Brain Plasticity xx (20xx) x–xx 1

DOI 10.3233/BPL-190081
IOS Press

1 Modulation of Distinct Intrinsic Resting

2 State Brain Networks by Acute Exercise
Bouts of Differing Intensity

f
3

roo
4 Angelika Schmitta,∗ , Neeraj Upadhyayb , Jason Anthony Martina , Sandra Rojasc
5 Heiko Klaus Strüderc and Henning Boeckera
a Functional Neuroimaging Group, Department of Radiology, University Hospital Bonn, Bonn, Germany

rP
6

7
b DZNE, German Center for Neurodegenerative Diseases, Bonn, Germany
8
c Institute of Movement and Neurosciences, German Sport University Cologne, Cologne, Germany

tho
9 Abstract.
10 Acute exercise bouts alter resting state functional connectivity (rs-FC) within cognitive, sensorimotor, and affective
11
Au
networks, but it remains unknown how these effects are influenced by exercise intensity.
12 Twenty-five male athletes underwent individual fitness assessments using an incremental treadmill test. On separate days,
13 they performed ‘low’ (35% below lactate threshold) and ‘high’ (20% above lactate threshold) intensity exercise bouts of
14 30 min. Rs-fMRI and Positive and Negative Affect Scale (PANAS) were acquired before and after each exercise bout.
15 Networks of interest were extracted from twenty-two participants (3 dropouts). Pre-to-post changes and between conditions
16 effects were evaluated using FSL’s randomise by applying repeated measures ANOVA. Results were reported at p < 0.05,
d

17 corrected for multiple comparisons using threshold free cluster enhancement.

18 PANAS revealed a significant increase in positive mood after both exercise conditions. Significant effects were observed
cte

19 between conditions in the right affective and reward network (ARN), the right fronto parietal network (FPN) and the senso-
20 rimotor network (SMN). Pre-to-post comparisons after ‘low’ exercise intensity revealed a significant increase in rs-FC in the
21 left and right FPN, while after ‘high’-intensity exercise rs-FC decreased in the SMN and the dorsal attention network (DAN)
22 and increased in the left ARN.
23 Supporting recent findings, this study is the first to report distinct rs-FC alterations driven by exercise intensity: (i) Increased
rre

24 rs-FC in FPN may indicate beneficial functional plasticity for cognitive/attentional processing, (ii) increased rs-FC in ARN
25 may be linked to endogenous opioid-mediated internal affective states. Finally, (iii) decreased rs-FC in the SMN may signify
26 persistent motor fatigue. The distinct effects on rs-FC fit with theories of transient persistent network alterations after acute
27 exercise bouts that are mediated by different exercise intensities and impact differentially on cognitive/attentional or affective
28 responses.
co

29 Keywords: Physical activity, aerobic exercise, rs-fMRI, resting state network, exercise intensity
Un

INTRODUCTION 30

Acute physical exercise bouts are known to 31

∗ Correspondence to: Angelika Schmitt, Functional Neu- enhance affective, cognitive and attentional pro- 32
roimaging Group, Department of Radiology, University Hospital
Bonn – Building 7, Sigmund-Freud-Str. 25, 53127 Bonn, Ger-
cesses in a range of minutes to hours after exercise 33

many. Tel.: +49 228 287 15970; Fax: +49 228 287 14457; E-mail: bouts. One possible mechanism by which such tran- 34

angelika.schmitt@ukbonn.de. sient improvements appear to be mediated is an 35

ISSN 2213-6304/19/$35.00 © 2019 – IOS Press and the authors. All rights reserved
This article is published online with Open Access and distributed under the terms of the Creative Commons Attribution Non-Commercial License (CC BY-NC 4.0).
 2 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

36 alteration of coherent neuronal activity that persists more than 1 min, more intense exercise (very hard) 88

37 after exercise. Such coherent neuronal activity is appears to be more beneficial than lighter intensity 89

38 termed “resting state functional connectivity” (rs-FC) exercise [7], whereas, when cognitive testing is per- 90

39 and brain regions that are functionally connected via formed immediately after the exercise bout, lighter 91

40 distinct temporo-spatial properties are referred to as intensity exercise (very light, light, moderate) appears 92

41 “functional networks”. Resting state functional mag- to be more beneficial than high intensity exercise 93

42 netic resonance imaging (rs-fMRI) is a potential tool [7]. Chang and colleagues found that distinct types 94

43 to measure such transient coherent neuronal activity of brain executive control functions tested with the 95

f
44 alterations that persist after acute exercise bouts. ‘Tower of London’ task are differentially improved 96

roo
45 Most of the previous rs-fMRI literature was, how- by different intensities of acute exercise [8, 9]. 97

46 ever, not focused on investigating the transient effects Studies using MRI to investigate the neuronal 98

47 of acute exercise bouts, but rather on long-term underpinnings of transient effects of acute exercise 99

48 effects of repeated exercise training over several bouts are rare. To the best of our knowledge, there 100

49 months [1–4]: Stronger rs-FC was reported between have been only two studies comparing cerebral blood 101

rP
50 the right parahippocampal gyrus and motor, sensory, flow (CBF) after different intensities of acute exer- 102

51 and mood regulating areas after 16 weeks of train- cise [10, 11], showing exercise intensity dependent 103

52 ing [4]. On the other hand, 12 weeks of exercise changes in CBF. One of these studies found decreases 104

53 induced increased FC between the default mode net- in CBF after moderate exercise intensity in the ante- 105

tho
54 work (DMN) and primary motor areas [3], eventually rior cingulate, bilateral thalamus, and bilateral insula, 106

55 indicating improved motor performance. Con- but not after low exercise intensity [11]. The other 107

56 versely, another study found no significant changes study found intensity dependent changes in CBF in 108

57 in intrinsic brain activity after a six months of aerobic the right somatosensory cortex in individuals with 109

58 exercise [2]. cerebrovascular disease. The CBF in the sensori- 110

Au
59 Less is known about the transient effects of acute motor cortices was reduced following low-intensity 111

60 exercise bouts on rs-FC: A recently published study exercise and increased following moderate-intensity 112

61 using rs-fMRI in both older and younger healthy exercise [10]. A recently published study by our own 113

62 volunteers reported a modulation of rs-FC after a group comparing the effect of low and high intensity 114

63 moderate exercise session on a bicycle ergometer exercise bouts on emotional face processing could 115

compared to a passive control condition [5]. The show distinct changes in the brain during fearful face
d

64 116

65 authors found a significant increase in rs-FC in the processing [12]: Low-intensity exercise resulted in 117
cte

66 affective-reward network (ARN), the hippocampal reduced brain activations in PCC/precuneus and high 118

67 and the right executive control network (ECN) after intensity exercise was associated with reduced fear- 119

68 exercising at 65% of their HRmax (maximum heart related activity in ventral basal ganglia structures. 120

69 rate). Another study investigating the effects of a Taken together, the mentioned behavioral and imag- 121

70 single moderate aerobic exercise session in young ing studies indicate distinct changes in cognitive, 122
rre

71 healthy subjects found transient changes in FC in affective and brain function, dependent on exercise 123

72 the auditory resting state network (RSN), the sen- intensity. 124

73 sorimotor network (SMN) and the thalamic-caudate In order to determine how the intensity of acute 125

74 RSN [6]. Taken together, these initial findings in exercise transiently affects the intrinsic neuronal 126
co

75 the acute setting suggest that single exercise bouts activity at rest, we designed this study to assess 127

76 induce transient changes in intrinsic brain activity that rs-FC changes after defined ‘low’ and ‘high’ exer- 128

77 can be captured by rs-fMRI. Although, these stud- cise intensities (adapted to the individual fitness 129

78 ies generally report enhanced FC of intrinsic brain level of each participant). These two forms of exer- 130
Un

79 networks, it still remains unknown how these effects cise intensity (‘low’ and ‘high’) were chosen in this 131

80 are influenced by the intensity of the exercise bout. study in order to fill the open gap (i.e. effects of 132

81 It is relevant to point out, based on meta-analytical different exercise intensities on rs-fMRI) in the liter- 133

82 data (independent of the imaging context) that the ature. We predicted that ‘high’-intensity (20% above 134

83 effect on cognitive performance after exercise is influ- lactate threshold) exercise would alter the FC at 135

84 enced by the intensity of the exercise and notably rest differently than ‘low’-intensity (35% below lac- 136

85 also by the time between the end of the exercise tate threshold) exercise. This could be helpful in 137

86 bout and the start of the cognitive testing [7]. When understanding the relation between exercise inten- 138

87 cognitive performance is acquired with a delay of sity and to identify interventions that specifically 139
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 3

140 enhance the FC in different cognitive and affective Table 1

141 networks. Participants demographics
Variable N = 22
142 MATERIALS AND METHODS Age (years) 27.1 ± 4.0
Education (years) 17.9 ± 2.7
Height (cm) 182 ± 6
143 Participants Weight (kg) 78.3 ± 6.8
BMI (kg/m2 ) 23.7 ± 1.4
144 Twenty-five right-handed and trained male sub- Resting HR (bpm) 75.9 ± 15.1

f
HRmax (bpm) 194.2 ± 7.2
145 jects (exercising at least 3 times/week for 45 minutes
16.1 ± 1.3

roo
peak running speed (km/h)
146 for the past two years) were recruited. Require-
Note: Data are presented as mean (±standard deviation).
147 ments for participation in the study were no current BMI = body mass index; bpm = beats per minute; HRmax =
148 or prior history of head injury, no neurological or maximum heart rate.
149 psychiatric illness, no orthopedic or general health
150 problems that may have prohibited physical activity. Bonn (Lfd. 340/13), according to national legislation 189

rP
151 Participants with MRI-contraindications (i.e. metal and the Declaration of Helsinki. 190

152 and/or electronic implants, claustrophobia, etc.) were

153 excluded. To assemble a set of descriptive charac- Experimental procedure 191

154 teristics (e.g. age, education, handedness) several

tho
155 questionnaires were collected. Additionally, psychi- The study consisted of 4 examination days: a pre- 192

156 atric questionnaires were carried out, including the experimental session and three experimental sessions 193

