Front. Hum. Neurosci.Frontiers in Human NeuroscienceFront. Hum. Neurosci.1662-5161Frontiers Media S.A.10.3389/fnhum.2025.1525497Human NeuroscienceOriginal ResearchHubs, influencers, and communities of executive functions: a task-based fMRI graph analysisDavisAlexandra T.*†Baptist Medical Center, Department of Behavioral Health, Jacksonville, FL, United States
Edited by: Daniele Corbo, University of Brescia, Italy
Reviewed by: Shihao He, Peking Union Medical College Hospital (CAMS), China
Alexander Grove Belden, Northeastern University, United States
*Correspondence: Alexandra T. Davis, alexandratongmg@gmail.com
†ORCID: Alexandra T. Davis, orcid.org/0000-0002-8451-0205
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Introduction
This study investigates four subdomains of executive functioning—initiation, cognitive inhibition, mental shifting, and working memory—using task-based functional magnetic resonance imaging (fMRI) data and graph analysis.
Methods
We used healthy adults’ functional magnetic resonance imaging (fMRI) data to construct brain connectomes and network graphs for each task and analyzed global and node-level graph metrics.
Results
The bilateral precuneus and right medial prefrontal cortex emerged as pivotal hubs and influencers, emphasizing their crucial regulatory role in all four subdomains of executive function. Furthermore, distinct hubs and influencers were identified in cognitive inhibition and mental shifting tasks, elucidating unique network dynamics. Our results suggest a decentralized brain organization with critical hub regions pertinent to conditions such as stroke and traumatic brain injury.
Discussion
The precuneus and medial prefrontal cortex stand out as consistent, domain-general nodes in our findings, which show both unique and shared neural hubs across executive function subdomains. The presence of distinct hubs in cognitive inhibition and mental shifting tasks suggests flexible, task-specific network configurations. A decentralized yet structured brain network may also promote cognitive resilience.
Executive functioning refers to several mental processes vital to effective cognitive control, encompassing tasks such as planning and organizing, initiation, time management, task shifting, and emotion regulation (Friedman and Robbins, 2022; Nemeth and Chustz, 2020). The current study utilizes archival fMRI data on initiation to investigate four subdomains: initiation, cognitive inhibition, mental shifting, and working memory.
Initiation, associated with the dorsolateral prefrontal cortex (DLPFC) and anterior cingulate cortex (ACC), involves starting and executing actions or cognitive processes (Jobson et al., 2021; Friedman and Robbins, 2022; Menon and D’Esposito, 2022). Cognitive inhibition, or inhibitory control, involves the inferior frontal gyrus, insula, superior parietal lobule (SPL), and middle cingulate (Long et al., 2022), while response inhibition involves the anterior cingulate cortex (ACC) and the pre-supplementary motor area (pre-SMA) (Morein-Zamir et al., 2013). Mental shifting, tied to the DLPFC and parietal cortex, facilitates flexible cognitive switching between or among tasks (Friedman and Robbins, 2022; Menon and D’Esposito, 2022). Working memory involves the temporary storage and manipulation of information, engaging a distributed network of brain regions, including the dorsolateral prefrontal cortex (DLPFC), the posterior parietal cortex (PPC), and the ventrolateral prefrontal cortex (VLPFC) (Chai et al., 2018; Engstrom et al., 2013).
Cognitive paradigms, such as the go/no-go, local task-switching, and n-back tasks, systematically investigate the four subdomains of executive function. The go/no-go paradigm assesses initiation and inhibition, elucidating the crucial connection between executive functioning and inhibitory control, which are indispensable for goal-directed behavior (Diamond, 2013; Verbruggen and Logan, 2008). Similarly, the local task-switching paradigm evaluates cognitive flexibility by observing how individuals manage attentional resources when transitioning between tasks with shared characteristics (Huff et al., 2015). Finally, the n-back paradigm, mainly the 2-back task, challenges participants with recalling and matching stimuli across a sequence (Jaeggi et al., 2010; Niendam et al., 2012). By employing cognitive paradigms, the present study delves into graph analysis metrics that assess a network graph’s structure, connectivity, and relationships, explicitly focusing on initiation, cognitive inhibition, shifting, and working memory.
Investigations into executive functioning often concentrate on specific brain regions (Friedman and Robbins, 2022; Menon and D’Esposito, 2022) or adopt a topographical network perspective that may use inconsistent labels for overall executive control abilities (Witt et al., 2021). In contrast, our study utilizes graph-based network analysis techniques to explore well-defined subdomains of executive function. This approach offers a robust framework for understanding the brain’s interconnected networks, effectively addressing the limitations of traditional region-based methods in studying executive functions.
While earlier research frequently focused on isolated regions, such as the prefrontal cortex, graph theory allows for examining brain-wide interactions, revealing emergent properties like global efficiency (communication efficiency across network nodes) and modularity (network structure strength) (Bullmore and Sporns, 2012; Rubinov and Sporns, 2010). The graph-based network perspective emphasizes the significance of relationships among brain regions. This approach provides insights into functional connectivity and dynamics by representing interactions as graphs. A significant advantage of this method is its capacity to identify critical nodes (hubs and influencers), similar to a social network. These regions are essential for network communication and integration. In the context of executive functions, this reveals vital components often overlooked in region-specific studies (Medaglia et al., 2015; Baum et al., 2017).
Moreover, graph theory facilitates task-specific comparisons, illustrating how different executive functions recruit distinct or overlapping network features, such as community structure or task flexibility. It quantifies the balance between functional integration (efficient global communication) and segregation (specialized local processing), which are crucial metrics for understanding brain organization during executive tasks (Baum et al., 2017; Ramos-Nuñez et al., 2017). Graph theory also sheds light on disruptions observed in clinical populations by linking network-level metrics to individual differences in cognitive performance.
This holistic approach captures both local and global properties of brain interactions, filling the gaps left by traditional region-specific analyses. It advances our understanding of the neural basis of executive functions by focusing on system-wide organization rather than isolated activity. Despite the growing interest in this area, few studies have systematically compared these subdomains using network-based graph theory approaches. Our research aims to bridge this gap by integrating traditional regional methods with graph analyses, providing a more comprehensive understanding of executive functions.
Graph-based network analysis serves as a valuable framework for unraveling the complexities inherent in systems represented as graphs. This methodology enhances our comprehension of intricate brain network connectivities and patterns (Bullmore and Sporns, 2009). Certain cortical areas emerge as highly connected or centralized regions, critical focal points (Farahani et al., 2019). The application of graph-based theory in cognitive neuroscience, particularly in human connectome studies, has evolved significantly, correlating brain network properties with human intelligence, memory, attention, and emotional processing (Farahani et al., 2019). For instance, research has demonstrated a correlation between working memory performance and local/global measures in brain networks (Stanley et al., 2015). Additionally, disruptions in functional network topology have been implicated in various cognitive and psychiatric disorders (Reijneveld et al., 2007).
This study aims to investigate core graph metrics to understand the network properties of cortical areas crucial for executive functions, specifically initiation, inhibition, shifting, and working memory in healthy adults. Utilizing graph-based network analysis, we seek to address several critical research questions:
Are there significant differences in brain network graphs between these executive functions?
What specific network features are associated with each task?
Which brain regions are essential for these functions?
Moreover, we propose the following hypotheses:
Brain network graphs will display significant differences in their topological properties—such as clustering coefficient, modularity, and global efficiency—across tasks related to different executive functions. This variation will reflect the distinct neural processing demands of each task.
Each executive function task will yield unique network features. For instance, tasks emphasizing working memory are expected to exhibit higher modularity, while those involving cognitive inhibition will show increased connectivity in control-related regions. Additionally, we anticipate tasks focused on shifting demonstrate greater flexibility in inter-community connections.
Specific brain regions will serve as critical hubs or influencers across these tasks. We expect the dorsolateral prefrontal cortex to play a central role in working memory, the anterior cingulate cortex vital for cognitive inhibition, and the parietal regions to be key in task-shifting. These essential regions likely exhibit high centrality and betweenness values, underscoring their importance in network communication.
This study aims to enhance our understanding of the neural underpinnings of executive functioning and its relationship with brain network organization by addressing these questions.
MethodsData acquisition
This study employed a publicly available dataset derived from functional magnetic resonance imaging (fMRI) scans of healthy adults (Rieck et al., 2021). One hundred forty-four participants (ages 20 to 86) underwent scanning using a Siemens 3 T MRI scanner while engaging in cognitive paradigms to assess functional activity. These paradigms included a go/no-go task for examining inhibition and initiation, a local task-switching paradigm for shifting, and an n-back task with three load levels (0-back, 1-back, and 2-back) for working memory (Rieck et al., 2021).
Functional connectivity estimates (quantified with time-series correlations) between various brain regions were computed using three distinct brain atlases (Rieck et al., 2021): the Schaefer 100 parcel 17 network atlas (Schaefer et al., 2018; Thomas Yeo et al., 2011), the Power 229 node 10 network atlas (Power et al., 2011), and the Schaefer 200 parcel 17 network atlas (Schaefer et al., 2018; Thomas Yeo et al., 2011; Rieck et al., 2021). The present study utilizes the correlation data obtained from the Schaefer 200 parcel 17 network atlas.
