groups to perform statistical analyses. The individual participants’ averaged HbO signal for each channel was compared between groups. The channels were compared between groups with independent t-tests. The channels were also assessed within groups to detect active channels (significantly more HbO than HbR) using paired-t tests. The within group comparisons were made for both groups in each condition. Afterwards, we applied a multiple comparison correction to examine the corrected data, this calculation was performed by correcting the uncorrected p-value (0.05) by the number of channels used for each test.

In order to further explore the relationship between prefrontal activity and interference condition performance of the participants, a post-hoc Pearson-r correlation analysis was performed. The region of interest (ROI) for this analysis was the dlPFC, which has been associated with the task network in previous studies (e.g., [17]). Accordingly, two bilateral channels corresponding to the ROI were used to calculate mean HbO for the OA group and YA group (C-3 and F-6). The relationships between the mean HbO values within the ROI and the mean reaction time and accuracy during the interference condition were examined separately for older adults and younger adults. These analyses were corrected for multiple comparisons by taking the uncorrected p-value (0.05) divided by the number of channels (2) multiplied by the number of correlations (2, corresponding to accuracy and reaction time).

Results

Behavioural results

Accuracy. There was a significant main effect of both age (F = 15.59, p < 0.001) and condition (F = 39.42, p < 0.001). There was also a significant interaction effect between age and condition (F = 12.67, p > 0.001). In the control condition, the OA group (M = 0.99, SD = 0.012) and the YA group (M = 0.99, SD = 0.087) were statistically similar in accuracy (p < 0.33). In the interference condition, the OA group (M = 0.90, SD = 0.07) demonstrated lower accuracy than the YA group (M = 0.97, SD = 0.02; p < 0.0006).

Reaction Time. There was a main effect of both age (F = 51.61, p < 0.001) and condition (F = 124.52, p < 0.001). In the control condition, the OA group (M = 768.04, SD = 166.40) was significantly slower than YA group (M = 545.63, SD = 131.56, p < 0.001). Similarly, in the interference condition, the OA group (M = 1114.50, SD = 90.40) performed significantly slower than the YA group (M = 891.31, SD = 110.62, p < 0.001). There was not a significant interaction between age and condition (p = 0.99).

Functional results

A pruning filter was used to exclude participants with excess motion. The number of participants included in each test are outlined in each table.

Between-Group Control Condition. The OA group demonstrated significantly greater activation in comparison to the YA group in the following channels: G-15, H-15, H-16, and I-16, D-12, E-4, E-5, E-12, F-5, and J-12 localized over the left hemisphere and anterior prefrontal cortex. The full results for all channels are displayed in Table 2.

Table 2.

Mean and standard deviations for YA and OA oxygenated hemoglobin averages by channel for control condition. Results from the independent t-test compared between groups including: number of participants included in each test, t-value, degrees of freedom, obtained p value, effect size, and confidence intervals from each test. The uncorrected significant channels are denoted with an asterix. Channels that remained significant after a correction for multiple comparisons are denoted with a double asterix.

Channel	OA Mean	OA SD	YA Mean	YA SD	N-test	t-value	df	p	Cohen’s D	95% CI L	95% CI U
A-1	0.000022	0.00005	0.000015	0.00005	34	0.421	32	0.676	0.145	−0.530	0.817
A-9	0.00003	0.00006	0.000017	0.00006	33	0.584	31	0.564	0.203	−0.483	0.886
B-1	0.000029	0.00007	0.000027	0.00005	34	0.107	32	0.915	0.037	−0.636	0.709
B-2	0.000021	0.00006	0.000018	0.00004	34	0.149	32	0.883	0.051	−0.622	0.723
B-9	0.000029	0.00004	0.000012	0.00004	34	1.187	32	0.224	0.407	−0.276	1.084
B-10	0.000025	0.000053	0.00001	0.000042	34	0.878	32	0.387	0.301	−0.378	0.975
C-2	0.000024	0.000059	0.00001	0.000042	34	0.791	32	0.435	0.271	−0.406	0.945
C-3	0.000026	0.00005	0.000011	0.00004	34	1.017	32	0.317	0.349	−0.332	1.024
C-10	0.000029	0.00006	0.000007	0.00005	34	1.155	32	0.257	0.396	−0.286	1.072
C-11	0.000032	0.00004	0.000009	0.00003	34	1.045	32	0.304	0.359	−0.322	1.034
D-3	0.000032	0.00004	0.000014	0.00003	34	1.321	32	0.196	0.453	−0.232	1.131
D-4	0.000034	0.00004	0.000009	0.00003	34	1.94	32	0.061	0.666	−0.031	1.352
D-11	0.000021	0.00003	0.000008	0.00003	34	1.173	32	0.249	0.402	−0.280	1.079
D-12*	0.000037	0.00004	0.000004	0.00003	34	2.696	32	0.011	0.925	0.209	1.628
E-4*	0.000036	0.00003	0.000004	0.00004	34	2.535	32	0.016	0.870	0.159	1.568
E-5*	0.000034	0.00003	−0.000002	0.00004	34	2.982	32	0.005	1.023	0.299	1.733
E-12*	0.000034	0.00003	0.000002	0.00004	34	2.493	32	0.018	0.855	0.145	1.553
E-13	0.00001	0.00009	0.000005	0.00004	34	0.238	32	0.813	0.082	−0.592	0.754
F-5*	0.000026	0.00004	−0.000002	0.00004	34	2.158	32	0.039	0.740	0.039	1.431
F-6	0.000023	0.000047	0.0000001	0.000048	34	1.318	32	0.197	0.452	−0.233	1.130
F-13	0.000005	0.00009	0.000002	0.00005	33	0.15	31	0.882	0.052	−0.631	0.735
F-14	−0.000038	0.00025	−0.00000047	0.00004	33	−0.594	31	0.557	−0.207	−0.890	0.480
G-6	0.000023	0.00007	−0.0000000087	0.00005	33	1.126	31	0.269	0.392	−0.300	1.079
G-7	0.000034	0.00006	0.000018	0.00005	32	0.829	30	0.414	0.293	−0.406	0.988
G-14	−0.000051	0.0003	−0.000002	0.00003	34	−0.656	32	0.516	−0.225	−0.898	0.451
G-15*	0.000032	0.00004	0.0000005	0.00003	34	2.479	32	0.019	0.850	0.141	1.548
H-7	0.000033	0.00005	0.000022	0.00004	34	0.716	32	0.479	0.246	−0.431	0.919
H-8	0.000033	0.00005	0.000027	0.00004	33	0.359	31	0.722	0.125	−0.559	0.807
H-15*	0.000037	0.00003	0.000007	0.00003	34	2.554	32	0.016	0.876	0.164	1.575
H-16*	0.000037	0.00003	0.000012	0.00003	34	2.313	32	0.027	0.793	0.088	1.487
I-8	0.000028	0.00005	0.00001	0.00003	34	1.315	32	0.198	0.451	−0.234	1.129
I-16*	0.000044	0.00004	0.000009	0.00004	32	2.476	30	0.019	0.877	0.142	1.599
J-12*	0.000038	0.00003	0.00000012	0.00004	34	2.984	32	0.005	1.023	0.299	1.734
J-13	0.000027	0.00005	0.000009	0.00004	34	1.072	32	0.292	0.368	−0.313	1.043