157 Mini International Neuropsychiatric Interview, Ger- that were separated by at least 2 days (maximum 12 194

158 man Version 5.0.0 (MINI) [13], the State and Trait weeks; time period for completing the whole study 195

159 Anxiety Inventory (STAI) [14], and the Beck Depres- including all three exercise conditions: 58 ± 47 days). 196
Au
160 sion Inventory (BDI) [15]. As this manuscript focuses exclusively on the com- 197

161 Three of 25 subjects had to be excluded from our parison of standardized and metabolically defined 198

162 final analyses due to injuries that prohibited exer- exercise intensities, data from the self-selected con- 199

163 cise (caused by private activities, N = 1) or dropouts dition will not be presented here and will be reported 200

164 (N = 2), resulting in a final sample size of N = 22 elsewhere. The time period for completing only 201

subjects. According to the Edinburgh Handedness the ‘low’ and the ‘high’ exercise condition was
d

165 202

166 Inventory (EHI) [16], N = 21 subjects were right 37 ± 46 days. Subjects were instructed not to change 203

dominant (mean laterality quotient 83.4 ± 15.1) and

cte

167 their training behavior and to continue their training 204

168 N = 1 was ambidextrous (laterality quotient: 21.74). as usual throughout the study (in order to main- 205

169 None of the 22 participants showed results indicat- tain their fitness level). During the pre-experimental 206

170 ing psychiatric diseases (MINI). At study entrance, session, participants completed a screening by a 207

171 the volunteers’ mean BDI score was 1.1 ± 1.9, physician at the German Sport University Cologne, 208
rre

172 indicating that all participants scored below the including medical history, stethoscopy, and a 12 209

173 threshold value for clinically relevant depression channel resting electrocardiogram (ECG: MAC 1200 210

174 (score ≤9: no depression). STAI trait scores were ST, GE Medical Systems Information Technologies 211

175 on average 30.8 ± 6.2 (range: 20 = not being afraid GmbH, Freiburg, Germany). In the following fitness 212
co

176 to 80 = maximum intensity of anxiety). BDI and assessment in form of an incremental treadmill test 213

177 STAI state scores did not differ significantly between (without spiroergometry), the subjects’ physical per- 214

178 examination days. The mean verbal intelligence quo- formance was examined. The incremental treadmill 215

179 tient estimated by a German vocabulary test was test started at a speed of 5 km/h, with a 1% gradient set 216
Un

180 106.4 ± 9.8 [17], indicating a normal verbal intelli- throughout (Woodway PPS Med, Woodway GmbH, 217

181 gence distribution in this cohort. For a summary of Germany). Each stage lasted 3 minutes and running 218

182 the volunteer demographics see Table 1. speed was increased by 1 km/h per stage. After each 219

183 Participants were informed about the aim of the stage, a blood sample (20 ␮L) was taken from the 220

184 study and received a written detailed explanation earlobe and used to measure blood lactate concen- 221

185 regarding all tests, potential discomforts, risks, and tration. Additionally, at the end of each stage heart 222

186 procedures employed in the investigation, after which rate (POLAR, Kempele, Finland) and a rating of per- 223

187 they gave written consent. The study was approved by ceived exertion (RPE; via the BORG Scale) [18] were 224

188 the local Ethics Committee of the University Hospital recorded. The RPE scale ranges from “6 – no exertion 225
 4 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

226 at all / extremely light” to “20 – maximal exertion”. prior to examination and to refrain from drinking 278

227 Participants performed the incremental test until voli- caffeine and eating two hours before the experiment. 279

228 tional exhaustion. Exhaustion was considered with

229 the attainment of at least two of the following crite- MRI acquisition 280

230 ria: high levels of blood lactate (8–10 mmol·L-1);

231 a perceived rate of exertion of ≥18 and/or a HR All MRI examinations were performed at the 281

232 of ± 10 (beats per minute) of age-predicted maxi- Department of Radiology, University Hospital 282

233 mum (220 - age) [19]. Blood lactate samples were Bonn, with a 3T clinical MRI System (Ingenia 283

f
234 analyzed using the BIOSEN C-Line lactate analyzer 5.1.7, Philips Healthcare, Best, The Nether- 284

roo
235 (EKF-diagnostic GmbH, Barleben, Germany). lands), equipped with an 8-channel head coil. 285

236 To ensure comparable relative exercise intensities Rs-fMRI data were collected using a T2*-weighted 286

237 across participants, the lactate threshold (LT) was gradient-echo, echo-planar imaging (EPI) protocol 287

238 individually determined from a plot of blood lactate sensitive to the blood oxygenation level-dependent 288

239 concentration against running speed. LT was defined (BOLD) contrast with 250 volumes: TR = 2595 ms, 289

rP
240 as the first workload at which there was an increase TE = 35 ms, FOV = 230 × 230 × 147 mm, image 290

241 in blood lactate concentration of 0.4 mmol/l above matrix = 64 × 64, flip angle = 90◦ , 3.59 × 3.59 × 3.59 291

242 the baseline lactate [20]. During the three experimen- mm3 voxels, interleaved acquisition (1,3,5,7. . .41; 292

243 tal sessions, different 30 minutes exercise bouts (i.e. 2,4,6,8. . .40; first odd, then even) of 41 contiguous 293

tho
244 ‘low’, ‘high’ and ‘self-selected’) were performed on axial slices of 3.59 mm thickness. These images 294

245 a treadmill. Each exercise session consisted of a five were obtained parallel to the anterior-posterior 295

246 minutes warm up at 5 km/h, followed by the main 30 commissure plane with no inter-slice gap. For all 296

247 minutes exercise and ended with a five minutes cool- rs-fMRI scans, we instructed participants to lay still 297

248 down period at 5 km/h. Before exercising, resting with their eyes closed and to stay awake without 298
Au
249 blood pressure, HR, and blood lactate were recorded. thinking about anything in particular for 11 minutes. 299

250 During exercise, speed, HR, blood lactate, physical For anatomical reference a 3D-T1 weighted data 300

251 exertion were monitored. HR was recorded every set was acquired within each run with the following 301

252 minute, and blood lactate and physical exertion were parameters: slice orientation: sagittal, acquisition 302

253 recorded every 5 minutes. Exercise intensities were matrix: 256 × 256, acquired voxel size: 1 × 1 × 1 303

randomized across the three experimental sessions mm3 , sequence type: 3D FFE, TR: 7.6 ms, TE:
d

254 304

255 [21]. The ‘low’ exercise intensity was performed at 3.9 ms, flip angle: 15◦ , total scan duration: 4:39 min. 305
cte

256 35% under LT to ensure a very easy exercise con-

257 dition that challenges primarily the aerobic energy MRI data preprocessing 306

258 supply. Since an exercise intensity with a similar per-

259 centage above LT would not be feasible for a duration First, an average T1 image of the acquired struc- 307

260 of 30 min, exercise intensity for the ‘high’ condi- tural images was created using the paired longitudinal 308
rre

261 tion was set at 20% above LT. Directly pre and post registration [23] in SPM12 (Statistical Parametric 309

262 exercise the mood questionnaire Positive and Neg- Mapping version; Wellcome Department of Imaging 310

263 ative Affect Schedule (PANAS) [22] was acquired. Neuroscience, London, UK) implemented in Matlab 311

264 Rs-fMRI scans were performed on all three exami- (The Mathworks Inc., Sherborn, MA, USA). The 312

nation days pre and approximately 18 ± 5 min post

co

265 averaged T1 image was then skull-stripped using 313

266 exercise bouts. To account for possible prolonged FSL’s BET (Brain Extraction Technique) algorithm 314

267 effects of exercise on resting HR, HR was monitored [24]. Functional data from all time points were 315

268 during MRI scans using a pulse oximetry transducer evaluated using FSL toolbox (version 5.0.9; FMRIB 316
Un

269 placed on the left index finger (BIOPAC Systems Software Library, http://www.fmrib.ox.ac.uk/fsl). 317

270 Inc., Goleta, CA, USA). Data was recorded with Soft- The preprocessing of the individuals’ rs-fMRIs 318

271 ware of Biopac (AcqKnowledge 4.1). In addition to included motion correction, brain extraction, and 319

272 the rs-fMRI, we also acquired other fMRI sequences spatial smoothing with 6 mm FWHM (full width at 320

273 (B0, CASL, task-based, structural). All participants half maximum). Due to the interleaved acquisition 321

274 were instructed to avoid exercise training 24 hours mode and the minimal effects of slice timing on 322

275 prior each testing session to minimize the acute influ- rs-fMRI (as previously described in the literature 323

276 ences of exercise prior to assessments. They were also [25]), we did not perform slice time correction. 324

277 instructed to refrain from alcohol intake 24 hours Functional images from each subject were registered 325
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 5

326 to their average T1 structural image and then acute exercise on rs-FC, we selected the following 12 376

327 optimized using the boundary-based registration components (involved in affective processing, motor 377

328 approach. Later, structural images were transformed function, attention, executive and cognitive function) 378

329 to standard space using a non-linear registration by visually inspecting every component and matching 379

330 tool (FNIRT). The resulting warp fields were then them with previous defined network templates [28]. 380

331 applied to the functional images. An additional The selected networks were: 1) Default Mode Net- 381

332 motion correction was performed to remove the work (DMN); 2) Sensorimotor Network (SMN); 3) 382

333 physiological noise components (preserving the Executive Control Network (ECN); 4) Salience Net- 383

f
334 autocorrelations) by using a robust ICA-based work (SAL); 5) Left Affective and Reward Network 384

roo
335 strategy for automatic removal of motion artifacts (LARN); 6) Right Affective and Reward Network 385

336 (ICA-AROMA) [26]. This approach removes (RARN); 7) Left Fronto Parietal Network (LFPN); 386

337 (motion-related) spurious noise from rs-fMRI data 8) Right Fronto Parietal Network (RFPN); 9) Dorsal 387

338 to a larger extent than the regression approach by Attention Network (DAN), 10) Basal Ganglia Net- 388