Processing
To identify relevant brain regions (ROIs) with robust functional connectivity, ROIs exhibiting a high correlation coefficient exceeding 0.75 in the adjacency matrices of individual participants were selected. ROIs associated with the motor and visual networks were excluded to maintain the study’s focus on executive functioning. Additionally, ROIs with limited occurrences—those with connections observed in less than 10 participants—were also excluded to ensure the inclusion of reliably connected brain regions. The resulting ROIs and the corresponding aggregated frequency of functional connectivity incidents constitute an adjacency matrix for each task (inhibition, initiation, shifting, and 2-back).
Brain network construction
The analysis pipeline is depicted in Figure 1. A brain graph network comprises nodes (brain regions) and edges (functional connectivities) (Fornito et al., 2016). Nodes can be assigned binary or weighted values representing activity intensity. Considering the interindividual variability in brain connectomes (Sun et al., 2022), we aggregate weighted values across participants to obtain collective brain activity. This connectivity was utilized to construct an adjacency matrix (Figure 2), which signifies connections between nodes in a graph (Alper et al., 2013). The functional network was mapped using an adjacency matrix for each task and visualized on a connectome utilizing the Schaefer200_n17 atlas (Figures 2, 3). Furthermore, network graphs were generated for each task (Figure 4).
Pipeline for data analysis. (A) Obtain raw archival data. (B) Functional connectivity matrices are constructed, and connections with less frequency were excluded. (C) Adjacency matrices were derived based on step (B). (D) Graphs were generated and plotted on connectome. (E) Graph theoretical analysis. Metrics such as degree centrality, clustering coefficient, and modularity are computed to characterize the network’s local and global properties. (F) Statistical analyses are applied to identify significant patterns or differences across tasks. (G) Results were plotted on Schaefer for visualizations of the findings.
Flowchart of the process from an archival dataset to visual representation. It begins with an archival dataset (A), moves to aggregated connections (B), then to an adjacency matrix (C). This data is used to create a functional network (D) and network graph (E), followed by graph measures calculation (F). Statistical analysis (F) and visualization (G) of the results are shown, with various graphical representations and analyses of brain networks and their connections.
Adjacency matrix for each task. (a) Initiation, (b) Inhibition, (c) Shifting, (d) 2-back.
Four heatmap plots labeled (a) Initiation, (b) Inhibition, (c) Shifting, and (d) 2-back. Each plot has a dendrogram on two sides and contains a yellow-to-red color scale, indicating varying data intensities across a grid.
Graphs plotted on connectome for each task. (a) Initiation, (b) Inhibition, (c) Shifting, (d) 2-back.
Four brain diagrams labeled (a) Initiation, (b) Inhibition, (c) Shifting, and (d) 2-back show neural networks with colored nodes and connecting lines. Networks include ContA, ContB, DefaultA, and others. Connection weights are indicated by varying line thickness, represented by a legend with weights from 30 to 125. Each diagram highlights different network connections and weights.
Graphs for each task. (a) Initiation, (b) Inhibition, (c) Shifting, (d) 2-back.
Four network graphs labeled (a) Initiation, (b) Inhibition, (c) Shifting, and (d) 2-back. Nodes are labeled with ROI numbers and connected by edges. Edge color represents weight, and node size indicates frequency. A legend explains these visual codes.Graph measures
Due to a lack of directionality in fMRI data, the current study employs weighted undirected matrices and investigates the topological characteristics of functional brain networks for each task. To analyze topographical features, conventional graph measures such as node centrality measures (degree, strength, betweenness, and closeness), clustering coefficient, modularity, characteristic path length, and small-worldedness, among others, were employed (Tables 1–5) (Sporns and Betzel, 2016; van den Heuvel et al., 2008).
Node degree.
Initiation
Inhibition
Shifting
2-back
Node
Degree
Node
Degree
Node
Degree
Node
Degree
189
7
72
8
134
6
134
7
72
6
189
8
136
6
73
6
182
6
134
8
189
5
189
5
186
6
182
8
73
4
136
5
73
5
139
7
33
4
33
4
33
5
33
6
34
4
72
4
136
5
186
6
72
4
38
4
139
4
159
6
58
4
58
4
134
4
136
6
32
4
139
4
159
4
77
5
139
4
182
4
77
3
88
5
135
4
165
4
195
3
90
5
38
3
88
3
82
3
73
5
81
3
34
3
81
3
32
5
186
3
81
3
140
3
58
5
150
3
32
3
38
3
96
5
182
3
186
3
150
3
181
5
159
3
179
3
90
3
194
5
192
3
159
3
32
3
79
4
133
3
192
3
181
3
38
4
88
2
90
2
78
2
34
4
90
2
79
2
79
2
75
4
79
2
50
2
179
2
150
4
82
2
89
2
97
2
195
4
89
2
52
2
133
2
133
4
52
2
82
2
74
2
82
3
91
2
91
2
192
2
78
3
179
2
78
2
34
2
81
3
181
2
71
2
52
2
89
3
195
2
150
2
89
2
179
3
140
2
181
2
91
2
140
3
77
1
195
2
58
2
192
3
46
1
140
2
48
2
165
3
45
1
187
2
147
1
135
3
80
1
183
2
44
1
97
2
97
1
135
2
61
1
45
2
50
1
77
1
183
1
52
2
66
1
46
1
170
1
50
2
74
1
80
1
40
1
74
2
96
1
45
1
160
1
91
2
78
1
97
1
50
1
40
2
60
1
66
1
66
1
37
2
37
1
74
1
156
1
48
2
40
1
44
1
187
1
59
2
48
1
68
1
142
1
99
2
59
1
37
1
80
1
187
2
61
1
75
1
45
1
156
2
71
1
147
1
46
1
160
2
147
1
156
1
88
1
183
2
187
1
157
1
37
1
173
2
156
1
170
1
71
1
132
2
160
1
160
1
157
1
46
1
157
1
149
1
149
1
80
1
170
1
133
1
165
1
66
1
183
1
178
1
153
1
44
1
165
1
71
1
142
1
31
1
153
1
47
1
61
1
36
1
57
1
60
1
68
1
95
1
147
1
170
1
157
1
142
1
149
1
153
1
166
1
199
1
151
1
138
1
137
1
176
1
193
1
198
1
Node strength.
Initiation
Inhibition
Shifting
2-back
Node
Strength
Node
Strength
Node
Strength
Node
Strength
186
286
186
365
139
208
73
205
189
205
189
300
136
187
134
174
182
186
182
291
73
185
189
167
139
168
139
275
189
182
139
158
73
162
159
244
186
176
136
134
77
155
73
233
134
176
186
133
72
152
134
232
33
168
88
132
33
142
33
231
182
136
182
132
79
139
72
227
88
125
33
125
134
124
79
210
77
117
179
108
159
122
77
203
72
117
72
102
38
110
88
186
79
115
38
97
136
106
181
162
34
114
79
95
88
102
136
161
179
114
77
85
150
93
38
159
159
113
90
84
195
89
90
156
38
111
159
83
90
86
179
151
150
103
58
78
179
86
150
134
90
97
150
72
181
85
81
131
181
85
192
62
82
82
195
125
82
84
165
62
97
75
82
123
58
83
34
61
81
74
34
113
46
79
52
58
46
68
97
112
32
74
181
58
140
59
89
99
195
68
46
54
78
54
52
98
52
65
81
52
147
54
78
94
192
65
82
49
45
54
32
93
45
64
50
46
89
54
192
93
147
64
195
46
80
53
46
87
135
64
89
45
52
53
80
83
80
59
91
45
34
51
140
83
89
56
80
43
192
47
45
80
97
55
140
42
32
46
50
80
133
55
45
39
187
42
187
80
81
54
147
39
156
38
58
76
187
54
187
39
50
38
156
76
50
53
32
36
91
38
133
70
156
53
97
35
74
37
91
68
140
48
156
34
160
28
96
68
91
43
78
30
170
28
147
68
160
26
157
30
66
28
74
63
66
25
183
29
58
25
160
62
157
25
66
28
133
25
183
59
170
25
170
28
183
22
194
59
74
22
135
27
48
21
66
53
183
22
160
24
157
17
170
53
96
21
71
21
37
14
75
47
78
20
74
18
44
13
165
42
165
20
44
17
149
13
135
42
60
17
149
17
142
12
37
36
37
15
68
15
40
12
48
35
40
11
37
12
165
12
59
33
48
11
75
12
153
11
40
31
59
11
133
11
61
10
157
31
142
11
178
10
71
10
173
25
153
11
132
24
61
10
99
22
71
10
142
21
44
20
149
20
153
16
71
15
31
14
166
14
199
12
47
11
61
11
151
11
36
10
57
10
60
10
68
10
95
10
138
10
137
10
176
10
193
10
198
10
Node betweenness.