Between-Group Interference Condition. Similarly, to the control condition, the OA group demonstrated significantly greater activation in comparison to the YA group in the following channels: D-3, D-4, D-11, D-12, E-4, and J-12 localized over the anterior prefrontal cortex, with a slight orientation towards the right hemisphere. The full results for all channels are displayed in Table 3.

Table 3.

Mean and standard deviations for YA and OA oxygenated hemoglobin averages by channel for interference condition. Results from the independent t-test compared between groups including: number of participants included in each test, t-value, degrees of freedom, obtained p value, effect size, and confidence intervals from each test. The uncorrected significant channels are denoted with an asterix. Channels that remained significant after a correction for multiple comparisons are denoted with a double asterix.

Channel	OA Mean	OA SD	YA Mean	YA SD	N-test	t-value	df	p	Cohen’s D	95% CI L	95% CI U
A-1	0.000013	0.00005	−0.000001	0.000007	34	0.741	32	0.464	0.254	−0.423	0.927
A-9	0.00003	0.00007	0.000006	0.00006	33	1.076	31	0.29	0.375	−0.317	1.061
B-1	0.000031	0.00007	0.000027	0.00005	34	0.16	32	0.874	0.055	−0.618	0.727
B-2	0.000029	0.00006	0.000026	0.00004	34	0.205	32	0.839	0.070	−0.603	0.742
B-9	0.000035	0.00004	0.000023	0.00004	34	0.828	32	0.414	0.284	−0.394	0.958
B-10	0.000022	0.00004	0.000015	0.00004	34	0.526	32	0.602	0.181	−0.495	0.853
C-2	0.000037	0.00006	0.000015	0.00005	34	1.105	32	0.278	0.379	−0.303	1.055
C-3	0.00004	0.00006	0.000009	0.00005	34	1.632	32	0.113	0.560	−0.130	1.242
C-10	0.000033	0.00006	0.000001	0.00005	34	1.724	32	0.094	0.591	−0.101	1.275
C-11	0.000017	0.00007	0.0000005	0.00004	34	0.879	32	0.386	0.302	−0.377	0.975
D-3*	0.000045	0.00005	0.000008	0.00005	34	2.15	32	0.039	0.737	0.036	1.428
D-4*	0.00004	0.00006	0.000001	0.00004	34	2.259	32	0.031	0.775	0.071	1.468
D-11*	0.00003	0.00004	0.000003	0.00004	34	2.07	32	0.047	0.710	0.011	1.399
D-12*	0.000044	0.00006	0.000007	0.00004	34	2.164	32	0.038	0.742	0.041	1.433
E-4*	0.000041	0.00005	0.000008	0.00004	34	2.156	32	0.039	0.740	0.038	1.430
E-5	0.000039	0.00005	0.000009	0.00004	34	1.956	32	0.059	0.671	−0.026	1.358
E-12	0.000042	0.00005	0.00001	0.00004	34	1.981	32	0.056	0.679	−0.018	1.367
E-13	0.000000077	0.00016	0.000014	0.00004	34	−0.346	32	0.732	−0.119	−0.791	0.555
F-5	0.000031	0.00004	0.000007	0.00005	34	1.625	32	0.114	0.557	−0.133	1.239
F-6	0.000033	0.00005	0.00007	0.00005	34	1.576	32	0.125	0.540	−0.149	1.221
F-13	−0.000003	0.00015	−0.000003	0.00004	33	−0.36	31	0.722	−0.125	−0.808	0.559
F-14	−0.000043	0.00029	0.000002	0.00003	33	−0.622	31	0.539	−0.217	−0.900	0.470
G-6	0.000033	0.00006	0.00000084	0.00005	33	1.595	31	0.121	0.556	−0.145	1.248
G-7	0.00004	0.00007	0.000006	0.00004	32	1.838	30	0.076	0.650	−0.067	1.357
G-14	−0.000056	0.00033	0.000004	0.00004	34	−0.748	32	0.46	−0.257	−0.930	0.420
G-15	0.00003	0.00004	0.000004	0.00003	34	1.971	32	0.057	0.676	−0.021	1.363
H-7	0.000035	0.00006	0.000025	0.00004	34	0.575	32	0.569	0.197	−0.478	0.870
H-8	0.00003	0.00005	0.000022	0.00005	33	0.442	31	0.661	0.154	−0.531	0.837
H-15	0.000039	0.00004	0.000013	0.00004	34	1.773	32	0.086	0.608	−0.085	1.292
H-16	0.000038	0.00004	0.000018	0.00004	34	1.397	32	0.172	0.479	−0.207	1.158
I-8	0.000024	0.00005	0.000017	0.00004	34	0.442	32	0.661	0.152	−0.523	0.824
I-16	0.000046	0.00005	0.000027	0.00004	32	1.127	30	0.269	0.399	−0.306	1.098
J-12*	0.000045	0.00005	0.000014	0.00004	34	2.106	32	0.043	0.722	0.022	1.412
J-13	0.000024	0.00008	0.000018	0.00004	34	0.251	32	0.803	0.086	−0.587	0.758