339 using 24 motion parameters or spike regression. To work (BGN), 11) Cerebellar Network (CBN), and 389

rP
340 avoid the further loss of signals of interest, we did (12) Hippocampal Network (HIP) (Fig. 1). In order 390

341 not regress out white matter (WM) and cerebrospinal to create the individual FC maps of each selected 391

342 fluid (CSF) in our main analysis; rather we used RSN per condition, we applied the dual regression 392

343 a gray matter (GM) mask to extract the indepen- approach [29, 30]. This is a two stage process: In 393

tho
344 dent components, later on [27]. The gray matter the first stage, we regressed the averaged group map 394

345 (GM) mask was created from the GM segments of each network to the individual denoised filtered 395

346 available in the segmentation toolbox of SPM12 functional data to get the timeseries - one timeseries 396

347 (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/). per group-ICA component. In the second stage, we 397

348 We thresholded GM segment at 0.2 to further remove further generated the FC maps of each groupICA 398
Au
349 overlapping voxels with WM and CSF and binarized component for each individual subject from the 399

350 the mask. We additionally performed a second regressed time series. Later on, these FC maps of 400

351 analysis with nuisance regression before performing each ICA per individual subject were inserted in the 401

352 gICA. Results of this analysis are presented in the statistical analysis to evaluate the effect of exercise 402

353 supplement. condition. 403

Finally, we obtained denoised filtered functional

d

354

355 data by band pass filtering the non aggressive

Statistical analysis 404
cte

356 denoised data from the ICA-AROMA to remove fre-

357 quencies other than 0.009 to 0.1 Hz. Behavioral data 405

All statistical analyses of behavioral data were per- 406

358 Creating functional connectivity maps formed using SPSS 25 (SPSS Inc., Chicago, Illinois). 407

Using the Shapiro-Wilk-Test, data was tested for nor-

rre

408
359 The denoised filtered functional data from every mal distribution. Tests revealed that the data of the 409
360 participant pre and post exercise of the ‘high’
PANAS negative affect scale were not normally dis- 410
361 and ‘low’ exercise conditions were registered from
tributed. However, simulation studies have shown 411
362 native EPI space to the MNI152 template (2 mm3 that repeated measures ANOVA is relatively robust 412
co

363 resolution) by applying the previously computed EPI- against violations of normal distribution assump- 413
364 to-MNI transform matrices. After transformation, tions, provided it is the only assumption that has been 414
365 the denoised filtered functional data were inserted violated [31]. Therefore, PANAS scores for positive 415
366 to the group-level independent components analysis affect and negative affect were entered in two separate
Un

416
367 (gICA) to create data-driven multivariate compo- 2 (condition: ‘low’ / ’high’) × 2 (time: pre / post exer- 417
368 nents. To do so, we temporally concatenated the cise) repeated measures ANOVAs. Post hoc paired 418
369 time-points with the dimensional reduction approach t-tests with Bonferroni correction were performed.
370 in FSL melodic and constrained it to a GM mask to 419

371 avoid the noise from WM and CSF signals. Twenty-

372 five independent spatiotemporal components (ICs) Physiological data 420

373 were derived commonly across the study sample. To compare ‘low’ vs ‘high’ exercise intensity, 421

374 On the basis of the previous literature [5, 6] and exercise-related parameters were analyzed using a 422

375 to further add to the knowledge of the effects of paired t-test. For the exercise-related parameters 423
6 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

f
roo
rP
tho
d Au
cte
rre
co
Un

Fig. 1. Networks derived from group ICA decomposition. The colored network maps are presented at a threshold of Z > 2.3 based on
the group ICA. BGN = Basal Ganglia Network; ECN = Executive Control Network; CBN = Cerebellar Network; DAN = Dorsal Attention
Network; DMN = Default Mode Network; HIP = Hippocampal Network; LARN = Left Affective and Reward Network; LFPN = Left Fronto
Parietal Network; RARN = Right Affective and Reward Network; RFPN = Right Fronto Parietal Network; SMN = Sensorimotor Network;
R = right; L = left.
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 7

424 speed and HR, average values for the 30 minutes significant clusters of the ‘low’, ‘high’ and ‘low vs 475

425 of each exercise condition were used. Additionally, high’ conditions. 476

426 maximum values of blood lactate, HR and RPE were

427 taken. RESULTS 477

428 To analyze the HR monitored during MRI sessions,

429 means for every single session were calculated and Acute exercise manipulation 478

430 analyzed with a repeated-measures 2 × 2 ANOVA

431 (with factors condition: ‘low’ / ‘high’ and time: pre / During the ‘low’ exercise condition HR increased 479

f
432 post exercise). Due to artifacts we could only include in the mean from HRrest (resting heart rate) of 480

roo
433 18 subjects in the ANOVA that had a complete data 62.0 ± 6.3 to 119.2 ± 15.8 bpm, which corresponds 481

434 set for ‘low’ and ‘high’. Additionally, post hoc paired to 61.5 ± 8.4% of the subjects HRmax . In the ‘high’ 482

435 t-tests with Bonferroni correction were performed. exercise condition subjects HR increased from a 483

436 Significant results were reported at p-values of less HRrest of 65.5 ± 12.1 to 176.8 ± 6.9 bpm, which 484

437 than 0.05. corresponds to 91,1 ± 3.3% of the subjects HRmax . 485

rP
These HR correspond to light (57–63% HRmax ) and 486

438 fMRI-data vigorous (77–95% HRmax ) aerobic exercise inten- 487

439 To evaluate the modulation of rs-FC in the selected sity as defined by the American College of Sports 488

440 RSN’s in each condition from pre to post exer- Medicine (ACSM) [36]. Paired t-tests showed that the 489

HRmax during ‘low’ (128.2 ± 17.2 bpm) and ‘high’

tho
441 cise, we created within-subject difference maps (post 490

442 minus pre) for both exercise conditions. Then, we (186.5 ± 6.5 bpm) exercise intensity differ signifi- 491

443 applied a one-tailed one sample t-test on these indi- cantly (t(21) = –16.5; p < 0.001). The values of the 492

444 vidual difference maps for each network to measure mean perceived exertion of the ‘low’ exercise condi- 493

445 within condition effects. However, to evaluate the tion (9 ± 2) and the ‘high’ exercise condition (17 ± 1) 494
Au
446 rs-FC changes between conditions (‘high vs low’) also matched the intensity categorizations of the 495

447 in each RSN (N = 12) individual difference maps ACSM (light: 9–11; vigorous: 14–17) and differ sig- 496

448 were inserted into a paired t-test design. To apply nificantly from each other (t(21) = –21.1; p < 0.001). 497

449 for possible effects of HR changes induced by exer- Significant differences between the two exercise 498

450 cise, we transformed the mean HR-values during the intensities were also found for the mean speed 499

resting state scan via Fisher´s transformation into z- (‘low’: 7.1 ± 0.7 km/h; ‘high’: 12.8 ± 1.3 km/h;
d

451 500

452 values and inserted these as covariate of no interest in t(21) = –22.0; p < 0.001) and the maximum blood lac- 501

tate concentration (‘low’: 2.8 ± 1.9 mmol/L; ‘high’:

cte

453 both aforementioned designs. This analysis was only 502

454 performed for sub-groups as we needed to exclude 10.2 ± 3.2 mmol/L; t(21) = –10.6; p < 0.001). 503

455 some data with artifacts in the HR recording. Due

456 to missing HR data, the one sample t-test for ‘low’ Physiological monitoring 504

457 and ‘high’ included N = 20 subjects. As the two sub-

rre

458 jects excluded for the ‘low’ condition were not the Due to artifacts, N = 4 subjects needed to be 505

459 same two subjects which needed to be excluded in removed from the analysis between conditions, 506

460 the ‘high’ condition, the analysis for comparing the resulting in a sample size of N = 18. The 2 × 2 507

461 exercise conditions included only N = 18 subjects ANOVA showed a significant main effect of time 508
co

462 (results are presented in the supplementary material). (F(1,17) = 115.93, p < 0.001), a main effect of 509

463 Modulations were assessed by randomizing each condition (F(1,17) = 41.92, p < 0.001), and a signif- 510

464 component per subject using non-parametric 5,000 icant time × condition interaction (F(1,17) = 82.24, 511

465 random permutations, incorporating a threshold- p < 0.001). In the post hoc analysis N = 2 sub- 512
Un

466 free cluster enhancement technique (TFCE) [32]. jects needed to be excluded for the ‘low’ and also 513

467 The resulting clusters at p < 0.05 (surviving Z > 2.3 N = 2 for the ‘high’ exercise condition. Compar- 514

468 threshold after multiple comparison correction), were isons revealed that HR was significantly increased 515

469 labeled using the AAL WFU PickAtlas [33, 34] and after the ‘high’ exercise bout compared to pre exer- 516

470 the Harvard-Oxford Cortical and Subcortical Struc- cise (t(19) = –14.99, p < 0.001; pre ‘high’: 58.9 ± 9.0 517

471 tural Atlas provided by FSL’s software package [35]. bpm, post ‘high’: 78.0 ± 9.6 bpm). No significant 518

472 Z-statistic values from normalized differential FC difference in HR was found pre to post ‘low’ exer- 519

473 maps of pre to post scan in each condition were cise (t(19) = –0.88, p = 0.390; pre ‘low’: 57.7 ± 8.1 520

474 extracted per network using a mask created from the bpm, post ‘low’: 58.6 ± 8.7 bpm). Comparing the 521
 8 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

522 differences of pre-post ‘low’ vs pre-post ‘high’ between time and intensity of the exercise bout and 547

523 revealed a significant effect (t(17) = 9.07, p < 0.001), 2) the change from pre to post exercise bout for each 548

524 i.e. the increase from pre to post ‘high’ (19.1 ± 5.7 intensity separately. 549