Initiation
Inhibition
Shifting
2-back
Node
Betweenness
Node
Betweenness
Node
Betweenness
Node
Betweenness
189
138
159
895.3666667
189
101.5
189
132
182
84.151826
77
848.1891273
134
89.468074
182
132
90
81.5
189
572.5642857
182
88.648268
73
116.2
73
72.513161
182
550.4220114
136
75.97132
134
107.616667
72
68.605463
88
363.0404762
72
73.452381
79
105
179
64
179
312.9166667
192
72.5
88
96.5
186
61.710859
58
265.3270125
73
69.377489
186
91
136
59.64667
91
265
81
69
192
84.5
159
55
90
263.1595238
79
66
81
79
33
54.013823
72
227.7273555
90
65
72
70.9
91
51
150
221
179
56
159
53
81
45
82
189.8714286
58
50.816991
179
51.5
82
45
81
189.8714286
135
47.916667
38
50.733333
150
37
186
180.5773489
186
47
165
50.516667
134
33.551963
89
168.8166667
159
47
78
48
140
24.714596
135
167
38
46.708333
91
36
74
22
194
153.6120443
91
45
90
30
58
22
75
126.5622013
34
40.60303
89
30
48
19
134
125.977697
88
35
136
28.483333
52
19
139
114.5426742
89
35
33
27.733333
139
17.64667
192
111.102381
32
34.75184
195
25
77
14.227234
195
96.4863525
150
33
140
25
181
13.160497
33
76.6583392
139
30.957251
183
25
195
8.555808
73
63.8258992
33
28.289935
32
22.9
78
7.643687
45
57
133
24.791667
58
22.8
79
7.643687
52
57
195
24
50
19
38
6.861111
48
57
140
24
52
19
89
6.5
59
57
82
17
71
19
32
5.341834
173
57
52
17
150
19
192
3
96
54.0447632
181
7.246753
139
18.866667
133
2.861111
136
49.6574909
77
0
82
5.5
97
1.4
78
48.6184711
46
0
187
4
34
0.75
79
33.2078878
45
0
34
2.733333
80
0
38
27.9546775
80
0
135
1.516667
187
0
34
27.9546775
97
0
181
1
44
0
50
20.2404762
50
0
77
0
88
0
165
18.7331633
66
0
46
0
46
0
181
16.4903546
74
0
80
0
71
0
74
13.4422031
96
0
45
0
45
0
133
12.2607623
78
0
97
0
37
0
32
10.9993036
60
0
66
0
50
0
37
6.8630962
37
0
74
0
147
0
156
3.5833333
40
0
44
0
157
0
140
3.1626206
48
0
68
0
66
0
183
2.0774802
59
0
37
0
40
0
97
1.6249851
61
0
75
0
61
0
40
1
71
0
147
0
156
0
99
1
147
0
156
0
170
0
132
1
187
0
157
0
160
0
187
0.7333333
156
0
170
0
183
0
160
0.7333333
160
0
160
0
149
0
46
0
157
0
149
0
142
0
80
0
170
0
133
0
165
0
66
0
183
0
178
0
153
0
44
0
165
0
71
0
142
0
31
0
153
0
47
0
61
0
36
0
57
0
60
0
68
0
95
0
147
0
170
0
157
0
142
0
149
0
153
0
166
0
199
0
151
0
138
0
137
0
176
0
193
0
198
0
Node closeness.
Initiation
Inhibition
Shifting
2-back
Node
Closeness
Node
Closeness
Node
Closeness
Node
Closeness
156
1
66
1
45
1
45
1
147
1
44
1
50
1
66
1
40
1
47
1
66
1
44
1
45
1
36
1
74
1
147
1
66
1
95
1
40
1
170
1
170
1
170
1
48
1
149
1
149
1
149
1
147
1
189
0.0181818
50
1
151
1
156
1
88
0.0172414
44
1
138
1
170
1
81
0.015625
142
1
193
1
183
1
192
0.0153846
189
0.0217391
40
0.5
142
1
73
0.0151515
182
0.0196078
99
0.5
153
1
134
0.0149254
90
0.0188679
132
0.5
189
0.0181818
182
0.0144928
73
0.0188679
31
0.3333333
88
0.0172414
72
0.0138889
72
0.0181818
57
0.3333333
89
0.0172414
50
0.0138889
82
0.0181818
142
0.3333333
192
0.0163934
82
0.0138889
81
0.0181818
199
0.3333333
81
0.016129
71
0.0138889
33
0.0181818
137
0.3333333
134
0.015873
187
0.0133333
134
0.0181818
198
0.3333333
136
0.015873
90
0.0131579
181
0.0163934
77
0.0049751
73
0.0149254
89
0.0131579
77
0.015873
182
0.0048077
90
0.0147059
32
0.0131579
89
0.015873
159
0.0048077
82
0.0142857
159
0.012987
48
0.015873
186
0.0044248
34
0.0140845
136
0.012987
136
0.015873
194
0.0044248
72
0.0140845
139
0.0126582
159
0.015625
88
0.0044053
58
0.0140845
58
0.0121951
179
0.015625
82
0.004329
139
0.0140845
160
0.0120482
186
0.0153846
81
0.004329
71
0.0138889
79
0.0119048
79
0.0153846
195
0.0042735
33
0.0136986
38
0.0119048
88
0.0153846
72
0.0042553
32
0.0136986
165
0.0117647
78
0.0153846
58
0.0041494
159
0.0136986
33
0.0116279
71
0.0153846
33
0.0041152
182
0.0133333
179
0.0114943
139
0.0149254
73
0.0040816
179
0.012987
34
0.0113636
192
0.0142857
96
0.0040161
160
0.0126582
135
0.011236
140
0.0140845
189
0.004
38
0.0123457
156
0.010989
32
0.0136986
79
0.0039526
181
0.0123457
178
0.010989
195
0.0136986
75
0.0039216
187
0.0114943
52
0.0106383
187
0.0135135
78
0.0038911
135
0.0114943
181
0.0105263
160
0.0135135
192
0.0038168
52
0.0113636
195
0.0105263
133
0.0133333
52
0.0038023
91
0.0113636
80
0.0104167
91
0.012987
134
0.0037879
80
0.0111111
186
0.009901
38
0.0126582
80
0.0037736
133
0.010989
91
0.0098039
52
0.0123457
139
0.0036232
79
0.0107527
140
0.0093458
153
0.0121951
136
0.0035971
195
0.0107527
68
0.0090909
58
0.0120482
97
0.0035842
165
0.0105263
75
0.0090909
80
0.0120482
74
0.0035587
150
0.009901
133
0.009009
34
0.0120482
181
0.0035461
140
0.0097087
157
0.0088496
97
0.0117647
187
0.0035461
157
0.0095238
150
0.0084034
74
0.0117647
160
0.0035461
60
0.009009
97
0.0083333
150
0.0108696
165
0.0035461
59
0.009009
78
0.0081967
37
0.0107527
133
0.0035211
186
0.0088496
77
0.0079365
157
0.01
90
0.0034843
96
0.0086957
37
0.0075758
165
0.0095238
89
0.0034364
97
0.008547
46
0.0072464
183
0.0093458
135
0.0034247
46
0.0084746
183
0.0068966
46
0.009009
50
0.0032787
61
0.0084746
74
0.0058824
61
0.009009
48
0.0032787
37
0.007874
71
0.0032573
77
0.0072993
173
0.0032258
78
0.0072993
32
0.0032154
183
0.0032051
38
0.0031949
34
0.0031949
157
0.003125
179
0.0030303
37
0.0030211
156
0.002924
176
0.002907
59
0.0028818
60
0.0028653
153
0.0027624
140
0.0027322
68
0.0027248
91
0.0026525
166
0.0024752
150
0.0023474
45
0.002079
46
0.0020704
61
0.0020704
147
0.0018587
Graph metrics across four executive tasks and their interpretations.
Metrics
Initiation
Inhibition
Shifting
2-back
Value
Interpretation
Value
Interpretation
Value
Interpretation
Value
Interpretation
Modularity
0.66
Network divided into distinct communities
0.62
Network divided into distinct communities
0.68
Network divided into distinct communities
0.69
Network divided into distinct communities
Global Efficiency
0.13
Low efficiency in information transfer
0.16
Moderate efficiency in information transfer
0.12
Low efficiency in information transfer
0.14
Moderate efficiency in information transfer
Path length ratio
0.87
Observed path length is slightly lower than random path length
1.31
Observed path length is about 1.31 times longer than random path length
1.00
Observed path length is very close to random path length
0.96
Observed path length is slightly lower than random path length
Characteristic Path Length
3.35
Nodes are relatively close to each other in terms of network connectivity
5.17
Nodes are relatively distant from each other in terms of network connectivity
3.91
Nodes are relatively close to each other in terms of network connectivity
3.93
Nodes are relatively close to each other in terms of network connectivity
Assortativity
0.07
Slight assortativity, indicating a tendency for nodes with similar degrees to be connected.
0.34
Moderate tendency for nodes to attach to similar nodes
0.37
Moderate tendency for nodes to attach to similar nodes
0.35
Moderate tendency for nodes to attach to similar nodes
Edge Density
0.04
Low, indicating a sparse network with few connections.
0.04
Low, indicating a sparse network with few connections.
0.04
Low, indicating a sparse network with few connections.
0.04
Low, indicating a sparse network with few connections.