Within-Group Older Adults. The OA group demonstrated several active channels (i.e., where HbO was significantly greater than HbR) in both task conditions, including channels recording specifically over a key task region of interest (dlPFC): C-3 (t = 2.67, p < 0.0017) and F-6 (t = 2.32, p < 0.034). Several other channels also demonstrated significant activation across the anterior prefrontal cortex, including: A-1, A-9, B-1, B-9, C-2, C-11, D-3, D-4, D-11, D-12, E-4, E-5, E-12, F-5, G-7, G-15, H-7, H-8, H-15, H-16, J-12, J-13, I-8, and I-16. See Table 4 for full results.

Table 4.

Mean and standard deviations for OA oxygenated and deoxygenated hemoglobin averages by channel for the control condition. Results from the paired t-test compared within groups include: number of participants included in each test, t-value, degrees of freedom, obtained p value, effect size, and confidence intervals from each test. The uncorrected significant channels are denoted with an asterix. Channels that remained significant after a correction for multiple comparisons are denoted with a double asterix.

Channel	HbO Mean	HbO SD	HbR Mean	HbR SD	N-test	t-value	df	p	Cohen’s D	95% CI L	95% CI U
A-1*	0.0002	0.00005	−0.000004	0.00001	17	2.144	16	0.048	0.520	0.005	1.021
A-9*	0.00003	0.00006	−0.00001	0.00001	17	2.368	16	0.031	0.574	0.052	1.081
B-1*	0.00003	0.00007	−0.00001	0.00002	17	2.265	16	0.038	0.549	0.030	1.053
B-2	0.00002	0.00006	−0.00001	0.00001	17	2.091	16	0.053	0.507	−0.006	1.006
B-9*	0.00003	0.00004	−0.000004	0.0001	17	2.86	16	0.011	0.694	0.153	1.217
B-10*	0.00002	0.0005	−0.000003	0.0002	17	2.204	16	0.043	0.534	0.018	1.037
C-2*	0.00002	0.00006	−0.00001	0.00001	17	2.422	16	0.028	0.587	0.063	1.096
C-3*	0.00003	0.00005	−0.00001	0.00001	17	2.671	16	0.017	0.648	0.115	1.164
C-10	0.00003	0.00006	−0.00001	0.00002	17	1.982	16	0.065	0.481	−0.029	0.977
C-11*	0.00002	0.00004	−0.00001	0.00002	17	2.603	16	0.019	0.631	0.101	1.146
D-3*	0.00003	0.00004	−0.00001	0.00001	17	3.407	16	0.004	0.826	0.263	1.370
D-4**	0.00003	0.00004	−0.00001	0.0001	17	3.88	16	0.001	0.941	0.356	1.506
D-11**	0.00002	0.00003	−0.00001	0.00002	17	4.044	16	<0.001	0.981	0.388	1.553
D-12**	0.00004	0.00004	−0.00001	0.00001	17	4.507	16	<0.001	1.093	0.477	1.688
E-4**	0.00004	0.00003	−0.00001	0.00001	17	4.675	16	<0.001	1.134	0.509	1.738
E-5**	0.00003	0.00003	−0.00001	0.00001	17	4.137	16	<0.001	1.003	0.406	1.580
E-12**	0.00003	0.00003	−0.00001	0.00001	17	4.254	16	<0.001	1.032	0.428	1.614
E-13	0.00001	0.00009	−0.00001	0.00003	17	1.109	16	0.248	0.269	−0.219	0.749
F-5*	0.00003	0.00004	−0.00001	0.00001	17	3.189	16	0.006	0.773	0.220	1.309
F-6*	0.00002	0.00005	−0.00001	0.00001	17	2.324	16	0.034	0.564	0.043	1.069
F-13	0.00001	0.00009	−0.00002	0.00004	17	1.189	16	0.086	0.444	−0.061	0.937
F-14	−0.00004	0.00025	−0.00003	0.0001	17	−0.197	16	0.846	−0.048	−0.523	0.429
G-6	0.00002	0.00007	−0.00001	0.00001	16	1.941	15	0.071	0.485	−0.041	0.998
G-7*	0.00003	0.00006	−0.00001	0.000001	16	2.89	15	0.011	0.723	0.160	1.266
G-14	−0.00005	0.0003	−0.00004	0.00013	17	−0.35	16	0.731	−0.085	−0.560	0.393
G-15*	0.00003	0.00004	−0.00005	0.00001	17	3.611	16	0.002	0.876	0.304	1.429
H-7*	0.00003	0.00005	−0.00001	0.00002	17	3.218	16	0.005	0.781	0.226	1.317
H-8*	0.00003	0.0005	−0.00001	0.0001	17	3.301	16	0.005	0.801	0.242	1.340
H-15**	0.00004	0.00003	−0.00001	0.00001	17	4.769	16	<0.001	1.157	0.526	1.765
H-16**	0.00004	0.00003	−0.00001	0.00001	17	4.461	16	<0.001	1.126	0.502	1.728
I-8*	0.00003	0.00005	−0.000003	0.00002	17	2.991	16	0.009	0.726	0.180	1.253
I-16**	0.00004	0.00004	−0.00001	0.00001	15	4.961	14	<0.001	1.281	0.578	1.959
J-12**	0.00004	0.00003	−0.00001	0.00001	17	4.932	16	<0.001	1.196	0.557	1.814
J-13**	0.00003	0.00005	−0.00001	0.00004	17	4.442	16	<0.001	1.077	0.464	1.669