525 bpm) was significantly greater than the increase from

526 pre to post ‘low’ (1.0 ± 5.5 bpm). No significant dif- Right affect and reward network 550
527 ference was found for HR between the two conditions
528 pre exercise (baseline) (t(19) = –0.65, p = 0.524). In the RARN a significant time × condition inter- 551

action in rs-FC in the right insula has been observed 552

f
529 PANAS (Fig. 2A, Table 2). This interaction was driven by an

roo
553

increase from pre to post ‘high’ exercise condition 554

530 Analyzing the positive affect scale using a 2 × 2 (trend only). 555

531 ANOVA revealed a significant main effect of time

532 (F(1,21) = 27.09, p < 0.001). Post-hoc comparisons
Right fronto parietal network 556
533 showed a significant increase for the ‘low’ exer-

rP
Results of the rs-FC analysis in the RFPN also indi-
cise condition pre (34.0 ± 8.5) to post (38.7 ± 6.6)
557
534
cated a significant time × condition interaction in the 558
535 (t(21) = –5.47, p < 0.001) and the ‘high’ exercise
right superior frontal gyrus (SFG) (Fig. 2B, Table 2).
condition pre (32.3 ± 10.6) to post (38.8 ± 6.9)
559
536
This effect was driven by a strong increase in rs-FC 560
537 (t(21) = –3.94, p = 0.001).
from pre to post ‘low’ exercise condition (Fig. 3E, 561

tho
538 The ANOVA for the negative affect scale only
Table 3). 562
539 revealed a significant main effect of condition
540 (F(1,21) = 6.29, p = 0.020). Comparing ‘low’ versus
541 ‘high’ post-hoc did not show any significant differ- Sensorimotor network 563

542 ences. The rs-FC analysis in the SMN revealed a signif- 564
Au
icant time × condition interaction in the postcentral 565

543 Acute exercise-induced changes in the functional gyrus and, the supplementary motor area (SMA) and 566

544 connectivity the precuneus (Fig. 2C, Table 2). Comparing pre 567

versus post ‘high’ exercise condition showed a sig- 568

545 In our data analysis we focused on two aspects nificant decrease in rs-FC in the postcentral gyrus 569

(Fig. 3F, Table 3).

d

546 for each RSN of interest: 1) the interaction effects 570

cte
rre
co
Un

Fig. 2. FC changes observed significantly different between conditions (‘low’ and ‘high’ intensity). Significant clusters in the interaction of
the: A) RARN in the right anterior insula; B) RFPN in the right superior frontal gyrus (SFG); C) SMN in the left and right postcentral gyrus,
the precuneus and the right supplementary motor area (SMA); Data shown is corrected for p < 0.05, after multiple comparison correction
using TFCE approach; z-values were normalized by dividing them by SD; For enhancing the contrast between the activation clusters and
the RSN masks, we chose either blue or red for the activation clusters. The colors do not indicate increases or decreases. N = 22; R = right;
L = left.
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 9

Table 2
Brain regions observed significantly different between conditions (‘low’ and ‘high’ intensity)
Brain region side k x y z p-value (tfce)
RARN – ‘high’ > ‘low’
anterior insula R 37 30 14 –16 0.013
RFPN – ‘low’>‘high’
superior frontal gyrus R 103 30 26 54 0.022
SMN - ‘low’>‘high’
precuneus L 184 –6 –42 56 0.015

f
supplementary motor area R 111 10 –18 50 0.038

roo
postcentral gyrus L 38 –30 –34 70 0.026
postcentral gyrus R 23 26 –32 66 0.040
postcentral gyrus L 11 –18 –40 76 0.046
postcentral gyrus L 10 –14 –34 64 0.046
Coordinates presented are in MNI space; k = cluster size; R = right; L = left.

rP
Table 3
Brain regions observed significantly different between pre to post within each exercise condition
Brain region side k x y z p-value (tfce)
ECN – ‘low’ post>pre
anterior cingulate cortex R 19 4 40 18 0.078∗∗

tho
DAN – ‘high’ pre>post
postcentral gyrus R 13 48 –22 44 0.045
LARN – ‘high’ post> pre
cerebellar lobule 6 L 7∗ –14 –70 –18 0.015
LFPN – ‘low’ post>pre
inferior parietal lobule L 15 –40 –50 44 0.033
RFPN – ‘low’ post>pre
Au
middle frontal gyrus R 568 44 36 34 0.004
middle orbitofrontal gyrus R 23 36 50 –8 0.034
inferior frontal gyrus (p. opercularis) R 10 46 12 38 0.046
superior frontal gyrus L 9∗ –2 30 52 0.041
SMN – ‘high’ pre > post
postcentral gyrus R 15 14 –38 78 0.027
d

Coordinates presented are in MNI space; k = cluster size; R = right; L = left; ∗ small cluster size, needs to be treated

with caution; ∗∗ represents a strong trend.

cte

571 Dorsal attention network Other networks 588

572 The DAN revealed no significant time × condition In the other examined RSNs we could not detect 589

573 interaction in rs-FC, although we detected a trend any further significant changes in FC. However, 590

(p = 0.053). However, there was a significant decrease we could detect statistical trends in some of the
rre

574 591

575 in FC from pre to post ‘high’ exercise condition in RSNs. The ECN showed a trend in the interaction 592

576 postcentral gyrus (Fig. 3B, Table 3). time × condition (p = 0.053) and an increase in rs-FC 593

from pre to post ‘low’ exercise condition in the ante- 594

Left affect and reward network rior cingulate cortex (ACC) (Fig. 3A). In the Salience 595
co

577

578 The LARN did not reveal a significant network we could also detect a strong trend in form of 596

579 time × condition interaction in rs-FC. However, an increase from pre to post ‘low’ exercise condition 597

580 there was a significant increase in FC from pre to (p = 0.06). 598

Un

581 post ‘high’ exercise condition in the left cerebellar

582 lobule VI (Fig. 3C, Table 3). Sub-group analysis including HR as covariate of 599

no interest 600

583 Left fronto parietal network

584 The LFPN did not reveal a significant Including the HR as covariate of no interest in our 601

585 time × condition interaction, but a significant analysis led to almost the same results, which we 602

586 increase in rs-FC from pre to post ‘low’ exercise found for the whole group analysis without HR as 603

587 condition in the left inferior parietal lobule (Fig. 3D, covariate. The RARN showed a significant interac- 604

Table 3). tion effect in the analysis without HR, but only a 605
 10 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

f
roo
rP
tho
d Au
cte
rre
co

Fig. 3. Differential effects in rs-FC from pre to post within exercise condition. Rs-FC changes in the: A) ECN within the ‘low’ condition
(post>pre) in the anterior cingulate cortex (ACC) (trend only, results are presented at p = 0.1), B) DAN within the ‘high’ condition (pre>post)
in the postcentral gyrus, C) LARN within the ‘high’ condition (post>pre) in the cerebellar lobule 6, D) LFPN within the ‘low’ condition
(post > pre) in the inferior parietal lobule (IPL), E) RFPN within the ‘low’ condition (post>pre) in the middle frontal gyrus (MFG), middle
Un

orbitofrontal gyrus (MOFG), inferior frontal gyrus (IFG) and superior frontal gyrus (SFG), F) SMN within the ‘high’ condition (pre>post)
in the postcentral gyrus; data shown is corrected for p < 0.05, after multiple comparison correction using TFCE approach; z-values were
normalized by dividing them by SD; For enhancing the contrast between the activation clusters and the RSN masks, we chose either blue or
red for the activation clusters. The colors do not indicate increases or decreases. N = 22; R = right; L = left.

606 very strong trend in the analysis with HR (p = 0.051) In the direct pre versus post comparisons the results 611

607 (see supplementary material: Figure S1, Table S1). for the ‘high’ exercise condition stayed the same, 612

608 The interaction effects in the RFPN and the SMN whereas the results of the pre versus post ‘low’ exer- 613

609 remained stable (see supplementary material: Figure cise condition changed slightly (see supplementary 614

610 S1, Table S1). material: Figure S2, Table S2). Where we initially 615
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 11

616 found a trend only in the ECN, results got signifi- effects on rs-FC suggest transient persistent network 663

617 cant in the ECN after inserting the HR as covariate. alterations after acute exercise bouts that are medi- 664

618 The significant change in the RFPN and the LFPN ated by different exercise intensities in a distinct 665

619 stayed the same when including HR. Detailed results manner and are discussed in the following with refer- 666

620 are summarized in the supplement. ence to cognitive/attentional (FPN), affective/reward 667

(RARN), and sensorimotor (SMN) function. 668

621 Additional analysis including nuisance

regression Modulation of the FPN after ‘low’-intensity 669

f
622
exercise 670

roo
623 Including nuisance regression (WM and CSF)
624 in our analysis pipeline led to similar results as The FPN plays a fundamental role in atten- 671

625 without nuisance regression. Our main findings tional monitoring [37], cognitive processing [38], and 672

626 could be replicated: the RARN showed a significant higher order motor control [39–41]. Our data revealed 673

interaction effect in the analysis without HR (see an increase in rs-FC after ‘low’-intensity exercise in 674

rP
627

628 supplementary material: Figure S3, Table S3), but the left and right FPN, which is a large-scale net- 675

629 no significances with HR. The interaction effects in work representing a major functional hub in terms 676

630 the RFPN and the SMN remained stable with and of distributed connectivity [42–44]. Meta-analyses 677

631 without HR as well as with and without nuisance also support the notion that rs-FC strength across the 678

tho
632 regression(see supplementary material: Figure S3 + wide-range FPN network is crucial for cognitive and 679

633 S4, Table S3 + S4). emotional self-regulation [45, 46]. Top-down atten- 680

634 In the direct pre versus post comparisons we tional control processes are regulated by the superior 681

635 could not detect significant changes in the DAN and frontal gyrus [47, 48] and the inferior parietal lob- 682

636 ECN anymore, however we could detect significant ule [49]; for instance, Yantis et al. suggested that the 683
Au
637 chances in the DMN, which we haven´t in the anal- inferior parietal lobule is a source of brief attention- 684

638 ysis without nuisance regression (see supplementary related control signals that indicate top-down shifts 685

639 material: Figure S5 + S6, Table S5 + S6). Detailed in attentive states [50]. 686