Small-worldness (sigma)
0.00
Not exhibiting small-world properties
0.00
Not exhibiting small-world properties
0.00
Not exhibiting small-world properties
0.00
Not exhibiting small-world properties
Transitivity
0.00
Absence of clustering
0.00
Absence of clustering
0.00
Absence of clustering
0.00
Absence of clustering
Clustering Coefficient
0.00
Absence of clustering
0.00
Absence of clustering
0.00
Absence of clustering
0.00
Absence of clustering
While it is frequently challenging to ascertain the most appropriate metrics for investigating brain networks (Bullmore and Sporns, 2009), centrality measures and small-world characteristics (e.g., high clustering coefficient and short characteristic path length) are indispensable for this process (He and Evans, 2010). Although there are no established criteria for “hub status,” most studies consider nodes with high centrality measures as hubs (Farahani et al., 2022; Fornito et al., 2016). In this study, weights denote the aggregate frequency of connections among participants. Local network measures were computed for each node (Brian region), including nodal strength, betweenness centrality, and closeness centrality. Nodes exhibiting the top 20% values for strength and betweenness were designated as hubs.
Furthermore, this study employed a novel centrality measure known as “expected force,” which quantifies a node’s potential influence within a network by summing the weights of its connections. This measure identifies critical nodes facilitating information flow, such as key brain regions in executive functions (Bullmore and Bassett, 2011). Nodes exhibiting high expected force are likely to influence other network nodes significantly. Unlike other centrality measures, expected force maintains reliability in network alterations, ensuring accuracy for incomplete or noisy systems (Lawyer, 2015). Therefore, potentially identify brain regions or connections that could be of interest in reorganization post-TBI or seizure disorders. Consequently, the top 20% of nodes with high expected force were identified as influencers.
Global network measures were also computed, including community detection, density, clustering coefficient, modularity, assortativity, characteristic path length, and small-worldedness. Community detection algorithms, such as Louvain and Infomap, identify subnetworks or modules (Hric et al., 2014). Assortativity measures the tendency of nodes in a network to connect with other nodes that have similar or dissimilar properties. In a brain network, it can be used to understand connectivity patterns (Rubinov and Sporns, 2010) and indirectly reflect network resilience (Farahani et al., 2019). Density reflects network connectivity, and modularity assesses community strength. The clustering coefficient indicates node clustering, while characteristic path length gauges information transfer efficiency.
Statistical analysis
Nonparametric (Kolmogorov–Smirnov) tests were employed to analyze the degree distribution of the graphs. The Kruskal-Wallis test was utilized to compare node strength and hubs across the tasks. In contrast, a pairwise comparison (Wilcoxon rank sum test) was used to elucidate their differences further.
Results
Tables 1–4 present each task’s node degree, strength, betweenness, and closeness. Table 6 and Figure 5 identify hubs with high strength and betweenness; Table 7 and Figure 6 highlight influencer nodes with high expected force. Figure 7 depicts the dendrogram for each graph’s edge betweenness community, and Tables 8–11 present the Louvain community for each graph. Figures 7, 8 illustrate the Louvain communities in the Schaefer atlas. Both algorithms yield comparable results.
Hubs.
All 4 graphs
ROI
Schaefer node label
Cortical areas
72
LH_ContC_pCun_1
LH Control Network Precuneus 1
182
RH_ContC_pCun_2
RH Control Network Precuneus 2
189
RH_DefaultA_PFCm_3
RH Default Network medial prefrontal 3
Initiation
ROI
Schaefer node label
Cortical areas
33
LH_DorsAttnA_SPL_2
LH Dorsal Attention Network Superior Parietal Lobule 2
Four brain maps show activation areas for different cognitive tasks: (a) Initiation, (b) Inhibition, (c) Shifting, and (d) 2-back. Each map highlights regions using color-coded hubs. These correspond to specific brain networks indicated by a color legend on the right of each map. Each map displays lateral and medial views of the left and right brain hemispheres.
Influencers.
Initiation
ROI
Schaefer node label
Cortical areas
33
LH_DorsAttnA_SPL_2
LH Dorsal Attention Network Superior Parietal Lobule 2
72
LH_ContC_pCun_1
LH Control Network precuneus 1
73
LH_ContC_pCun_2
LH Control Network precuneus 2
134
RH_DorsAttnA_SPL_2
RH Dorsal Attention Network Superior Parietal Lobule 2
136
RH_DorsAttnA_SPL_4
RH Dorsal Attention Network Superior Parietal Lobule 4
139
RH_DorsAttnB_PostC_3
RH Dorsal Attention Network Post Central gyrus (medial segment)
Four panels show brain models with corresponding color-coded network maps illustrating high expected force in various brain functions. Panel (a) Initiation shows activation in specific networks. Panel (b) Inhibition displays different activated networks. Panel (c) Shifting highlights another set of networks. Panel (d) 2-back shows networks related to memory tasks. Each panel includes lateral and medial views for both the right and left sides of the brain. A legend on the right of each panel identifies networks by color.
Main Louvain communities plotted on brain atlas. (a) Initiation, (b) Inhibition, (c) Shifting, (d) 2-back.
Four diagrams show brain maps divided into colored regions, labeled as communities, depicting lateral and medial views for different cognitive tasks: (a) Initiation, (b) Inhibition, (c) Shifting, and (d) 2-back. Each diagram distinguishes different brain regions using various colors, corresponding to distinct community functions.
A comparative analysis of graph network measures across subdomains of executive functioning, including initiation, cognitive inhibition, mental shifting, and working memory, unveiled nuanced variations and commonalities in the organization and functional connectivity patterns. Specific brain regions consistently emerged as prominent hubs (Table 6), facilitating information exchange within the graph of all four subdomains of executive function. These regions include the bilateral precuneus (LH_ContC_pCun_1, RH_ContC_pCun_2) and the right medial prefrontal cortex (DMN) (RH_DefaultA_PFCm_3). Notably, these regions exhibited high node strength and betweenness centrality and are connected to each other, indicating their pivotal role in facilitating executive functioning processes. Closeness centrality analysis revealed that several nodes in each graph exhibit a closeness of 1 (Table 4), yet they form isolated subnetworks. Nodes with the following highest closeness values are notably low, suggesting their relative distance from other nodes regarding functional connectivity. This implies reduced efficiency in transmitting information or influence across the broader brain network.
A Kolmogorov–Smirnov test compared the observed network properties with those of randomly generated networks, revealing significant differences in degree distributions (Table 12). Considerable variations were identified among the shared ROIs across the four subdomains. Utilizing a Kruskal-Wallis test, we observed a chi-squared value of 21.634 and a p-value of 7.772e-05, indicating statistically significant discrepancies in strength across the four tasks. As depicted using the Wilcoxon rank sum test (Table 13), notable disparities in the strength of common nodes (ROIs) were evident, particularly between inhibition and the other functions. Conversely, no statistically significant distinctions between initiation, shifting, and the 2-back tasks were observed (Table 14).
Degree distribution.
Initiation
Inhibition
Shifting
2-back
Degree
Distribution
Degree
Distribution
Degree
Distribution
Degree
Distribution
1
0.4
1
0.3461538
1
0.4736842
1
0.3518519
2
0.2363636
2
0.2179487
2
0.1929825
2
0.2962963
3
0.1818182
3
0.1153846
3
0.1403509
3
0.1481481
6
0.0545455
4
0.0897436
4
0.1403509
4
0.1296296
5
0.0545455
5
0.1153846
5
0.0175439
5
0.037037
7
0.0181818
6
0.0512821
6
0.0350877
6
0.0185185
7
0.0128205
7
0.0185185
8
0.0512821
Pairwise comparisons using the Wilcoxon rank sum test.
Tasks
Initiation
Inhibition
Shifting
Inhibition
0.0026
–
–
Shifting
1
0.0154
–
2back
1
0.0001
0.9116
Kolmogorov–Smirnov test results.
Initiation
Inhibition
Shifting
2-back
D
p-value
D
p-value
D
p-value
D
p-value
0.85455
2.01E-05
0.96154
3.11E-09
0.87719
2.65E-05
0.87037
8.14E-06
Unique regions of interest (ROIs) specific to each task (Table 15) are also analyzed. ROIs 194, 135, 96, and 75 exhibit high degrees but low strength. The inhibition graph possesses the highest number of distinct ROIs compared to the other graphs and appears to be the most distinctive among the four. Kruskal-Wallis rank sum tests were performed for a list of hubs from each graph, resulting in a p-value of 0.7965. This indicates no significant difference in the hubs among the graphs.
Distinct nodes across tasks.
Node
Initiation
Inhibition
Shifting
Two back
31
NA
1
NA
NA
36
NA
1
NA
NA
40
1
2
1
NA
44
1
1
NA
1
47
NA
1
NA
NA
48
2
2
1
NA
57
NA
1
NA
NA
59
NA
2
1
NA
60
NA
1
1
NA
61
1
1
1
NA
68
NA
1
NA
1
75
NA
4
NA
1
95
NA
1
NA
NA
96
NA
5
1
NA
99
NA
2
NA
NA
132
NA
2
NA
NA
135
NA
3
4
2
137
NA
1
NA
NA
138
NA
1
NA
NA
142
1
1
1
NA
149
1
1
NA
1
151
NA
1
NA
NA
153
1
1
1
NA
166
NA
1
NA
NA
167
NA
NA
2
NA
169
NA
NA
1
NA
173
NA
2
NA
NA
175
NA
NA
1
NA
176
NA
1
NA
NA
178
NA
NA
NA
1
193
NA
1
NA
NA
194
NA
5
NA
NA
198
NA
1
NA
NA
199
NA
1
NA
NA
Unique to the cognitive inhibition graph are regions LH_DefaultA_pCunPCC_1 (left posterior cingulate cortex/precuneus), LH_DefaultB_PFCd_1 (left dorsal prefrontal cortex 1), and LH_DefaultB_PFCd_3 (left dorsal prefrontal cortex 3) (ROIs 77, 88, and 90), which emerge as hubs. Conversely, in the mental shifting graph, regions LH_DefaultA_pCunPCC_3 (left posterior cingulate cortex/precuneus 3), RH_DorsAttnA_SPL_2 (right superior parietal lobule 2), and RH_DorsAttnA_SPL_4 (right superior parietal lobule 4)(ROIs 134, 136, and 79) assume hub roles. Regions unique to cognitive inhibition and mental shifting (Table 15) were detected. In the inhibition graph, ROIs 75, 96, and 194 (LH_DefaultA_IPL_1, LH_DefaultC_IPL_1, and RH_DefaultC_IPL_1), specifically, bilateral inferior parietal lobule, were identified as key regions of connectivity. In contrast, the shifting graph exhibited uniquely high connections involving ROI 135 (RH_DorsAttnA_SPL_3), the right superior parietal lobule.