The OA group demonstrated fewer active channels in the interference condition than the control condition. Again, both channels associated with the task network demonstrated significant activation (C-3, t = 3.44, p < 0.003; F-6, t = 3.09, p < 0.007). While twenty-three other channels across the anterior prefrontal cortex did demonstrate significant activation (B-1, B-2, B-9, C-2, C-10, D-3, D-4, D-11, D-12, E-4, E-5, E-12, F-5, G-6, G-7, G-15, H-7, H-8, H-15, H-16, J-12, J-13, and I-16). See Table 5 for full results.

Table 5.

Mean and standard deviations for OA oxygenated and deoxygenated hemoglobin averages by channel for the interference condition. Results from the paired t-test compared within groups include: number of participants included in each test, t-value, degrees of freedom, obtained p value, effect size, and confidence intervals from each test. The uncorrected significant channels are denoted with an asterix. Channels that remained significant after a correction for multiple comparisons are denoted with a double asterix.

Channel	HbO Mean	HbO SD	HbR Mean	HbR SD	N-test	t-value	df	p	Cohen’s D	95% CI L	95% CI U
A-1	0.000013	0.00005	−0.000007	0.00002	17	1.563	16	0.138	0.379	−0.119	0.866
A-9	0.00003	0.00007	−0.000013	0.00002	17	2.001	16	0.063	0.485	−0.025	0.982
B-1*	0.000031	0.00007	−0.000012	0.00002	17	2.273	16	0.037	0.551	0.032	1.056
B-2*	0.000029	0.00006	−0.000005	0.00002	17	2.176	16	0.045	0.528	0.012	1.029
B-9*	0.000035	0.00004	−0.000001	0.00001	17	2.918	16	0.01	0.708	0.165	1.233
B-10	0.000022	0.00004	−0.000002	0.00001	17	2.015	16	0.061	0.489	−0.022	0.986
C-2*	0.000037	0.00006	−0.000011	0.00001	17	3.01	16	0.008	0.730	0.184	1.259
C-3*	0.00004	0.00006	−0.00001	0.00001	17	3.44	16	0.003	0.834	0.270	1.380
C-10*	0.000033	0.00006	−0.0000022	0.00001	17	2.638	16	0.018	0.640	0.108	1.155
C-11	0.0000017	0.00007	−0.0000008	0.00002	17	1.896	16	0.076	0.460	−0.048	0.954
D-3**	0.000045	0.00005	−0.0000009	0.00001	17	3.846	16	0.001	0.933	0.350	1.496
D-4*	0.00004	0.00006	−0.000008	0.00001	17	3.246	16	0.005	0.787	0.231	1.325
D-11*	0.00003	0.00004	−0.0000015	0.00004	17	3.638	16	0.002	0.893	0.318	1.449
D-12*	0.000044	0.00006	−0.0000066	0.00001	17	3.382	16	0.004	0.820	0.258	1.364
E-4*	0.00004	0.00005	−0.000003	0.00001	17	3.692	16	0.002	0.880	0.307	1.434
E-5*	0.00004	0.00005	−0.0000024	0.00001	17	3.578	16	0.003	0.868	0.297	1.419
E-12*	0.00004	0.00005	−0.000005	0.00001	17	3.644	16	0.002	0.884	0.310	1.438
E-13	7.6E-08	0.00016	−0.000012	0.00004	17	0.417	16	0.682	0.101	−0.377	0.576
F-5*	0.00003	0.00004	−0.0000038	0.00001	17	3.175	16	0.006	0.770	0.217	1.305
F-6*	0.000033	0.00005	−0.000006	0.00001	17	3.087	16	0.007	0.749	0.199	1.280
F-13	−0.0000027	0.00015	−0.000002	0.00005	17	0.59	16	0.564	0.143	−0.337	0.619
F-14	−0.00004	0.00029	−0.00005	0.0002	17	0.49	16	0.631	0.119	−0.360	0.594
G-6**	0.000033	0.00006	−0.000011	0.00001	16	2.765	15	0.014	0.691	0.134	1.230
G-7*	0.00004	0.00007	−0.00001	0.00001	16	3.324	15	0.005	0.831	0.249	1.393
G-14	−0.00005	0.00033	−0.00004	0.00016	17	−0.345	16	0.735	−0.084	−0.559	0.394
G-15*	0.00003	0.00004	−0.000004	0.00001	17	3.118	16	0.007	0.756	0.205	1.289
H-7*	0.000035	0.00006	−0.000008	0.00001	17	2.978	16	0.009	0.722	0.177	1.250
H-8*	0.00003	0.00005	−0.0000099	0.00002	17	2.825	16	0.012	0.685	0.146	1.207
H-15**	0.000039	0.00004	−8.9E-07	0.00001	17	3.196	16	0.001	0.950	0.363	1.516
H-16*	0.000038	0.00004	−0.000006	0.00001	17	3.724	16	0.002	0.903	0.326	1.461
I-8	0.000024	0.00005	0.0000002	0.00001	17	1.873	16	0.079	0.454	−0.052	0.948
I-16*	0.000046	0.00005	−0.0000096	0.00002	15	3.465	14	0.004	0.895	0.280	1.487
J-12**	0.000045	0.00005	−0.000006	0.00001	17	4.353	16	<0.001	1.056	0.447	1.643
J-13**	0.000023	0.00008	−0.000002	0.00006	17	4.048	16	<0.001	0.982	0.389	1.554

Within-Group Younger Adults. Within younger adults, there were three channels across the anterior prefrontal cortex (B-1, D-3, H-8) that did demonstrate activation (p < 0.05). See Table 6 for full results.