640 results are summarized in the supplement. Unfortunately, we have no neuropsychological test 687

data on attention or executive control tasks (e.g. 688

Stroop, Flanker, Simon tasks, etc.) available. It is thus

d

689

641 DISCUSSION not possible to directly infer on the impact of exercise 690
cte

on FPN-associated behavioral effects. Nevertheless, 691

642 This is the first study to investigate the differen- it is highly interesting that the connectivity strength 692

643 tial effects of aerobic exercise bouts of ‘low’ (35% within the FPN was increased by physical exercise 693

644 below lactate threshold) and ‘high’ (20% above lac- of ‘low’ intensity: The superior frontal gyrus and the 694

645 tate threshold) intensity on RSN activations in young inferior parietal lobule showed a significant increase 695
rre

646 and healthy recreational athletes. Behavioral data in rs-FC after ‘low’-intensity exercise, while a signif- 696

647 revealed a significant increase in positive affect after icant decrease in rs-FC in the superior frontal gyrus 697

648 both exercise intensities. Moreover, while provid- was found after ‘high’-intensity exercise. Our find- 698

649 ing further evidence for persisting rs-FC increases ings on elevated FC in networks mediating attentional 699
co

650 in cognitive, sensorimotor, and affective networks control processes specifically after ‘low’-intensity 700

651 after acute exercise bouts [5, 6], the results of this exercise bouts are well in line with a number of 701

652 study suggest additional exercise intensity-dependent studies in the literature that have reported improved 702

653 modulatory effects on distinct resting state brain net- attention and cognitive performance after low to mod- 703
Un

654 works. Interaction effects depending on the factors erate exercise bouts [51–53]. According to several 704

655 pre / post and ‘high’ / ‘low’ were found in the RARN, studies and meta-analysis the relationship between 705

656 the SMN and the RFPN. Whereas ‘low’-intensity exercise intensity and cognitive performance is U- 706

657 exercise bouts resulted in increased rs-FC in brain shaped [54–57], i.e. suggesting that the beneficial 707

658 networks implicated in cognitive and attentional pro- effects occur preferentially with light to moderate 708

659 cessing (LFPN and RFPN), ‘high’-intensity exercise exercise intensities, while they then deteriorate with 709

660 increased rs-FC in affective brain networks (LARN) high-intensity anaerobic exercise [58]. Also, effects 710

661 and decreased FC in the sensorimotor (SMN) net- of different exercise intensities on cerebral oxygena- 711

662 work. In general terms, the observed locoregional tion tested with near infrared spectroscopy (NIRS) 712
 12 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

713 point in this direction, i.e. indicating a quadratic occipital areas [62]. This network promotes skeleton- 763

714 relation of prefrontal cortex oxygenation with incre- motor body orientation, environmental monitoring, 764

715 mental exercise: first rising between moderate and and response selection [67–71]. Reiman et al. had 765

716 hard intensities, and then falling at very hard intensi- posited that the anterior insula may function as an 766

717 ties [59]. alarm center for internally-sensed dangers or homeo- 767

718 In conclusion, our data seem to agree with the static changes [72]. Indeed, this hypothesis is also 768

719 general notion, that beneficial effects of exercise are supported by meta-analytical data, reporting that 769

720 preferentially found in low- to intermediate exercise the anterior insula is more involved in cognitively 770

f
721 ranges (according to physiological exercise recom- demanding emotional tasks, and suggesting prefer- 771

roo
722 mendations). Whether the quadratic relation between ential recruitment for evaluative, experimental, or 772

723 exercise intensity and functional outcomes described expressive aspects of internally generated emotions 773

724 above can conclusively explain why enhanced FPN [73]. Gasquoine reported that the insula is the pri- 774

725 rs-FC changes in our study were found exclusively for mary reception region of the brain for interoceptive 775

726 the ‘low’-intensity condition remains uncertain: First information, in particular for deciphering changes in 776

rP
727 of all, one needs to consider that several studies have autonomic nervous system activity levels [74], and 777

728 reported improvements in cognitive states also in the Gu et al. showed that the anterior insula responds 778

729 context of high intensity anaerobic exercise bouts [54, to the interaction of cognitive demand and stimulus 779

730 60, 61]. Moreover, our rs-fMRI measurements were valence [75]. Based on the role of the anterior insula in 780

tho
731 undertaken on average 18 minutes after the acute encoding interoceptive signals from the body’s inter- 781

732 exercise bout, i.e. when any deteriorating impact of nal milieu that reflect autonomic activity, Damasio 782

733 ‘high’-intensity exercise on cognitive states should argued that the insula is the site within the brain where 783

734 have already leveled off (see introduction). Thus, subjective feelings of emotion are generated [76]. 784

735 future studies will need to address the relationship Our findings might reflect elevated processing of 785
Au
736 between rs-FC in dedicated cognitive / attentional interoceptive signals and internally generated emo- 786

737 networks and exercise intensity and/or delay time, tions, and suggest a possible neural mechanism 787

738 before conclusive interpretations can be drawn. underlying the subjective affective experiences after 788

strenuous exercise [77, 78]. Why these findings in the 789

739 Modulation of the ARN after ‘high’-intensity RARN are reduced when including HR as covariate 790

of no interest remains unclear. In fact, it could either

d

740 exercise 791

be due to the HR difference from pre to post exer- 792

cte

741 The significant interaction effect in the ARN was cise, or it could be related to the reduced power after 793

742 driven by an almost significant increase of rs-FC the exclusion of 4 subjects from this analysis due to 794

743 after the ‘high’-intensity exercise condition and a artifacts. 795

744 decrease after the ‘low’ exercise condition in the The assessments of mood changes using the 796

745 right anterior insula. Weng et al. reported a signifi- PANAS revealed a significant increase after both 797
rre

746 cant rs-FC increase in the right anterior insula after exercise interventions (‘low’ and ‘high’) in the pos- 798

747 moderate aerobic exercise, as compared to a passive itive affect scale. These findings are in line with 799

748 control condition, and a previous study investigat- previous literature reporting PANAS ratings after 800

749 ing the connectivity of the insula at rest showed it to a low and a high exercise intervention in students 801
co

750 be functionally integrated in two distinct neural net- [79]. Pre-exercise, the values of the negative affect 802

751 works [5]: 1) The ventral most anterior insula is part variable were already near the bottom of the range, 803

752 of the anterior network, which is connected to the thus representing a floor effect and leaving little/no 804

753 rostral anterior cingulate cortex, the middle and infe- room for further decreases in negative affect. This 805
Un

754 rior frontal cortex, and the temporo parietal cortex might explain why we could not detect any signifi- 806

755 [62]. This anterior network is involved in emotional cant changes in negative affect. The results of these 807

756 salience detection and attentional control [63–66] and behavioral data are not completely in line with the 808

757 is mostly related to the integration of multiple cog- imaging findings in the ARN, as we found a signif- 809

758 nitive, homeostatic and emotional (i.e. interoceptive) icant increase in positive affect after both exercise 810

759 functions. 2) The dorsal posterior insula is part of intensities in the mood questionnaire, but differen- 811

760 the posterior network and is connected to the dorsal tial effects induced by exercise intensity in the ARN. 812

761 posterior cingulate cortex, along with sensorimo- One possible reason might be the timing of the acqui- 813

762 tor, premotor, supplementary motor, temporal, and sition of the data. While the PANAS positive scale 814
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 13

815 was acquired directly post exercise, the rs-fMRI was bly relate to the persistent changes in brain function 863

816 acquired approximately 18 minutes after exercise. after acute exercise bouts [85]. Best described are the 864

mechanisms underlying mood changes. Potentially, 865

817 Modulation of the SMN after ‘high’-intensity these affective responses are mediated by levels of 866

818 exercise endogenous opioids, which are found to be elevated 867

after acute bouts of exercise in both humans and ani- 868

819 Interestingly, ‘high’-intensity exercise was also mals in the peripheral blood; this effect seems to 869

820 associated with decreased intrinsic FC in the SMN. be intensity-dependent and has been linked to mood 870

f
821 The SMN includes the sensorimotor cortex and improvements [86–90]. Independent PET ligand acti- 871

roo
822 extends to the SMA. Studies have indicated that vation studies from different groups were able to 872

823 the SMN is activated during motor tasks like finger report local exercise-induced central opioid release 873

824 tapping [80]. Previous work investigating the effect and related opioid tracer binding changes in the ante- 874

825 of moderate exercise compared to rest could detect rior insula [91–93] after strenuous exercise, which fits 875

826 increased co-activation in the pre and/or postcentral well with brain areas found in the ARN in our study. 876

rP
827 gyri [6]. Together with our results, different exercise We also found FC changes in the FPN, which 877

828 intensities seem to alter SMN brain activity in a dif- is involved in cognitive processes. Acute exercise 878

829 ferential manner, namely increasing rs-FC after low stimulates the hypothalamic-pituitary-adrenal (HPA) 879

830 and moderate exercise, but decreasing rs-FC after axis in an intensity-dependent fashion, resulting in 880

tho
831 high intensity exercise. This may be related to per- increased cortisol levels [94–96]. Some studies have 881

832 sistent fatigue after strong exercise intensities, i.e. shown that moderate increases in cortisol enhance 882

833 either long & enduring or short & of high intensity as working memory [97, 98], whereas the elevation of 883

834 in the current study. Indeed, acute exhaustive exer- cortisol over a certain level serves to impair memory 884

835 cise has been shown to cause reduced excitability [99, 100]. Prefrontal cortex is characterized by a high 885
Au
836 of the SMA [81], which could provide an expla- cortisol receptor density [101]. 886

837 nation for the decrease in functional connectivity Another possible mechanism would be the level of 887

838 within the SMN after ‘high’-intensity exercise. In the the brain derived neurotrophic factor (BDNF), which 888

839 same line, reduced connectivity between the SMA is one of the main regulators of neurogenesis. Studies 889

840 and the sensorimotor cortex was found in patients investigating the effect of BDNF after acute exercise 890

with chronic fatigue syndrome during motor prepa- bouts on cognitive function reported mixed effects.
d

841 891

842 ration [82]. However, rs-FC analyses in multiple Some reported an impairment, some an enhancement 892
cte

843 sclerosis patients with fatigue also reported opposite and other reported no effect on cognition [102–104]. 893