Furthermore, each executive subdomain was associated with distinct influencers (Table 12). Notably, many of the identified influencers also function as hubs. Additionally, initiation function is influenced by ROIs 136, 134, and 139 (RH Dorsal Attention Network Superior Parietal Lobule 2, RH Dorsal Attention Network Superior Parietal Lobule 4, and RH Dorsal Attention Network Post Central gyrus (medial segment)) as influencers. Cognitive inhibition is influenced by several regions of interest (ROIs) in the human brain, including:
Left Hemisphere: Superior Parietal Lobule 1 (LH Dorsal Attention Network), Inferior Parietal Sulcus 1 (LH Control Network), and Inferior Parietal Lobule 1 (LH Default Network).
Right Hemisphere: Superior Parietal Lobule 1 (RH Dorsal Attention Network), Superior Parietal Lobule 4 (RH Dorsal Attention Network), Post Central Gyrus (medial segment) (RH Dorsal Attention Network), Precuneus 1 (RH Control Network), and Inferior Parietal Lobule 1 (RH Default Network).
Mental shifting requires influences from ROIs 32, 33, 34, 58, 133, and 135 (LH Dorsal Attention Network Superior Parietal Lobule 1, LH Dorsal Attention Network Superior Parietal Lobule 2, LH Dorsal Attention Network Superior Parietal Lobule 3, LH Control Network Inferior Parietal Sulcus 1, RH Dorsal Attention Network Superior Parietal Lobule 1, and RH Dorsal Attention Network Superior Parietal Lobule 3).
Working memory requires influences from ROIs 38, 58, and 165 (LH Dorsal Attention Network Post Central Gyrus 4, LH Control Network Inferior Parietal Sulcus 1, and RH Control Network Inferior Parietal Sulcus 1).
Brain graphs with a short characteristic path length are believed to integrate information more efficiently between nodes (Paldino et al., 2016). In contrast, inhibition graphs exhibit a relatively long characteristic path length of 5.1681 compared to the initiation, mental shifting, and working memory graphs (3.348269, 3.9143, and 3.9273, respectively). Initiation has a lower assertiveness value, indicating a more neutral or distributed balance, balancing local and global connectivity. However, it might rely more on specific hubs for overall functionality, making it vulnerable to hub damage, as seen in conditions like stroke or traumatic brain injury. Global efficiency is inversely proportional to the topological distance between nodes and is typically interpreted as a measure of the capacity for parallel information transfer and integrated processing (Bullmore and Sporns, 2009). The observation that all graphs exhibit low to moderate levels of global efficiency suggests that the brain regions are not highly interconnected, thereby limiting the efficiency of information transmission across the network. Furthermore, the degree distribution (Table 5) exhibited characteristics of an exponentially truncated power law distribution. In other words, most nodes have relatively low degrees, while some have extremely high degrees.
The clustering coefficient provides insight into the local connectivity of nodes within a network, reflecting the extent to which neighboring nodes are interconnected (Bullmore and Sporns, 2009). It assesses the prevalence of clustered connections among nearby nodes, indicating the likelihood of forming local clusters or communities. Path transitivity evaluates the number of local detours along a path, contributing to understanding how efficiently information flows within the network. Graphs with a high small-world value exhibit densely clustered local connections and optimal long-range connections, facilitating efficient information processing at minimal cost (Bassett and Bullmore, 2006; Bullmore and Sporns, 2009). The clustering coefficient, transitivity, and small-worldedness sigma of 0 indicate a decentralized structure. However, modularity values of all four graphs suggest the presence of distinct communities across all subdomains of executive function, highlighting the absence of local clustering. Indeed, the community detection analysis unveils a rich modular structure within each graph (Figures 3, 4; Tables 8–11).
Discussion and clinical implications
The evolution of graph theory in cognitive neuroscience has provided valuable insights into the intricate connections within the human brain, offering a robust framework for understanding cognitive processes and their neural underpinnings (Bullmore and Sporns, 2009; Farahani et al., 2019; Medaglia, 2017; Medaglia et al., 2015). Executive functioning, essential for daily activities (Zelazo et al., 2004), encompasses various cognitive processes. This study enhances our understanding of executive functioning in healthy adults by identifying key hub/influencer regions and analyzing local and global properties of subdomains of executive functioning, namely, initiation, cognitive inhibition, mental shifting, and working memory.
Hubs and influencers
Our hypothesis that specific brain regions will serve as critical hubs or “influencers” across these tasks was confirmed. The precuneus and right medial prefrontal cortex (mPFC) emerged as crucial hubs for all four subdomains of executive function. Our findings also support previous research highlighting the dorsolateral prefrontal cortex (DLPFC), anterior cingulate cortex (ACC), and parietal regions as key components of executive function.
Both the precuneus and mPFC have been identified as integral components of the default mode network (DMN), which is typically active during rest and internally directed thought (Cavanna and Trimble, 2006; Yeager et al., 2022; Friedman and Robbins, 2022; Jobson et al., 2021; Menon and D’Esposito, 2022). The DMN deactivates during cognitively demanding tasks, allowing for more focused information processing (Jin et al., 2012; Leech and Sharp, 2014; Salgado-Pineda et al., 2021; Billette et al., 2022; Xu et al., 2019).
In our study, the precuneus exhibited connectivity with the posterior cingulate cortex (PCC), which deactivates alongside the precuneus during executive function tasks (Raichle, 2015), as well as with the inferior and superior parietal cortices, which exhibit task-dependent activation levels (Yeo et al., 2015). Similarly, the right mPFC showed strong connectivity with other prefrontal regions, which are generally activated during executive functions.
Given their role as hubs and “influencers,” the mPFC and precuneus likely regulate network-wide activity, influencing when to engage or suppress cognitive processes depending on task demands. Dysfunction in these regions is associated with attention deficits, impaired self-referential thinking, and decision-making difficulties and has been linked to neurological disorders such as Alzheimer’s disease, schizophrenia, and depression (Buckner et al., 2009; Menon, 2011). These findings highlight the potential of these regions as targets for neuromodulation techniques, such as transcranial magnetic stimulation (TMS), to enhance executive function in individuals affected by stroke, neurodegeneration, or cognitive impairments (Guse et al., 2010).
While the literature on the right hemisphere is less extensive, surgical mapping studies have indicated the involvement of the right ventromedial prefrontal cortex (vmPFC) and orbital frontal areas in facial emotion recognition and theory of mind (Bernard et al., 2018). Our findings suggest a hub and influencer role for the right mPFC in executive functioning, contributing to our understanding of right hemisphere involvement in cognitive processes.
In addition to hub regions shared by all four subdomains of executive functions, our findings also identified unique hubs for cognitive inhibition as the left posterior cingulate cortex/precuneus and the left dorsal prefrontal cortex. Moreover, the mental shifting function relied on hub regions such as the left posterior cingulate cortex/precuneus and the right superior parietal lobule. The superior parietal lobule had previously been studied for the function of attentional shifting (Wang et al., 2014). Interestingly, bilateral inferior parietal lobules exhibited high connectivity in cognitive inhibition, aligning with the previous study on the parietal cortex’s contribution to inhibitory processes (Kolodny et al., 2017). Potential treatment strategies could be developed by targeting these regions for executive function disorders such as ADHD and inhibitory control disorders.
In addition to the “influencer” regions shared by all four subdomains, the bilateral superior parietal lobule and the right post-central gyrus (medial segment) are also identified as “influencers.” The superior parietal lobule is considered to play a pivotal role in numerous cognitive functions (Wang et al., 2014). The bilateral inferior parietal sulcus (IPS) plays an “influencer” role in cognitive inhibition, mental shifting, and working memory. This is similar to another study suggesting that IPS plays an essential role in executive functioning, particularly inhibition (Osada et al., 2019). Working memory appears to have a segment of the left post-central gyrus (DAN) as an “influencer,” similar to the findings on working memory among early Parkinson’s patients (Alsakaji et al., 2021). A segment of the right inferior parietal lobule (IPSL) appears to be an “influencer” of cognitive inhibition, which could be attributed to its involvement in visual attention (Corbetta and Shulman, 2002). As mentioned, “influencers” are more resilient to network reorganization. Therefore, these regions could be potential targets for the treatment of executive function deficits post-TBI, seizure disorders, or post-tumor resection.