Table 6.

Mean and standard deviations for YA oxygenated and deoxygenated hemoglobin averages by channel for the control condition. Results from the paired t-test compared within groups include: number of participants included in each test, t-value, degrees of freedom, obtained p value, effect size, and confidence intervals from each test. The uncorrected significant channels are denoted with an asterix. Channels that remained significant after a correction for multiple comparisons are denoted with a double asterix.

Channel	HbO Mean	HbO SD	HbR Mean	HbR SD	N-test	t-value	df	p	Cohen’s D	95% CI L	95% CI U
A-1	0.000015	0.00005	0.000009	0.00004	17	0.317	16	0.756	0.077	−0.401	0.552
A-9	0.000017	0.00006	0.000005	0.00003	16	0.588	15	0.566	0.147	−0.348	0.637
B-1*	0.000027	0.00005	−0.000002	0.00002	17	2.274	16	0.037	0.552	0.032	1.056
B-2	0.000018	0.00004	−0.000002	0.00001	17	1.918	16	0.073	0.465	−0.043	0.960
B-9	0.000012	0.00004	−0.000004	0.00001	17	1.477	16	0.159	0.358	−0.138	0.844
B-10	0.000001	0.00004	−0.000009	0.00001	17	1.923	16	0.072	0.466	−0.042	0.962
C-2	0.00001	0.00004	−0.000008	0.00001	17	1.595	16	0.13	0.387	−0.112	0.875
C-3	0.000011	0.00004	−0.000006	0.00001	17	1.581	16	0.133	0.383	−0.115	0.871
C-10	0.000007	0.00005	−0.000008	0.00002	17	1.443	16	0.168	0.350	−0.146	0.835
C-11	0.000003	0.00004	−0.000003	0.00002	17	0.536	16	0.599	0.130	−0.349	0.605
D-3*	0.000014	0.00003	−0.000006	0.00001	17	2.219	16	0.041	0.538	0.021	1.041
D-4	0.000009	0.00003	−0.000003	0.00001	17	1.381	16	0.186	0.335	−0.159	0.819
D-11	0.000008	0.00003	−0.000008	0.00001	17	1.816	16	0.088	0.440	−0.065	0.933
D-12	0.000004	0.00003	−0.000009	0.00001	17	1.322	16	0.205	0.321	−0.172	0.804
E-4	0.000004	0.00004	0.0000001	0.00001	17	0.324	16	0.75	0.079	−0.399	0.554
E-5	−0.000002	0.00004	−0.000002	0.00001	17	−0.026	16	0.98	−0.006	−0.481	0.469
E-12	0.000002	0.00004	0.000011	0.00002	17	0.91	16	0.376	0.221	−0.264	0.699
E-13	0.000005	0.00004	−0.000005	0.00001	17	9.25	16	0.369	0.224	−0.261	0.703
F-5	−0.000002	0.00004	−0.000004	0.00001	17	0.266	16	0.749	0.064	−0.412	0.539
F-6	0.000001	0.00005	−0.000005	0.00001	17	0.505	16	0.621	0.122	−0.357	0.598
F-13	0.0000001	0.00005	−0.000003	0.00001	16	0.473	15	0.643	0.118	−0.376	0.608
F-14	0.000018	0.00004	−0.000002	0.00001	16	0.415	15	0.684	0.104	−0.389	0.593
G-6	−8E-08	0.00005	−0.000003	0.00002	17	0.243	16	0.811	0.059	−0.418	0.534
G-7	0.000018	0.00005	−0.000002	0.00002	16	1.425	15	0.175	0.356	−0.155	0.857
G-14	−0.000002	0.00003	−0.000003	0.00002	17	0.121	16	0.905	0.029	−0.447	0.504
G-15	0.00000005	0.00003	−0.000007	0.00002	17	0.834	16	0.417	0.202	−0.281	0.680
H-7	0.000022	0.00004	−0.000002	0.00001	17	1.986	16	0.064	0.482	−0.028	0.978
H-8*	0.000027	0.000004	−0.000006	0.000002	16	2.53	15	0.023	0.632	0.085	1.163
H-15	0.000007	0.00003	−0.000008	0.00002	17	1.57	16	0.136	0.381	−0.118	0.868
H-16	0.000012	0.00003	−0.000004	0.00001	17	1.77	16	0.096	0.429	−0.075	0.921
I-8	0.00001	0.00003	0.000005	0.00002	17	0.506	16	0.62	0.123	−0.356	0.598
I-16	0.000009	0.00004	0.000004	0.00003	17	0.286	16	0.779	0.069	−0.408	0.544
J-12	0.0000001	0.00004	−0.0000005	0.00001	17	0.414	16	0.685	0.100	−0.378	0.575
J-13	0.000009	0.00004	−0.000006	0.00001	17	1.356	16	0.194	0.329	−0.165	0.813

The YA group demonstrated more active channels in the interference condition than the control condition. Again, none of the channels associated with the task region demonstrated significant activation. However, only ten other channels across the probe array demonstrated significant activation (B-1, B-2, B-9, B-10, H-7, H-15, H-16, J-12, J-13, and I-16). See Table 7 for full results.

Table 7.

Mean and standard deviations for YA oxygenated and deoxygenated hemoglobin averages by channel for the interference condition. Results from the paired t-test compared within groups include: number of participants included in each test, t-value, degrees of freedom, obtained p value, effect size, and confidence intervals from each test. The uncorrected significant channels are denoted with an asterix. Channels that remained significant after a correction for multiple comparisons are denoted with a double asterix.