844 effects [83]. Therefore, it is interesting to consider Protocols that were too exhaustive rather led to an 894

845 experimental data on ‘fatigue caused by sustaining impairment in cognitive function [105]. Although, 895

846 submaximal-intensity muscle contraction(s)’ during there are more neurotransmitters that are involved in 896

847 scanning: a significant increase in FC was reported cognitive processes (e.g. norepinephrine and acetyl- 897
rre

848 in M1, S1, and SMA [84]. The authors interpreted choline) [106, 107], little has been done to investigate 898

849 their finding of strengthened FC with increasing the influence of these transmitters after acute exercise 899

850 muscle fatigue as a compensatory phenomenon ‘by on cognitive function. 900

851 enhancing the descending command for greater mus-

co

852 cle recruitment to maintain the same force’ [84]. As Limitations 901

853 our data were acquired 18 minutes after the strenuous

854 exercise, such compensatory effects are likely to be There are some limitations that must be considered 902

855 no longer in place, leaving the sensorimotor output in the current study. The study population consisted 903
Un

856 system in a persisting post-exhaustive state. only of males in order to exclude potential hormone 904

cycle-linked humoral factors. To generalize findings, 905

857 Possible underlying mechanisms future studies need to investigate gender-balanced 906

cohorts. Additionally, no randomized control con- 907

858 A variety of possible underlying mechanisms, dition was implemented, so it cannot be excluded 908

859 coming from both human and animal studies, are to with certainty that effects shown in the pre-post 909

860 be considered to explain our findings, notably neu- comparisons are also driven by factors other than 910

861 rochemical changes (lactate, cortisol, neurotrophins, exercise. We consider this to be unlikely, as the mod- 911

862 neurotransmitters and neuromodulators) that possi- ulations of rs-FC maps induced by exercise were 912
 14 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

913 generally in line with previous initial findings in the REFERENCES 957

914 literature (as mentioned above), with the additional

915 description of exercise intensity effects. Future stud- [1] Chirles TJ, Reiter K, Weiss LR, Alfini AJ, Nielson KA, 958
Smith JC. Exercise training and functional connectivity 959
916 ies will also need to combine rs-FC measurements,
changes in mild cognitive impairment and healthy elders. 960
917 exercise conditions of different intensities, and ded- J Alzheimers Dis. 2017;57(3):845-56. 961
918 icated neuropsychological tests to shed further light [2] Flodin P, Jonasson LS, Riklund K, Nyberg L, Boraxbekk 962

919 on the prolonged neurobiological mechanisms at the CJ. Does aerobic exercise influence intrinsic Bbrain activ- 963
ity? An aerobic exercise intervention among healthy old 964
920 brain network level mediated by acute exercise bouts. adults. Front Aging Neurosci. 2017;9:267.

f
965
921 Finally, direct comparisons with other studies are [3] McGregor KM, Crosson B, Krishnamurthy LC, Krish- 966

roo
922 hampered by the fact, that standardization is lacking namurthy V, Hortman K, Gopinath K, et al. Effects of 967

923 throughout the exercise and imaging literature with a 12-week aerobic spin intervention on resting state net- 968
works in previously sedentary older adults. Front Psychol. 969
924 regards to age, fitness level, exercise intensity, exer- 2018;9:2376. 970
925 cise duration, and the time point of brain imaging [4] Tozzi L, Carballedo A, Lavelle G, Doolin K, Doyle 971

926 after the exercise conditions. M, Amico F, et al. Longitudinal functional connectivity 972

rP
changes correlate with mood improvement after regular 973
exercise in a dose-dependent fashion. Eur J Neurosci. 974
2016;43(8):1089-96. 975
[5] Weng TB, Pierce GL, Darling WG, Falk D, Magnotta VA, 976
927 CONCLUSIONS Voss MW. The acute effects of aerobic exercise on the 977
functional connectivity of human brain networks. Brain 978

tho
928 This is the first study demonstrating differential Plast. 2017;2(2):171-90. 979

929 effects of two different individually titrated exercise [6] Rajab AS, Crane DE, Middleton LE, Robertson AD, 980
Hampson M, MacIntosh BJ. A single session of exercise 981
930 intensities on resting state networks. The ‘low’ and increases connectivity in sensorimotor-related brain net- 982
931 ‘high’ exercise intensities affect prolonged network works: A resting-state fMRI study in young healthy adults. 983

alterations in attentional / cognitive and affective Front Hum Neurosci. 2014;8:625. 984
932
Au
[7] Chang YK, Labban JD, Gapin JI, Etnier JL. The effects of 985
933 / reward networks in a differential manner. Future
acute exercise on cognitive performance: A meta-analysis. 986
934 research in this area will help to provide neurobiolog- Brain Res. 2012;1453:87-101. 987
935 ical evidence which exercise intensity is best suited [8] Chang YK, Chu IH, Chen FT, Wang CC. Dose-response 988

936 for either cognitive / attentional or affective modu- effect of acute resistance exercise on Tower of Lon- 989
don in middle-aged adults. J Sport Exerc Psychol. 990
937 lation and may pave the way for supportive clinical
d

2011;33(6):866-83. 991
938 applications in patients or for enhancing brain plas- [9] Chang YK, Etnier JL. Exploring the dose-response 992
939 ticity. relationship between resistance exercise intensity and cog- 993
cte

940 The authors have no conflict of interest to report. nitive function. J Sport Exerc Psychol. 2009;31(5):640-56. 994
[10] Robertson AD, Crane DE, Rajab AS, Swardfager W, Mar- 995
941 This research did not receive any specific grant zolini S, Shirzadi Z, et al. Exercise intensity modulates the 996
942 from funding agencies in the public, commercial, or change in cerebral blood flow following aerobic exercise 997

943 not-for-profit sectors. in chronic stroke. Exp Brain Res. 2015;233(8):2467- 998
75. 999
rre

[11] Williamson JW, McColl R, Mathews D. Changes in 1000

regional cerebral blood flow distribution during postex- 1001
ercise hypotension in humans. J Appl Physiol (1985). 1002
944 SUPPLEMENTARY MATERIAL 2004;96(2):719-24. 1003
[12] Schmitt A, Martin JA, Rojas S, Vafa R, Scheef L, Struder 1004
co

945 Analogue to the described results in the main HK, et al. Effects of low and high intensity exercise on 1005

946 manuscript, the supplement includes the results of emotional face-processing: A fMRI face-matching study. 1006
Soc Cogn Affect Neurosci. 2019. 1007
947 the rs-FC analysis including HR as covariate of no [13] Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs 1008
948 interest. Additionally, we present the results from an J, Weiller E, et al. The Mini-International Neuropsychi- 1009
Un

949 analysis using nuisance regression (WM and CSF). atric Interview (M.I.N.I.): The development and validation 1010

950 The supplement describes the interaction effects and of a structured diagnostic psychiatric interview for 1011
DSM-IV and ICD-10. J Clin Psychiatry. 1998;59(Suppl 1012
951 the pre/post comparison within each exercise condi- 20):22-33;quiz 4-57. 1013
952 tion for the RSN that revealed significant changes in [14] Spielberger CD, Gorsuch RL, Lushene R, Vagg PR, Jacobs 1014

953 FC. GA. Manual for the State-Trait Anxiety Inventory: Palo 1015
Alto, CA: Consulting Psychologists Press, 1983. 1016
954 The supplementary material is available in the
[15] Hautzinger M, Bailer M, Worall H, Keller F. Beck- 1017
955 electronic version of this article: http://dx.doi.org/ Depressions-Inventar: (BDI); Testhandbuch. [Beck 1018
956 10.3233/BPL-190081. Depression Inventory (BDI): Test manual]: Bern; 1994. 1019
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 15

1020 [16] Oldfield RC. The assessment and analysis of hand- labeling of activations in SPM using a macroscopic 1085
1021 edness: The Edinburgh inventory. Neuropsychologia. anatomical parcellation of the MNI MRI single-subject 1086
1022 1971;9(1):97-113. brain. Neuroimage. 2002;15(1):273-89. 1087
1023 [17] Schmidt KH, Metzler P. Wortschatztest: WST: Weinheim: [35] Desikan RS, Segonne F, Fischl B, Quinn BT, Dick- 1088
1024 Beltz Test GmbH; 1992. erson BC, Blacker D, et al. An automated labeling 1089
1025 [18] Borg GA. Psychophysical bases of perceived exertion. system for subdividing the human cerebral cortex on MRI 1090
1026 Med Sci Sports Exerc. 1982;14(5):377-81. scans into gyral based regions of interest. Neuroimage. 1091
1027 [19] Hollmann W, Knigge H, Knicker A, Strüder HK. Meth- 2006;31(3):968-80. 1092
1028 ods for Measurement of Physical Fitness and Training [36] American College of Sports Medicine, Riebe D, Ehrman 1093
1029 Recommendations in Studies on Humans. Functional Neu- JK, Liguori G, Magal M. ACSM’s guidelines for exercise 1094

f
1030 roimaging in Exercise and Sport Sciences. New York, NY: testing and prescription. Tenth edition. ed. Philadelphia: 1095

roo
1031 Springer New York; 2012, pp. 79-107. Wolters Kluwer; 2018. xxx, p. 472. 1096
1032 [20] Bishop D, Jenkins DG, Mackinnon LT. The relation- [37] Ptak R. The frontoparietal attention network of the human 1097
1033 ship between plasma lactate parameters, Wpeak and 1-h brain: Action, saliency, and a priority map of the environ- 1098
1034 cycling performance in women. Med Sci Sports Exerc. ment. Neuroscientist. 2012;18(5):502-15. 1099
1035 1998;30(8):1270-5. [38] Marek S, Dosenbach NUF. The frontoparietal network: 1100
1036 [21] Williams EJ. Experimental designs balanced for the esti- Function, electrophysiology, and importance of indi- 1101