Efficiency and communities
Unlike our hypothesis, each subdomain of the executive function showed similar network features except for inhibition. We also did not find increased connectivity in control-related regions or higher modularity for working memory. However, the cognitive inhibition graph exhibits slightly longer characteristic path lengths and greater overall region involvement than its counterparts, such as the parieto-occipital cortex (DAN) and temporal–parietal regions. These findings suggest cognitive inhibition involves a brain network organization that prioritizes specialized information transfer between regions, emphasizing the distinct nature of inhibitory control processes within the executive network. While this may result in less efficient overall network function, it may also reflect a more targeted and specialized approach to cognitive processing in inhibition.
Conversely, initiation, mental shifting, and working memory have shorter path lengths, which could minimize the metabolic cost associated with routing action potentials across axons and synaptic contacts and, hence, could provide faster, more direct, and less noisy information transfer (Bullmore and Sporns, 2009). Our analysis of graph metrics collectively implies a decentralized and modular functioning organization, wherein information processing occurs across distributed networks rather than being confined to specific localized regions.
Distinct subsystems of communities consistently emerge across four subdomains, prominently featuring medial parietal regions and the posterior medial frontal area across all four executive functioning subdomains. Echoing established findings, the medial prefrontal region exhibits a recurring presence during executive tasks yet notably delineates into two discernible subsystems: the dorsomedial prefrontal cortex (dmPFC) and ventromedial prefrontal cortex (vmPFC), particularly during mental shifting and working memory processes. This partition may stem from the dmPFC’s primary connections to the neocortex, while the vmPFC primarily interfaces with the limbic system (Jobson et al., 2021).
Vulnerability and resilience
Unlike our hypothesis, we did not find significant differences in their topological properties across tasks related to these four subdomains of executive functions. Our findings support a more distributed network topology for all four subdomains of executive functions. A distributed network, which does not rely heavily on single central components, offers resilience to random damage, as observed in the human brain’s robust response to lesions (Achard et al., 2006; Aerts et al., 2016), especially in a pediatric population (Guan et al., 2024). This resilience provides a framework for understanding the brain’s ability to maintain cognitive functions even after injury.
On the other hand, The vulnerability of hub regions to targeted damage highlights their critical role. Lesions in these hubs, such as those occurring in stroke or traumatic brain injury, can significantly impair executive functioning. Initiation stood out as it exhibits a low assortativity value, indicating a distributed network with balanced local and global connectivity but relying on hubs and communities (mPFC). Therefore, it could be more vulnerable than the other executive functions. Indeed, motivational and initiation deficits frequently occur in individuals with acquired brain injury, where prefrontal areas are more vulnerable (Palmisano et al., 2020). This understanding could guide clinical interventions, such as targeted behavioral therapy or deep brain stimulation, to restore function in affected regions (Aerts et al., 2016). Moreover, alterations in global network topology observed in conditions like Alzheimer’s disease, multiple sclerosis, and epilepsy suggest that these pathologies may function as “disconnection syndromes,” where disrupted connectivity underlies cognitive deficits (Guye et al., 2010). Understanding brain networks’ distributed and resilient nature can inform rehabilitation strategies to leverage intact pathways to compensate for lost functions. Further research is needed to explore how these network characteristics evolve across different conditions and stages of brain damage.
Recent studies have further highlighted the clinical implications of distributed network topology in executive functions. For instance, research shows that the topological properties of the frontoparietal network (FPN) and default mode network (DMN) are associated with executive function performance across the lifespan, with the DMN showing greater sensitivity to age-related changes (Menardi et al., 2024). Additionally, alterations in network topology have been observed in patients with mild cognitive impairment (MCI), suggesting that changes in network organization could serve as imaging markers for early diagnosis and intervention before Alzheimer’s disease onset (Xue et al., 2024).
Recent research has also emphasized the role of hub regions in neurological disorders. For example, in Parkinson’s disease, the spread of α-synuclein through connected brain regions leads to neuronal loss and network disruptions, with hub regions playing a significant role in this process (Frigerio et al., 2024). Understanding the involvement of hub regions is becoming increasingly important for clinical practice, as these hubs are critical for maintaining normal brain function and enabling complex behavior (Stam, 2024). These findings reinforce the importance of network topology in developing targeted interventions and rehabilitation strategies for various neurological conditions.
Limitations and future direction
Several limitations exist besides the small sample size and exclusive focus on the brain’s cortical areas. A key concern is that the cognitive paradigms used may not adequately capture the complex nuances of the four subdomains of executive functioning: initiation, inhibition, mental shifting, and working memory. While these paradigms provide valuable insights, they may not fully reflect the intricacies of these cognitive processes. This limitation underscores the need for future research to employ various cognitive tasks for a more thorough assessment of executive functioning.
Additionally, while graph analysis yields important insights into the dynamics of brain networks associated with cognitive tasks, several significant limitations exist. Reducing the brain into nodes and edges oversimplifies its inherent complexity, and the decisions made regarding the parcellation schemes and network construction parameters can significantly impact the results. Factors such as the spatial and temporal resolution of neuroimaging data, individual variability, and subjective thresholding methods introduce potential confounding variables. Furthermore, the cross-sectional nature of the analysis limits our understanding of how these dynamics change over time. Interpreting graph metrics concerning neural processes also requires caution due to their context-dependent nature. Addressing these limitations is critical for enhancing our understanding of brain network organization and functionality.
Despite these constraints, the study significantly contributes to our understanding of how brain networks support various cognitive processes. Future research should explore the subdomains of executive functioning with diverse cognitive paradigms, expand data collection to include subcortical activities and examine the complex interplay between brain networks and cognitive processes.
Conclusion
This study enhances our understanding of executive functioning by identifying key hubs, influencers, and communities while examining local and global network characteristics across four subdomains of executive function. Central areas such as the bilateral precuneus and the right medial prefrontal area are indispensable for integrating, transmitting information, and regulating activities within distributed networks, rendering them essential to executive functioning. Damage to these hubs can disrupt the executive function network.
Rehabilitation strategies can capitalize on neuroplasticity to preserve or enhance the functionality of these critical hubs. Techniques such as transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (tDCS) may target these regions to facilitate the restoration of connectivity and enhance cognitive outcomes. Furthermore, task-specific cognitive training designed to activate these hubs can promote network reorganization, enabling compensatory pathways to develop and improve recovery.
The distributed nature of executive function networks also suggests resilience in cognitive recovery. Even when a hub is compromised, strengthening other regions or connections within the network may mitigate deficits. Incorporating insights into hub functionality facilitates more targeted and effective rehabilitation, improving outcomes for individuals with brain injuries.
Moreover, this study elucidates distinct hubs and influencers specific to each executive function subdomain, underscoring the unique characteristics of these cognitive processes. Consistent with prior research, the bilateral precuneus is reaffirmed as a pivotal hub and influencer in executive functioning. Our finding on the central role of the right mPFC in executive functioning could point to a new direction in research in the right hemisphere.
The resilience of distributed brain networks to damage holds significant implications for conditions such as stroke and traumatic brain injury, guiding interventions aimed at preserving executive function. Further research is necessary to elucidate how network organization adapts to various types of brain damage, including epilepsy and neurodegenerative diseases, and to develop targeted therapeutic strategies that enhance recovery.
Author’s note
R Studio was used as an integrated development environment for R programming. The data were processed and analyzed using various R packages, including the `sqldf` package for SQL-like data manipulation (Grothendieck, 2007), the `brainGraph` package for brain network analysis (Watson, 2015), the `igraph` package for graph theory analysis (Csárdi et al., 2006), the `ggraph` package for advanced graph visualization (Pedersen, 2021), the `ggplot2` package for creating plots (Pedersen and RStudio, 2024), and the `brainconn` package for connectivity analysis (Chopra, n.d.). The analysis reports were generated using the `knitr` package for dynamic report generation (Xie, 2025), and tables were formatted using the `kableExtra` package (Zhu, 2024).
Data availability statement
Publicly available datasets were analyzed in this study. This data can be found at: https://pubmed.ncbi.nlm.nih.gov/34877370/.
Ethics statement
The studies involving humans were approved by IRB at Rotman Research Institute at Baycrest. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.
Author contributions
AD: Writing – original draft, Writing – review & editing.
Funding
The author(s) declare that no financial support was received for the research and/or publication of this article.