Channel	HbO Mean	HbO SD	HbR Mean	HbR SD	t-value	N-test	df	p	Cohen’s D	95% CI L	95% CI U
A-1	−0.000001	0.00007	0.000008	0.00003	−0.414	17	16	0.684	−0.100	−0.575	0.378
A-9	0.000006	0.00006	0.000003	0.00004	0.159	16	15	0.876	0.040	−0.451	0.529
B-1*	0.000027	0.00005	−0.000009	0.00002	2.5	17	16	0.024	0.606	0.079	1.118
B-2*	0.000026	0.00004	−0.000007	0.00001	3.276	17	16	0.005	0.795	0.237	1.333
B-9*	0.000023	0.00004	−0.000009	0.00001	3.109	17	16	0.007	0.754	0.204	1.286
B-10*	0.000015	0.00004	−0.000013	0.00001	2.582	17	16	0.02	0.626	0.096	1.140
C-2	0.000015	0.00005	−0.000008	0.00002	1.649	17	16	0.119	0.400	−0.101	0.889
C-3	0.000009	0.00005	−0.000008	0.00002	1.236	17	16	0.234	0.300	−0.191	0.782
C-10	0.00001	0.00005	−0.000009	0.00002	1.016	17	16	0.325	0.246	−0.240	0.726
C-11	0.000001	0.00004	−0.000005	0.00001	0.644	17	16	0.529	0.156	−0.325	0.632
D-3	0.000008	0.00005	−0.000007	0.00001	1.19	17	16	0.251	0.289	−0.201	0.770
D-4	0.000001	0.00004	−0.000004	0.00001	0.479	17	16	0.638	0.116	−0.363	0.591
D-11	0.000003	0.00004	−0.000004	0.00001	0.651	17	16	0.524	0.158	−0.323	0.634
D-12	0.000007	0.00004	−0.000006	0.00001	1.242	17	16	0.232	0.301	−0.190	0.783
E-4	0.000008	0.00004	−0.000005	0.00001	1.153	17	16	0.266	0.280	−0.209	0.761
E-5	0.000009	0.00004	−0.000005	0.00001	1.225	17	16	0.238	0.297	−0.193	0.779
E-12	0.00001	0.00004	−0.000006	0.00001	3.99	17	16	0.181	0.339	−0.155	0.824
E-13	0.000014	0.00004	−0.000005	0.00001	1.677	17	16	0.113	0.407	−0.095	0.896
F-5	0.000007	0.00005	−0.000005	0.00001	0.975	17	16	0.344	0.236	−0.249	0.715
F-6	0.000007	0.00005	−0.000006	0.00001	0.941	17	16	0.361	0.228	−0.257	0.706
F-13	0.000011	0.00004	−0.00001	0.00002	1.696	16	15	0.111	0.424	−0.095	0.930
F-14	0.000002	0.00003	−0.000005	0.00001	0.66	16	15	0.519	0.165	−0.331	0.656
G-6	0.000001	0.00005	−0.000006	0.00001	0.446	17	16	0.661	0.108	−0.370	0.583
G-7	0.000006	0.00004	−0.000003	0.00001	0.773	16	15	0.452	0.193	−0.305	0.685
G-14	0.000004	0.00004	−0.000004	0.00001	0.928	17	16	0.367	0.225	−0.260	0.703
G-15	0.000004	0.00003	−0.000008	0.00002	1.296	17	16	0.213	0.314	−0.178	0.797
H-7*	0.000025	0.00004	−0.000004	0.00001	2.481	17	16	0.025	0.602	0.075	1.112
H-8	0.000022	0.00005	−0.000011	0.00002	2.044	16	15	0.059	0.511	−0.019	1.026
H-15*	0.000013	0.00004	−0.000009	0.00001	2.125	17	16	0.049	0.515	0.001	1.016
H-16*	0.000018	0.00004	−0.0000005	0.00001	2.536	17	16	0.022	0.615	0.087	1.127
I-8	0.000017	0.00004	−0.000001	0.00002	1.514	17	16	0.149	0.367	−0.130	0.854
I-16*	0.000027	0.00004	−0.000005	0.00002	2.481	17	16	0.025	0.602	0.075	1.112
J-12*	0.000014	0.00004	−0.000009	0.00001	2.356	17	16	0.032	0.571	0.050	1.078
J-13*	0.000018	0.00004	−0.000009	0.00001	2.51	17	16	0.023	0.609	0.081	1.120

Post hoc correlation analysis

There was a significant positive correlation between the average HbO (M = 0.000036) and reaction time (M = 1114.50) for the OA interference condition (r = 0.615, p < 0.009). However, there was no significant correlation between the average HbO and accuracy (p = 0.291).

There were no significant correlations between the average HbO and reaction time (p = 0.753) or accuracy (p = 0.170) for the YA interference condition.

Corrected data

While the data were initially computed without a correction for multiple comparisons, follow up analyses were completed to correct for multiple comparisons, with the results presented below.

Between-Groups. There were no channels that remained significant after correction for multiple corrections in the control or interference conditions.

Within-Groups Older Adults. In the control condition, the following channels across the prefrontal cortex were significant after correction: D-4, D-11, D-12, E-4, E-5, E-12, H-15, H-16, I-16, J-12, and J-13. In the interference condition, the following channels across the prefrontal cortex remained significant: D-3, G-6, H-15, J-12, J-13.

Within-Groups Younger Adults. There were no channels that remained significant after a comparison for multiple corrections in the control or interference conditions.

Discussion

The purpose of the current study was to examine the neurophysiological and behavioural differences between healthy older and younger adults on the MSIT using fNIRS. We aimed to answer the following questions.