rP
1037 mation of residual effects of treatments. Aust J Sci Res Ser vidual precision mapping. Dialogues Clin Neurosci. 1102
1038 A. 1949;2(2):149-68. 2018;20(2):133-40. 1103
1039 [22] Krohne HW, Egloff B, Kohlmann CW, Tausch A. Unter- [39] Wise SP, Boussaoud D, Johnson PB, Caminiti R. Pre- 1104
1040 suchungen mit einer deutschen Version der "Positive motor and parietal cortex: Corticocortical connectivity 1105
1041 and Negative Affect Schedule“ (PANAS). Diagnostica. and combinatorial computations. Annu Rev Neurosci. 1106
1042 1996;42:139-56. 1997;20:25-42. 1107

tho
1043 [23] Ashburner J, Ridgway GR. Symmetric diffeomorphic [40] Wu T, Wang L, Chen Y, Zhao C, Li K, Chan P. Changes of 1108
1044 modeling of longitudinal structural MRI. Front Neurosci. functional connectivity of the motor network in the resting 1109
1045 2012;6:197. state in Parkinson’s disease. Neurosci Lett. 2009;460(1):6- 1110
1046 [24] Smith SM. Fast robust automated brain extraction. Hum 10. 1111
1047 Brain Mapp. 2002;17(3):143-55. [41] Wymbs NF, Bassett DS, Mucha PJ, Porter MA, Grafton 1112
1048 [25] Wu CW, Chen CL, Liu PY, Chao YP, Biswal BB, Lin CP. ST. Differential recruitment of the sensorimotor puta- 1113
Au
1049 Empirical evaluations of slice-timing, smoothing, and nor- men and frontoparietal cortex during motor chunking in 1114
1050 malization effects in seed-based, resting-state functional humans. Neuron. 2012;74(5):936-46. 1115
1051 magnetic resonance imaging analyses. Brain Connect. [42] Marek S, Hwang K, Foran W, Hallquist MN, Luna B. 1116
1052 2011;1(5):401-10. The contribution of network organization and integra- 1117
1053 [26] Pruim RHR, Mennes M, van Rooij D, Llera A, Buitelaar tion to the development of cognitive control. PLoS Biol. 1118
1054 JK, Beckmann CF. ICA-AROMA: A robust ICA-based 2015;13(12):e1002328. 1119
1055 strategy for removing motion artifacts from fMRI data. [43] Power JD, Cohen AL, Nelson SM, Wig GS, Barnes KA, 1120
d

1056 Neuroimage. 2015;112:267-77. Church JA, et al. Functional network organization of the 1121
1057 [27] Bright MG, Tench CR, Murphy K. Potential pitfalls when human brain. Neuron. 2011;72(4):665-78. 1122
cte

1058 denoising resting state fMRI data using nuisance regres- [44] Power JD, Schlaggar BL, Lessov-Schlaggar CN, Petersen 1123
1059 sion. Neuroimage. 2017;154:159-68. SE. Evidence for hubs in human functional brain networks. 1124
1060 [28] Smith SM, Fox PT, Miller KL, Glahn DC, Fox PM, Neuron. 2013;79(4):798-813. 1125
1061 Mackay CE, et al. Correspondence of the brain’s func- [45] Chen T, Becker B, Camilleri J, Wang L, Yu S, Eickhoff 1126
1062 tional architecture during activation and rest. Proc Natl SB, et al. A domain-general brain network underlying 1127
1063 Acad Sci U S A. 2009;106(31):13040-5. emotional and cognitive interference processing: Evi- 1128
rre

1064 [29] Beckmann CF, Mackay CE, Filippini N, Smith SM. Group dence from coordinate-based and functional connectivity 1129
1065 comparison of resting-state FMRI data using multi-subject meta-analyses. Brain Struct Funct. 2018;223(8):3813- 1130
1066 ICA and dual regression. NeuroImage. 2009;47:S148. 40. 1131
1067 [30] Filippini N, MacIntosh BJ, Hough MG, Goodwin GM, [46] Langner R, Leiberg S, Hoffstaedter F, Eickhoff SB. 1132
1068 Frisoni GB, Smith SM, et al. Distinct patterns of brain Towards a human self-regulation system: Common and 1133
co

1069 activity in young carriers of the APOE-epsilon4 allele. distinct neural signatures of emotional and behavioural 1134
1070 Proc Natl Acad Sci U S A. 2009;106(17):7209-14. control. Neurosci Biobehav Rev. 2018;90:400-10. 1135
1071 [31] Berkovits I, Hancock GR, Nevitt J. Bootstrap resampling [47] Casey BJ, Thomas KM, Welsh TF, Badgaiyan RD, Eccard 1136
1072 approaches for repeated measure designs: Relative robust- CH, Jennings JR, et al. Dissociation of response con- 1137
1073 ness to sphericity and normality violations. Educ Psychol flict, attentional selection, and expectancy with functional 1138
Un

1074 Meas. 2000;60(6):877-92. magnetic resonance imaging. Proc Natl Acad Sci U S A. 1139
1075 [32] Smith SM, Nichols TE. Threshold-free cluster enhance- 2000;97(15):8728-33. 1140
1076 ment: Addressing problems of smoothing, threshold [48] Colcombe SJ, Kramer AF, Erickson KI, Scalf P, McAuley 1141
1077 dependence and localisation in cluster inference. Neu- E, Cohen NJ, et al. Cardiovascular fitness, cortical 1142
1078 roimage. 2009;44(1):83-98. plasticity, and aging. Proc Natl Acad Sci U S A. 1143
1079 [33] Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An auto- 2004;101(9):3316-21. 1144
1080 mated method for neuroanatomic and cytoarchitectonic [49] Hopfinger JB, Buonocore MH, Mangun GR. The neural 1145
1081 atlas-based interrogation of fMRI data sets. Neuroimage. mechanisms of top-down attentional control. Nat Neu- 1146
1082 2003;19(3):1233-9. rosci. 2000;3(3):284-91. 1147
1083 [34] Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Criv- [50] Yantis S, Schwarzbach J, Serences JT, Carlson RL, Stein- 1148
1084 ello F, Etard O, Delcroix N, et al. Automated anatomical metz MA, Pekar JJ, et al. Transient neural activity in 1149
 16 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI

1150 human parietal cortex during spatial attention shifts. Nat [67] Craig AD. How do you feel? Interoception: The sense of 1215
1151 Neurosci. 2002;5(10):995-1002. the physiological condition of the body. Nat Rev Neurosci. 1216
1152 [51] Coleman M, Offen K, Markant J. Exercise similarly facil- 2002;3(8):655-66. 1217
1153 itates men and women’s selective attention task response [68] Flynn FG. Anatomy of the insula functional and clinical 1218
1154 times but differentially affects memory task performance. correlates. Aphasiology. 1999;13(1):55-78. 1219
1155 Front Psychol. 2018;9:1405. [69] Kurth F, Eickhoff SB, Schleicher A, Hoemke L, Zilles 1220
1156 [52] Chang YK, Alderman BL, Chu CH, Wang CC, Song TF, K, Amunts K. Cytoarchitecture and probabilistic maps 1221
1157 Chen FT. Acute exercise has a general facilitative effect on of the human posterior insular cortex. Cereb Cortex. 1222
1158 cognitive function: A combined ERP temporal dynamics 2010;20(6):1448-61. 1223
1159 and BDNF study. Psychophysiology. 2017;54(2):289-300. [70] Kurth F, Zilles K, Fox PT, Laird AR, Eickhoff SB. A link 1224

f
1160 [53] Loprinzi PD, Kane CJ. Exercise and cognitive function: A between the systems: Functional differentiation and inte- 1225

roo
1161 randomized controlled trial examining acute exercise and gration within the human insula revealed by meta-analysis. 1226
1162 free-living physical activity and sedentary effects. Mayo Brain Struct Funct. 2010;214(5-6):519-34. 1227
1163 Clin Proc. 2015;90(4):450-60. [71] Taylor KS, Seminowicz DA, Davis KD. Two systems of 1228
1164 [54] Alves CR, Tessaro VH, Teixeira LA, Murakava K, resting state connectivity between the insula and cingulate 1229
1165 Roschel H, Gualano B, et al. Influence of acute high- cortex. Hum Brain Mapp. 2009;30(9):2731-45. 1230
1166 intensity aerobic interval exercise bout on selective [72] Reiman EM, Lane RD, Ahern GL, Schwartz GE, David- 1231

rP
1167 attention and short-term memory tasks. Percept Mot Skills. son RJ, Friston KJ, et al. Neuroanatomical correlates of 1232
1168 2014;118(1):63-72. externally and internally generated human emotion. Am J 1233
1169 [55] Chmura J, Nazar K, Kaciuba-Uscilko H. Choice reaction Psychiatry. 1997;154(7):918-25. 1234
1170 time during graded exercise in relation to blood lactate [73] Phan KL, Wager T, Taylor SF, Liberzon I. Functional 1235
1171 and plasma catecholamine thresholds. Int J Sports Med. neuroanatomy of emotion: A meta-analysis of emo- 1236
1172 1994;15(4):172-6. tion activation studies in PET and fMRI. Neuroimage. 1237

tho
1173 [56] McMorris T, Graydon J. The effect of exercise on 2002;16(2):331-48. 1238
1174 the decision-making performance of experienced and [74] Gasquoine PG. Contributions of the insula to cognition 1239
1175 inexperienced soccer players. Res Q Exerc Sport. and emotion. Neuropsychol Rev. 2014;24(2):77-87. 1240
1176 1996;67(1):109-14. [75] Gu X, Liu X, Van Dam NT, Hof PR, Fan J. Cognition- 1241
1177 [57] McMorris T, Sproule J, Turner A, Hale BJ. Acute, interme- emotion integration in the anterior insular cortex. Cereb 1242
1178 diate intensity exercise, and speed and accuracy in working Cortex. 2013;23(1):20-7. 1243
Au
1179 memory tasks: A meta-analytical comparison of effects. [76] Damasio A. Feelings of emotion and the self. Ann N Y 1244
1180 Physiol Behav. 2011;102(3-4):421-8. Acad Sci. 2003;1001:253-61. 1245
1181 [58] Brisswalter J, Collardeau M, Rene A. Effects of acute [77] Niven A, Thow J, Holroyd J, Turner AP, Phillips SM. Com- 1246
1182 physical exercise characteristics on cognitive perfor- parison of affective responses during and after low volume 1247
1183 mance. Sports Med. 2002;32(9):555-66. high-intensity interval exercise, continuous moderate- and 1248
1184 [59] Rooks CR, Thom NJ, McCully KK, Dishman RK. Effects continuous high-intensity exercise in active, untrained, 1249
1185 of incremental exercise on cerebral oxygenation measured healthy males. J Sports Sci. 2018;36(17):1993-2001. 1250
d