Conflict of interest
The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declare that no Gen AI was used in the creation of this manuscript.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnhum.2025.1525497/full#supplementary-material
ReferencesAchardS.SalvadorR.WhitcherB.SucklingJ.BullmoreE. (2006). A resilient, low-frequency, small-world human brain functional network with highly connected association cortical hubs. J. Neurosci.26, 63–72. doi: 10.1523/JNEUROSCI.3874-05.2006AertsH.FiasW.CaeyenberghsK.MarinazzoD. (2016). Brain networks under attack: Robustness properties and the impact of lesions. Brain139, 3063–3083. doi: 10.1093/brain/aww194, PMID: 27497487AlperB.BachB.Henry RicheN.IsenbergT.FeketeJ.-D. (2013). Weighted graph comparison techniques for brain connectivity analysis. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 483–492. Available online at: https://doi.org/10.1145/2470654.2470724 (Accessed September, 2023).AlsakajiR.JonesS.IbikO.CollettaK.LivengoodS.PapeT. B.. (2021). Working Memory Deficits Related to Brain Atrophy in Early Stage Parkinson’s Disease. Arch. Phys. Med. Rehabil.102:e38. doi: 10.1016/j.apmr.2021.07.572BassettD. S.BullmoreE. (2006). Small-world brain networks. Neuroscientist12, 512–523. doi: 10.1177/1073858406293182, PMID: 17079517BaumG. L.CiricR.RoalfD. R.BetzelR. F.MooreT. M.ShinoharaR. T.. (2017). Modular Segregation of Structural Brain Networks Supports the Development of Executive Function in Youth. Curr. Biol.27, 1561–1572.e8. doi: 10.1016/j.cub.2017.04.051, PMID: 28552358BernardF.LeméeJ.-M.Ter MinassianA.MeneiP. (2018). Right Hemisphere Cognitive Functions: From Clinical and Anatomic Bases to Brain Mapping During Awake Craniotomy Part I: Clinical and Functional Anatomy. World Neurosurg.118, 348–359. doi: 10.1016/j.wneu.2018.05.024, PMID: 29763748BilletteO. V.ZieglerG.AruciM.SchützeH.KizilirmakJ. M.RichterA.. (2022). Novelty-Related fMRI Responses of Precuneus and Medial Temporal Regions in Individuals at Risk for Alzheimer Disease. Neurology99, e775–e788. doi: 10.1212/WNL.0000000000200667, PMID: 35995589BucknerR. L.SepulcreJ.TalukdarT.KrienenF. M.LiuH.HeddenT.. (2009). Cortical Hubs Revealed by Intrinsic Functional Connectivity: Mapping, Assessment of Stability, and Relation to Alzheimer’s Disease. J. Neurosci.29, 1860–1873. doi: 10.1523/JNEUROSCI.5062-08.2009, PMID: 19211893BullmoreE. T.BassettD. S. (2011). Brain Graphs: Graphical Models of the Human Brain Connectome. Annu. Rev. Clin. Psychol.7, 113–140. doi: 10.1146/annurev-clinpsy-040510-143934, PMID: 21128784BullmoreE.SpornsO. (2009). Complex brain networks: Graph theoretical analysis of structural and functional systems. Nat. Rev. Neurosci.10, 186–198. doi: 10.1038/nrn2575, PMID: 19190637BullmoreE.SpornsO. (2012). The economy of brain network organization. Nat. Rev. Neurosci.13, 336–349. doi: 10.1038/nrn3214CavannaA. E.TrimbleM. R. (2006). The precuneus: A review of its functional anatomy and behavioural correlates. Brain: A. J. Neurol.129, 564–583. doi: 10.1093/brain/awl004, PMID: 16399806ChaiW. J.Abd HamidA. I.AbdullahJ. M. (2018). Working Memory From the Psychological and Neurosciences Perspectives: A Review. Front. Psychol.9:401. doi: 10.3389/fpsyg.2018.00401, PMID: 29636715ChopraS. (n.d.). brainconn: A R-package for plotting brain connectivity data • brainconn. Available online at: https://sidchop.github.io/brainconn/articles/brainconn.html (accessed May 20, 2025).CorbettaM.ShulmanG. L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nat. Rev. Neurosci.3, 201–215. doi: 10.1038/nrn755CsárdiG.NepuszT.TraagV.HorvátS.ZaniniF.NoomD.. (2006). igraph: Network Analysis and Visualization (p. 2.1.4) [Dataset]. Available online at: https://doi.org/10.32614/CRAN.package.igraph (Accessed September, 2023).DiamondA. (2013). Executive Functions. Annu. Rev. Psychol.64, 135–168. doi: 10.1146/annurev-psych-113011-143750, PMID: 23020641EngstromM.LandtblomA.-M.KarlssonT. (2013). Brain and effort: Brain activation and effort-related working memory in healthy participants and patients with working memory deficits. Front. Hum. Neurosci.7:140. doi: 10.3389/fnhum.2013.00140, PMID: 23616756FarahaniF. V.KarwowskiW.D’EspositoM.BetzelR. F.DouglasP. K.SobczakA. M.. (2022). Diurnal variations of resting-state fMRI data: A graph-based analysis. NeuroImage256:119246. doi: 10.1016/j.neuroimage.2022.119246, PMID: 35477020FarahaniF. V.KarwowskiW.LighthallN. R. (2019). Application of Graph Theory for Identifying Connectivity Patterns in Human Brain Networks: A Systematic Review. Front. Neurosci.13:585. doi: 10.3389/fnins.2019.00585, PMID: 31249501FornitoA.ZaleskyA.BullmoreE. T. (2016). Fundamentals of brain network analysis. Amsterdam: Elsevier/Academic Press.FriedmanN. P.RobbinsT. W. (2022). The role of prefrontal cortex in cognitive control and executive function. Neuropsychopharmacology47, 72–89. doi: 10.1038/s41386-021-01132-0, PMID: 34408280FrigerioI.BroedersT. A. A.LinC.-P.BouwmanM. M. A.KoubiyrI.BarkhofF.. (2024). Pathologic Substrates of Structural Brain Network Resilience and Topology in Parkinson Disease Decedents. Neurology103:e209678. doi: 10.1212/WNL.0000000000209678, PMID: 39042844GrothendieckG. (2007). sqldf: Manipulate R Data Frames Using SQL (p. 0.4-11) [Dataset]. Available online at: https://doi.org/10.32614/CRAN.package.sqldf (Accessed September, 2023).GuanX.HuB.ZhengW.ChenN.LiX.HuC.. (2024). Changes on Cognition and Brain Network Temporal Variability After Pediatric Neurosurgery. Neurosurgery96, 555–567. doi: 10.1227/neu.0000000000003124, PMID: 39023270GuseB.FalkaiP.WobrockT. (2010). Cognitive effects of high-frequency repetitive transcranial magnetic stimulation: A systematic review. J. Neural Transm. (Vienna)117, 105–122. doi: 10.1007/s00702-009-0333-7, PMID: 19859782GuyeM.BettusG.BartolomeiF.CozzoneP. J. (2010). Graph theoretical analysis of structural and functional connectivity MRI in normal and pathological brain networks. MAGMA23, 409–421. doi: 10.1007/s10334-010-0205-z, PMID: 20349109HeY.EvansA. (2010). Graph theoretical modeling of brain connectivity. Curr. Opin. Neurol.23, 341–350. doi: 10.1097/WCO.0b013e32833aa567, PMID: 20581686HricD.DarstR. K.FortunatoS. (2014). Community detection in networks: Structural communities versus ground truth. Phys. Rev. E90:062805. doi: 10.1103/PhysRevE.90.062805, PMID: 25615146HuffM. J.BalotaD. A.MinearM.AschenbrennerA. J.DuchekJ. M. (2015). Dissociative Global and Local Task-Switching Costs Across Younger Adults, Middle-Aged Adults, Older Adults, and Very Mild Alzheimer Disease Individuals. Psychol. Aging30, 727–739. doi: 10.1037/pag0000057, PMID: 26652720JaeggiS. M.BuschkuehlM.PerrigW. J.MeierB. (2010). The concurrent validity of the N-back task as a working memory measure. Memory18, 394–412. doi: 10.1080/09658211003702171, PMID: 20408039JinG.LiK.QinY.ZhongN.ZhouH.WangZ.. (2012). fMRI study in posterior cingulate and adjacent precuneus cortex in healthy elderly adults using problem solving task. J. Neurol. Sci.318, 135–139. doi: 10.1016/j.jns.2012.02.032, PMID: 22560874JobsonD. D.HaseY.ClarksonA. N.KalariaR. N. (2021). The role of the medial prefrontal cortex in cognition, ageing and dementia. Brain. Communications3:fcab125. doi: 10.1093/braincomms/fcab125, PMID: 34222873KolodnyT.MevorachC.ShalevL. (2017). Isolating response inhibition in the brain: Parietal versus frontal contribution. Cortex88, 173–185. doi: 10.1016/j.cortex.2016.12.012, PMID: 28142026LawyerG. (2015). Understanding the influence of all nodes in a network. Sci. Rep.5:8665. doi: 10.1038/srep08665, PMID: 25727453LeechR.SharpD. J. (2014). The role of the posterior cingulate cortex in cognition and disease. Brain137, 12–32. doi: 10.1093/brain/awt162, PMID: 23869106LongJ.SongX.WangY.WangC.HuangR.ZhangR. (2022). Distinct neural activation patterns of age in subcomponents of inhibitory control: A fMRI meta-analysis. Front. Aging Neurosci.14:938789. doi: 10.3389/fnagi.2022.938789, PMID: 35992590MedagliaJ. D. (2017). Graph Theoretic Analysis of Resting State fMRI. Neuroimaging Clin. N. Am.27, 593–607. doi: 10.1016/j.nic.2017.06.008, PMID: 28985931MedagliaJ. D.LynallM.