Question 1: How do older and younger adults differ in their behavioural performance of the MSIT?

With respect to the first aim, as expected, both older and younger adults demonstrated faster and more accurate performance in the control condition than the interference condition on the MSIT. These within-group findings effectively replicated the protocol paper for the MSIT [5] and other fMRI studies [3], [17]and fNIRS literature [12], [9], [10]. The decline in performance during the interference condition is consistent with the increase in task difficulty and indicates the additional cognitive processing time and resources required in order to successfully complete the task. Moreover, we also hypothesized that older adults would have lower accuracy and slower reaction times than younger adults in the interference condition, as well as slower reaction time during the control condition, but with no group differences in accuracy. In accordance with these age-related hypotheses, the behavioural results demonstrated that older adults were slower and less accurate than younger adults on the interference condition. Older adults also had slower reaction times in the control condition compared to younger adults, however there were no significant differences between the accuracy of the groups in either condition. The mean reaction times in the current study differed from those reported by Bush and Shin [5]. Notably, the MSIT was originally designed for fMRI research and prior research has been conducted in these conditions (although Bush and Shin [5]note that the task can be readily adapted for other imaging methods or pure behavioural data collection). However, some research suggests that the MRI scanner environment can create meaningful differences in performance compared to non-scanner environments [8]. For example, participants are typically laying in a supine position within a scanner and are exposed to the noise of the scanner, whereas in the current study, participants were sitting upright in an office environment. There were also some small variations between the original and current task protocols, including a smaller number of trials in the current study (60 trials of interference and control vs 96 of each in the original protocol). These differences in task administration could account for some of the differences between the range of reaction time values, particularly considering the older aged participants in the current sample.

Essentially, in the current study, older adults were able to maintain accuracy, which was comparable to younger adults during the low-demand control condition. There was a relatively small, but significant, decrease in accuracy between the control and interference conditions in the OA group. These findings are consistent with previous studies; for example, Salami et al. [17]also found a significant, but small, decrease in older aged participants accuracy between conditions. However, during the high-demand interference condition, older adults declined in accuracy and reaction time relative to younger adults. It is worth noting that the differences between-condition reaction times were identical for both OA and YA groups, suggesting that older adults may be experiencing slowed cognitive processing, overall. However, OA performed at lower accuracy than YA during the interference condition, but not the control condition. Taken together, these behavioural findings are consistent with the notion of increased proneness to interference with age, put forth by Salami et al. [17]who had very similar behavioural findings with the MSIT task. Previous fMRI studies have shown that older adults have consistently slower reaction times than younger adults in both the control and interference conditions [3], [17]. Together, the current findings are in line with the impact of normal aging on cognition, which describes a decline in processing speed, among other areas of cognition, which is associated with a decline in behavioural performance (eg., accuracy and response time) [11].

Question 2: How do older and younger adults differ in their functional activation during the MSIT

Regarding the second research question, as anticipated, significant age-related differences were detected across both conditions. Specifically, the older adults demonstrated significantly more oxygenated hemoglobin than the younger adults in ten channels spanning the central prefrontal cortex and left dorsal lateral prefrontal cortex. These results are discrepant from previous age-comparative fMRI studies: for instance, Salami and colleagues [17] did not find any significant increases or decreases in activation during the control condition of the MSIT between younger and older adults, despite having a similar mean age for each age group. It is possible that differences in participant characteristics may account for some differences in the observed findings. For example, the younger adult group in the current study ranged in age from 25 to 35 years and the older adult group ranged in age from 65 to 82 years; both of these groups had approximately 17 years of education on average. In comparison, the participants in the study by Salami et al. [17] had a younger group ranging from 20 to 31 years of age and an older group that was 65–74 years old, each with approximately 14 years of education. Such differences in age and education level may account for some of the differences in findings between these studies. Nevertheless, the additional recruitment of brain areas in order for older adults to perform at comparable levels to younger adults, aligns with many of the existing cognitive theories of aging (e.g. Compensation Related Utilization of Neural Circuits Hypothesis (CRUNCH), Scaffolding Theory of Aging and Cognition revised (STAC-r), and Hemispheric asymmetry Reduced in OLD (HAROLD)). In particular, CRUNCH focuses on the prefrontal regions of the brain and posits an age-related increase in activation at lower demand tasks [15]. The STAC-r model also indicates that recruitment of additional brain regions may occur to scaffold cognitive performance (Reuter-Lornez and Park, 2014). Finally, the recruitment of areas outside of the task network suggests that there is a dedifferentiation effect from older adults in order to compensate for task performance, which is consistent with HAROLD [6]. Overall, our findings suggests that younger adults require less neural activity than older adults to perform the same control task with the same accuracy and speed, supporting the hypothesis of cognitive compensation [2].

In the interference condition, our results also demonstrated significantly greater activation in older adults in comparison to younger adults. In particular, there were six channels spanning the central and right anterior prefrontal cortex that demonstrated significant age-related increases in oxygenated hemoglobin. These findings are consistent with previous fMRI literature that has also detected age-related increases in activation. Both Avelar-Pereira et al. [3] and Salami and colleagues [17] found increased activation during the interference condition of the MSIT. Moreover, Salami and colleagues [17] found bilateral increases in activation in the dlPFC and ACC, with age-related over recruitment in older adults in comparison to younger adults. The current study partially replicated these results but did not see bilateral increases in activation or increases in the anterior cingulate cortex. Despite the similarities in theory and underpinnings of the measurements, the differences between these two methods could have led to disparate results. Specifically, fNIRS has limited spatial resolution and depth of measurement, and is not able to measure activation in the anterior cingulate cortex, for example. Furthermore, the data analysis techniques are unique for each method, which could have led to differences in the results.