1186 by near-infrared spectroscopy: A systematic review. Prog [78] Malik AA, Williams CA, Weston KL, Barker AR. 1251
1187 Neurobiol. 2010;92(2):134-50. Perceptual responses to high- and moderate-intensity 1252
cte

1188 [60] Du Rietz E, Barker AR, Michelini G, Rommel AS, interval exercise in adolescents. Med Sci Sports Exerc. 1253
1189 Vainieri I, Asherson P, et al. Beneficial effects of acute 2018;50(5):1021-30. 1254
1190 high-intensity exercise on electrophysiological indices of [79] Bixby WR, Spalding TW, Hatfield BA. Temporal dynam- 1255
1191 attention processes in young adult men. Behav Brain Res. ics and dimensional specificity of the affective response 1256
1192 2019;359:474-84. to exercise of varying intensity: Differing pathways to a 1257
1193 [61] Gmiat A, Micielska K, Kozlowska M, Flis DJ, Smaruj M, common outcome. J Sport Exercise Psy. 2001;23(3):171- 1258
rre

1194 Kujach S, et al. The impact of a single bout of high intensity 90. 1259
1195 circuit training on myokines’ concentrations and cogni- [80] Biswal B, Yetkin FZ, Haughton VM, Hyde JS. Func- 1260
1196 tive functions in women of different age. Physiol Behav. tional connectivity in the motor cortex of resting 1261
1197 2017;179:290-7. human brain using echo-planar MRI. Magn Reson Med. 1262
1198 [62] Cauda F, D’Agata F, Sacco K, Duca S, Geminiani G, Ver- 1995;34(4):537-41. 1263
co

1199 celli A. Functional connectivity of the insula in the resting [81] Coco M, Perciavalle V, Cavallari P, Perciavalle V. Effects 1264
1200 brain. Neuroimage. 2011;55(1):8-23. of an exhaustive exercise on motor skill learning and on the 1265
1201 [63] Corbetta M, Shulman GL. Control of goal-directed and excitability of primary motor cortex and supplementary 1266
1202 stimulus-driven attention in the brain. Nat Rev Neurosci. motor area. Medicine (Baltimore). 2016;95(11):e2978. 1267
1203 2002;3(3):201-15. [82] van der Schaaf ME, Roelofs K, de Lange FP, Geurts DEM, 1268
Un

1204 [64] Dosenbach NU, Visscher KM, Palmer ED, Miezin FM, van der Meer JWM, Knoop H, et al. Fatigue is associated 1269
1205 Wenger KK, Kang HC, et al. A core system for the imple- with altered monitoring and preparation of physical effort 1270
1206 mentation of task sets. Neuron. 2006;50(5):799-812. in patients with chronic fatigue syndrome. Biol Psychiatry 1271
1207 [65] Fox MD, Corbetta M, Snyder AZ, Vincent JL, Raichle Cogn Neurosci Neuroimaging. 2018;3(4):392-404. 1272
1208 ME. Spontaneous neuronal activity distinguishes human [83] Bisecco A, Nardo FD, Docimo R, Caiazzo G, d’Ambrosio 1273
1209 dorsal and ventral attention systems. Proc Natl Acad Sci A, Bonavita S, et al. Fatigue in multiple sclerosis: The 1274
1210 U S A. 2006;103(26):10046-51. contribution of resting-state functional connectivity reor- 1275
1211 [66] Seeley WW, Menon V, Schatzberg AF, Keller J, Glover ganization. Mult Scler. 2018;24(13):1696-705. 1276
1212 GH, Kenna H, et al. Dissociable intrinsic connectivity [84] Jiang Z, Wang XF, Kisiel-Sajewicz K, Yan JH, Yue GH. 1277
1213 networks for salience processing and executive control. Strengthened functional connectivity in the brain during 1278
1214 J Neurosci. 2007;27(9):2349-56. muscle fatigue. Neuroimage. 2012;60(1):728-37. 1279
 A. Schmitt et al. / Acute Effects of Exercise on Resting State fMRI 17

1280 [85] Basso JC, Suzuki WA. The effects of acute exercise on responses to the stress of treadmill exercise. Physio- 1321
1281 mood, cognition, neurophysiology, and neurochemical logic adaptations to physical training. N Engl J Med. 1322
1282 pathways: A review. Brain Plast. 2017;2(2):127-52. 1987;316(21):1309-15. 1323
1283 [86] Brooks S, Burrin J, Cheetham ME, Hall GM, Yeo T, [97] Yuen EY, Liu W, Karatsoreos IN, Feng J, McEwen BS, 1324
1284 Williams C. The responses of the catecholamines and beta- Yan Z. Acute stress enhances glutamatergic transmission 1325
1285 endorphin to brief maximal exercise in man. Eur J Appl in prefrontal cortex and facilitates working memory. Proc 1326
1286 Physiol Occup Physiol. 1988;57(2):230-4. Natl Acad Sci U S A. 2009;106(33):14075-9. 1327
1287 [87] Colt EW, Wardlaw SL, Frantz AG. The effect of running on [98] Yuen EY, Liu W, Karatsoreos IN, Ren Y, Feng J, McEwen 1328
1288 plasma beta-endorphin. Life Sci. 1981;28(14):1637-40. BS, et al. Mechanisms for acute stress-induced enhance- 1329
1289 [88] Goldfarb AH, Hatfield BD, Armstrong D, Potts J. ment of glutamatergic transmission and working memory. 1330

f
1290 Plasma beta-endorphin concentration: Response to inten- Mol Psychiatry. 2011;16(2):156-70. 1331

roo
1291 sity and duration of exercise. Med Sci Sports Exerc. [99] de Quervain DJ, Roozendaal B, Nitsch RM, McGaugh JL, 1332
1292 1990;22(2):241-4. Hock C. Acute cortisone administration impairs retrieval 1333
1293 [89] McMurray RG, Forsythe WA, Mar MH, Hardy CJ. Exer- of long-term declarative memory in humans. Nat Neurosci. 1334
1294 cise intensity-related responses of beta-endorphin and 2000;3(4):313-4. 1335
1295 catecholamines. Med Sci Sports Exerc. 1987;19(6):570-4. [100] Tollenaar MS, Elzinga BM, Spinhoven P, Everaerd WA. 1336
1296 [90] Schwarz L, Kindermann W. Changes in beta-endorphin The effects of cortisol increase on long-term memory 1337

rP
1297 levels in response to aerobic and anaerobic exercise. Sports retrieval during and after acute psychosocial stress. Acta 1338
1298 Med. 1992;13(1):25-36. Psychol (Amst). 2008;127(3):542-52. 1339
1299 [91] Boecker H, Sprenger T, Spilker ME, Henriksen G, Kop- [101] Heffelfinger AK, Newcomer JW. Glucocorticoid effects 1340
1300 penhoefer M, Wagner KJ, et al. The runner’s high: on memory function over the human life span. Dev Psy- 1341
1301 Opioidergic mechanisms in the human brain. Cereb Cor- chopathol. 2001;13(3):491-513. 1342
1302 tex. 2008;18(11):2523-31. [102] Gutin B, Di Gennaro J. Effect of a treadmill run to 1343

tho
1303 [92] Hiura M, Sakata M, Ishii K, Toyohara J, Oda K, Nariai exhaustion on performance of long addition. Res Q. 1344
1304 T, et al. Central mu-opioidergic system activation evoked 1968;39(4):958-64. 1345
1305 by heavy and severe-intensity cycling exercise in humans: [103] Pontifex MB, Hillman CH, Fernhall B, Thompson KM, 1346
1306 A pilot study using positron emission tomography with Valentini TA. The effect of acute aerobic and resistance 1347
1307 11C-carfentanil. Int J Sports Med. 2017;38(1):19-26. exercise on working memory. Med Sci Sports Exerc. 1348
1308 [93] Saanijoki T, Tuominen L, Tuulari JJ, Nummenmaa L, 2009;41(4):927-34. 1349
Au
1309 Arponen E, Kalliokoski K, et al. Opioid release after [104] Tomporowski PD. Effects of acute bouts of exercise on 1350
1310 high-intensity interval training in healthy human subjects. cognition. Acta Psychol (Amst). 2003;112(3):297-324. 1351
1311 Neuropsychopharmacology. 2018;43(2):246-54. [105] Wrisberg CA, Herbert WG. Fatigue effects on the 1352
1312 [94] Hill EE, Zack E, Battaglini C, Viru M, Viru A, Hackney timing performance of well-practiced subjects. Res Q. 1353
1313 AC. Exercise and circulating cortisol levels: The intensity 1976;47(4):839-44. 1354
1314 threshold effect. J Endocrinol Invest. 2008;31(7):587-91. [106] Ferreira-Vieira TH, Guimaraes IM, Silva FR, Ribeiro FM. 1355
1315 [95] Kraemer RR, Blair S, Kraemer GR, Castracane VD. Alzheimer’s disease: Targeting the Cholinergic System. 1356
d

1316 Effects of treadmill running on plasma beta-endorphin, Curr Neuropharmacol. 2016;14(1):101-15. 1357
1317 corticotropin, and cortisol levels in male and female [107] McMorris T. Developing the catecholamines hypoth- 1358
cte

1318 10K runners. Eur J Appl Physiol Occup Physiol. esis for the acute exercise-cognition interaction in 1359
1319 1989;58(8):845-51. humans: Lessons from animal studies. Physiol Behav. 1360
1320 [96] Luger A, Deuster PA, Kyle SB, Gallucci WT, Montgomery 2016;165:291-9. 1361
LC, Gold PW, et al. Acute hypothalamic-pituitary-adrenal
rre
co
Un