-E.BassettD. S. (2015). Cognitive Network Neuroscience. J. Cogn. Neurosci.27, 1471–1491. doi: 10.1162/jocn_a_00810, PMID: 25803596MenardiA.SpoaM.VallesiA. (2024). Brain topology underlying executive functions across the lifespan: Focus on the default mode network. Front. Psychol.15:1441584. doi: 10.3389/fpsyg.2024.1441584, PMID: 39295768MenonV. (2011). Large-scale brain networks and psychopathology: A unifying triple network model. Trends Cogn. Sci.15, 483–506. doi: 10.1016/j.tics.2011.08.003, PMID: 21908230MenonV.D’EspositoM. (2022). The role of PFC networks in cognitive control and executive function. Neuropsychopharmacology47, 90–103. doi: 10.1038/s41386-021-01152-w, PMID: 34408276Morein-ZamirS.Simon JonesP.BullmoreE. T.RobbinsT. W.ErscheK. D. (2013). Prefrontal Hypoactivity Associated with Impaired Inhibition in Stimulant-Dependent Individuals but Evidence for Hyperactivation in their Unaffected Siblings. Neuropsychopharmacology38, 1945–1953. doi: 10.1038/npp.2013.90, PMID: 23609131NemethD. G.ChustzK. M. (2020). Chapter 6—Executive functions defined. In NemethD. G.GlozmanJ. (Eds.), Evaluation and Treatment of Neuropsychologically Compromised Children (pp. 107–120). Academic Press. Available online at: https://doi.org/10.1016/B978-0-12-819545-1.00006-0 (Accessed September, 2023).NiendamT. A.LairdA. R.RayK. L.DeanY. M.GlahnD. C.CarterC. S. (2012). Meta-analytic evidence for a superordinate cognitive control network subserving diverse executive functions. Cogn. Affect. Behav. Neurosci.12, 241–268. doi: 10.3758/s13415-011-0083-5, PMID: 22282036OsadaT.OhtaS.OgawaA.TanakaM.SudaA.KamagataK.. (2019). An Essential Role of the Intraparietal Sulcus in Response Inhibition Predicted by Parcellation-Based Network. J. Neurosci. Off. J. Soc. Neurosci.39, 2509–2521. doi: 10.1523/JNEUROSCI.2244-18.2019, PMID: 30692225PaldinoM. J.ZhangW.ChuZ. D.GolrizF. (2016). Metrics of brain network architecture capture the impact of disease in children with epilepsy. Neuroimage Clin.13, 201–208. doi: 10.1016/j.nicl.2016.12.005, PMID: 28003958PalmisanoS.FasottiL.BertensD. (2020). Neurobehavioral Initiation and Motivation Problems After Acquired Brain Injury. Front. Neurol.11:23. doi: 10.3389/fneur.2020.00023, PMID: 32153486PedersenT. L. (2021). ggraph: An Implementation of Grammar of Graphics for Graphs and Networks (p. 2.2.1) [Dataset]. doi: 10.32614/CRAN.package.ggraphPedersenT. L.RStudio. (2024). ggraph: An Implementation of Grammar of Graphics for Graphs and Networks (Version 2.2.1) [Computer software]. Available online at: https://cran.r-project.org/web/packages/ggraph/index.html (Accessed September, 2023).PowerJ. D.CohenA. L.NelsonS. M.WigG. S.BarnesK. A.ChurchJ. A.. (2011). Functional network organization of the human brain. Neuron72, 665–678. doi: 10.1016/j.neuron.2011.09.006, PMID: 22099467RaichleM. E. (2015). The brain’s default mode network. Annu. Rev. Neurosci.38, 433–447. doi: 10.1146/annurev-neuro-071013-014030, PMID: 25938726Ramos-NuñezA. I.Fischer-BaumS.MartinR. C.YueQ.YeF.DeemM. W. (2017). Static and Dynamic Measures of Human Brain Connectivity Predict Complementary Aspects of Human Cognitive Performance. Front. Hum. Neurosci.11:420. doi: 10.3389/fnhum.2017.00420, PMID: 28883789ReijneveldJ. C.PontenS. C.BerendseH. W.StamC. J. (2007). The application of graph theoretical analysis to complex networks in the brain. Clin. Neurophysiol.118, 2317–2331. doi: 10.1016/j.clinph.2007.08.010, PMID: 17900977RieckJ. R.BaracchiniG.NicholD.AbdiH.GradyC. L. (2021). Dataset of functional connectivity during cognitive control for an adult lifespan sample. Data Brief39:107573. doi: 10.1016/j.dib.2021.107573, PMID: 34877370RubinovM.SpornsO. (2010). Complex network measures of brain connectivity: Uses and interpretations. NeuroImage52, 1059–1069. doi: 10.1016/j.neuroimage.2009.10.003, PMID: 19819337Salgado-PinedaP.Rodriguez-JimenezR.Moreno-OrtegaM.DompabloM.Martínez de AragónA.SalvadorR.. (2021). Activation and deactivation patterns in schizophrenia during performance of an fMRI adapted version of the stroop task. J. Psychiatr. Res.144, 1–7. doi: 10.1016/j.jpsychires.2021.09.039, PMID: 34583085SchaeferA.KongR.GordonE. M.LaumannT. O.ZuoX.-N.HolmesA. J.. (2018). Local-Global Parcellation of the Human Cerebral Cortex from Intrinsic Functional Connectivity MRI. Cereb. Cortex28, 3095–3114. doi: 10.1093/cercor/bhx179, PMID: 28981612SpornsO.BetzelR. F. (2016). Modular Brain Networks. Annu. Rev. Psychol.67, 613–640. doi: 10.1146/annurev-psych-122414-033634, PMID: 26393868StamC. J. (2024). Hub overload and failure as a final common pathway in neurological brain network disorders. Netw. Neurosci.8, 1–23. doi: 10.1162/netn_a_00339, PMID: 38562292StanleyM. L.SimpsonS. L.DagenbachD.LydayR. G.BurdetteJ. H.LaurientiP. J. (2015). Changes in Brain Network Efficiency and Working Memory Performance in Aging. PLoS One10:e0123950. doi: 10.1371/journal.pone.0123950, PMID: 25875001SunL.LiangX.DuanD.LiuJ.ChenY.WangX.. (2022). Structural insight into the individual variability architecture of the functional brain connectome. NeuroImage259:119387. doi: 10.1016/j.neuroimage.2022.119387, PMID: 35752416Thomas YeoB. T.KrienenF. M.SepulcreJ.SabuncuM. R.LashkariD.HollinsheadM.. (2011). The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J. Neurophysiol.106, 1125–1165. doi: 10.1152/jn.00338.2011, PMID: 21653723van den HeuvelM. P.StamC. J.BoersmaM.Hulshoff PolH. E. (2008). Small-world and scale-free organization of voxel-based resting-state functional connectivity in the human brain. NeuroImage43, 528–539. doi: 10.1016/j.neuroimage.2008.08.010, PMID: 18786642VerbruggenF.LoganG. D. (2008). Automatic and Controlled Response Inhibition: Associative Learning in the Go/No-Go and Stop-Signal Paradigms. J. Exp. Psychol. Gen.137, 649–672. doi: 10.1037/a0013170, PMID: 18999358WangJ.YangY.FanL.XuJ.LiC.LiuY.. (2014). Convergent functional architecture of the superior parietal lobule unraveled with multimodal neuroimaging approaches. Hum. Brain Mapp.36, 238–257. doi: 10.1002/hbm.22626, PMID: 25181023WatsonC. G. (2015). brainGraph: Graph Theory Analysis of Brain MRI Data (p. 3.1.0) [Dataset]. Available online at: https://doi.org/10.32614/CRAN.package.brainGraph (Accessed September, 2023).WittS. T.Van Ettinger-VeenstraH.SaloT.RiedelM. C.LairdA. R. (2021). What Executive Function Network is that? An Image-Based Meta-Analysis of Network Labels. Brain Topogr.34, 598–607. doi: 10.1007/s10548-021-00847-z, PMID: 33970388XieY. (2025). Knitr.pdf. Available online at: https://cran.r-project.org/web/packages/knitr/knitr.pdf (Accessed September, 2023).XuP.ChenA.LiY.XingX.LuH. (2019). Medial prefrontal cortex in neurological diseases. Physiol. Genomics51, 432–442. doi: 10.1152/physiolgenomics.00006.2019, PMID: 31373533XueC.ZhengD.RuanY.CaoX.ZhangX.QiW.. (2024). Reorganized brain functional network topology in stable and progressive mild cognitive impairment. Front. Aging Neurosci.16:1467054. doi: 10.3389/fnagi.2024.1467054, PMID: 39624168YeagerB. E.BrussJ.DuffauH.HerbetG.HwangK.TranelD.. (2022). Central precuneus lesions are associated with impaired executive function. Brain Struct. Funct.227, 3099–3108. doi: 10.1007/s00429-022-02556-0, PMID: 36087124YeoB. T. T.KrienenF. M.EickhoffS. B.YaakubS. N.FoxP. T.BucknerR. L.. (2015). Functional Specialization and Flexibility in Human Association Cortex. Cereb. Cortex25, 3654–3672. doi: 10.1093/cercor/bhu217, PMID: 25249407ZelazoP. D.CraikF. I. M.BoothL. (2004). Executive function across the life span. Acta Psychol.115, 167–183. doi: 10.1016/j.actpsy.2003.12.005ZhuH. (2024). kableExtra.pdf. Available online at: https://cran.r-project.org/web/packages/kableExtra/kableExtra.pdf (Accessed September, 2023).