Although the increased activation in older adults was expected, contrary to our hypotheses, older adults also had a larger extent of activation in the control condition than younger adults that was more robust than the group differences observed in the interference condition. Taken together with the behavioural findings, these results suggest that the compensation mechanisms employed by older adults in the control condition are less effective in the interference condition. In other words, older adults are employing compensation mechanisms that allow them to compensate for functional inefficiencies at lower task levels, but not as effectively at higher task levels. These findings are in line with previous comparative aging fMRI literature [3], [17]and the aforementioned models of aging (CRUNCH, STAC-r, HAROLD). In particular, the CRUNCH model (Ruter-Lornez and Cappell, 2008) suggests that over-activation at lower task levels is a compensatory mechanism for inefficiency of performance, but that after a certain threshold is reached, the systems exhibit a decline in activation, demonstrating a failure to compensate or support the task load.

The current study also re-examined the results after correcting for multiple comparisons. Although these corrections are not yet common place within the fNIRS literature, they are becoming more common and the intention of presenting these results is to serve as a comparison for future studies. Although there were no significant differences between the OA and YA group or within the YA group, an interesting finding emerged within the OA group. Specifically, the OA group maintained significantly higher oxygenated hemoglobin levels across the prefrontal cortex in both the control and interference conditions. These findings underscore that older adults require significantly more oxygenated hemoglobin to complete the task.

Question 3: Are older and younger adult brains working proportionately hard within task Conditions?

Finally, regarding the third research question, older adults showed more active channels in the control condition, while younger adults demonstrated the opposite pattern, with greater activation in the interference condition. Within the older adult group, the control condition elicited activation (oxygenated hemoglobin > deoxygenated hemoglobin) in twenty-eight channels, whereas the interference condition yielded activation in twenty-five channels. In contrast, within the younger adult group, the control condition activated three channels and the interference condition elicited activation in ten channels. Thus, the younger adults could be said to display a typical pattern of activation for the MSIT. As per the MSIT protocol, it follows that the interference condition is more difficult and therefore requires additional activation or compensation to complete the task [5]. As such, it was hypothesized that older adults would demonstrate greater activation in the interference condition to compensate for increased performance deficiencies, although the reverse pattern was observed.

Interestingly, the correlation analyses revealed a significant positive correlation between OA interference HbO and reaction time. There was no correlation between OA HbO and accuracy nor YA interference HbO and reaction time or accuracy. These findings suggest that increased prefrontal activity is related to increased reaction time for older adults, specifically. In particular, OA needed more time and greater HbO to perform the interference condition, which was not the case for YA. Further, despite these compensatory mechanisms, OA still demonstrated significantly less accuracy that YA during the interference condition. These results align with CRUNCH, given that the OA group may demonstrate increased activity in an effort to compensate for declining reaction time [15].

The results for the older adult group also demonstrated recruitment of areas outside of the task network, which aligns with the STAC-R model [16]. These results are similar to those of Halliday and colleauges (2018), where the active channels of fallers and non-fallers were compared within groups. They found that fallers demonstrated significantly more active channels in both conditions than the control group- suggesting that fallers required additional functional resources compared to a less-cognitively declined group [10].

Impact, Limitations, and Future directions

The current study represents the first comparison of fNIRS activation during an executive function task in healthy older and younger adults. The results demonstrated that older adults are able to compensate for age-related inefficiencies by recruiting additional areas of activation. However, this compensatory strategy is only effective in low demand conditions and is not attainable when the task demands are high. These findings largely fit with the theoretical models of healthy aging. However, there were some inconsistencies between the current findings and the existing aging literature. It is critical to understand that there are a variety of trajectories associated with aging. Particularly in older adults, the disparity between cognitive abilities can vary greatly between individuals [11]. The current older adult sample had a relatively low standard deviation in both accuracy and reaction time, suggesting a relatively consistent group. However, it is important to note that in a larger sample size, there would likely be a larger variability of performance, even among healthy adults. Due to the increasing importance of healthy cognitive aging, more research is needed to better understand the impact of interindividual differences on cognitive aging, and how these differences are reflected in neuroimaging modalities such as fNIRS.

There are also several limitations to the current study that must be considered. Despite our sample size being comparable to previous studies [10], our findings and their interpretation are limited by our reduced sample size. According to Turner et al. [21], most neuroimaging studies are generally underpowered. The present study could have benefitted from a larger sample size in terms of greater heterogeneity amongst participants, increased generalizability, and robustness from additional power. Future studies should include larger sample sizes, as well as a comparison with resting state fNIRS data in order to validate the task-based data. Finally, another limitation is that the current study was restricted to the spatial area covered by the probe. It is typical for fNIRS studies to focus on a particular region of interest; given the current focus on executive functions, the probe was appropriately centered on the prefrontal cortex. However, it is possible that additional activation in regions beyond the probe occurred that was not captured. Encouragingly, however, findings from the current study were compared to fMRI literature with superior spatial resolution and produced compatible results. Future studies should continue to compare these techniques to exploit the advantages of each; with fNIRS offering superior temporal resolution, lower cost and greater mobility.

Conclusion

The current study was the first to use fNIRS to compare functional activation between older and younger adults during the MSIT. The results lend support to some of the previous fMRI literature suggesting that older adults demonstrate a greater extent of activation in comparison to younger adults [17]. Further, by comparing the proportion of active channels in older and younger adults, the current study showed that while older adults had more active channels, they showed fewer active channels in the interference condition relative to the control condition, suggesting reduced ability to compensate for task demands. Taken together, the use of fNIRS to compare the cognitive and functional performance of healthy adults across the lifespan is an important addition to the cognitive aging literature. As the aging population continues to grow, it will become increasingly important to understand the cognitive changes that healthy older adults undergo and the variables that influence these changes [4].

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.