Table 1 Information summaries of studies focusing on associations between dynamic GABA modulation and behavior
From: MRS-assessed brain GABA modulation in response to task performance and learning
Author, year | Sample size | Strength Coil Sequence | Voxel (mm3) Time Points | Metabolite/RC | Data analysis software | Task/intervention | Behavioral measures | Modulation and correlation |
|---|---|---|---|---|---|---|---|---|
Perceptual performance and learning | Â | Â | Â | Â | Â | |||
Bang et al. 2018 | 24 H | 3T 32-channel MEGA-PRESS PRESS | OCC 20 × 20 × 20 3 time points: pre-, post-, 3.5 h post-learning | GABA Glu/NAA | LCModel | An orientation detection learning task | (1) detection threshold | E/I modulation: E/I ratio increased after reactivation and returned to baseline 3.5 h after reactivation |
BednaÅ™Ãk et al. 2015 | 12 H | 7T Gradient/shim sLASER | OCC 20 × 20 × 20 continuously measure during visual stimulation and rest | Glu Lac GABA/ | LCModel | Red–black flickering checkerboard | No | GABA and Glu modulations: GABA did not change significantly with visual stimulation Glu increased with visual stimulation |
Boillat et al. 2020 | 20 H (posBOLD) 21 H (negBOLD) | 7T quadrature coil Semi-adiabatic SPECIAL | OCC 18 × 18 × 18 continuously measure during STIM and REST | GABA Glu La/water | Home-built MATLAB functions and LCModel | A full screen radial checkerboard (posBOLD) A small central checkerboard (negBOLD) | No | GABA and Glu modulations: GABA decreased during the stimulation of negBOLD GABA did not change significantly during the stimulation of posBOLD Glu decreased during the stimulation of negBOLD but increased during posBOLD |
Frangou et al. 2018 | Cohort 1: SN 17H FD 21H cohort 2: SN 17H FD 21H | 3T 32-channel 2D J-PRESS | Right OCT 25 × 25 × 25 2 time points: Pre- and post-task OCT and PPC 25 × 25 × 25 2 time points: Pre- and post-task | GABA/tCr | ProFit | A radial or concentric Glass patterns learning task | (1) Learning rate of perceptual sensitivity (d’) across runs (2) Change in perceptual sensitivity (d’) (early vs late run) | GABA modulation: GABA did not change significantly in either voxel GABA correlation: cohort 1: (r = − 0.43, CI = [− 0.74, 0.07]) measure (1) in SN & OCT GABA change (r = 0.54, CI = [0.05, 0.85]) measure (2) in FD—OCT GABA change cohort 2: (r = − 0.43, CI = [− 0.75, − 0.02]) measure (1) in SN—OCT GABA (r = 0.55, CI = [0.10, 0.78]) measure (2) in FD—OCT GABA |
Frangou et al. 2019 | 47 H | 7T 32-channel sLASER | Left OCT and PPC 20 × 20 × 20 4 time points: baseline, first (block 1,2), second (3,4), third (5,6) | GABA Glu/tCr | LCModel | A radial or concentric Glass patterns learning task | (1) Learning improvement in each block (2) Individual biggest improvement | GABA modulation: In OCT, GABA levels decreased in SN and increased in FD In PPC, GABA levels increased in both SN and FD, no significant difference between tasks GABA correlation: (n = 16, r = − 0.43, CI = [− 0.75, − 0.02]) measure (2) in SN—OCT GABA change (n = 19, r = 0.55, CI = [0.10, 0.78]) measure (2) in FD—OCT GABA change |
Frank et al. 2022 | 13 Children 14 Adults | 3T 64-channel MEGA-PRESS | Early visual cortex 20 × 20 × 20 Continuously measure at pre-, during and post-training | GABA/NAA + NAAG | LCModel | An orientation detection learning task | (1) Detection threshold | GABA modulation: GABA increased during visual training in children, and it persisted after training ended GABA did not change significantly during visual training in adults |
Heba et al. 2016 | 24 H | 3T 32-channel MEGA-PRESS | left and right SM1 30 × 30 × 30 2 time points: Baseline, post-stimulation | GABA/water | Gannet | Repetitive somatosensory stimulation | (1) 2ptD changes before and after the stimulation | GABA modulation: GABA did not change significantly after the perceptual stimulation |
Ip et al. 2019 | 24 H | 7T 32-channel semi-LASER | OCC 20 × 20 × 20 4 time points: Baseline, visual stimulation (4 types) | GABA Glu/water | MRspa LCModel | Full field checkerboards with contrast (3.1%, 12.5%, 50%, and 100%) | No | GABA and Glu modulations: GABA did not change significantly under different contrast levels as compared to baseline Glu increased significantly in stimulation with 100% contrast as compared to baseline |
Koush et al. 2021 | 20 H | 4T Single-channel quadrature coil MRGA-STEAM | OCC and PCC Continuously measure during visual stimuli and auditory emotion task | GABA Glx Lac/NAA | Â | A flashing checkboard (OCC activation) Identification of auditory emotion (PCC deactivation) | No | GABA and Glx modulations: GABA decreased during visual stimuli but increased during identification of auditory emotion task Glx increased during visual stimuli but did not change during identification of auditory emotion task |
Kurcyus et al. 2018 | 30 H | 3T 12-channel MEGA-PRESS | OCC 25 × 25 × 25 3 time points: In eye closed In open in dark In open with stimuli | GABA Glx/Cr | Gannet LCModel | An increasing visual input task | (1) Discrimination threshold | GABA and Glx modulations: GABA decreased significantly from eye-closed to eye-open state GABA did not change significantly with visual checkerboard stimuli Glx increased with visual checkerboard stimuli GABA correlation: [r (23) = − 0.54, p = 0.008] measure (1) and GABA with visual checkerboard stimuli |
Lea-Carnall et al. 2020 | 14 H | 3T 8-channel MEGA-PRESS | Left S1 20 × 30 × 20 2 time points: First 12 m of tactile stimulation, last 12 m of tactile stimulation | GABA Glx/NAA | jMRUI AMARES | Tactile stimulation at resonance frequency (23 Hz) tactile stimulation above resonance frequency (39 Hz) | (1) Tactile discrimination test | GABA and Glx modulations: GABA decreased in response to above-resonance stimulation GABA did not change significantly in response to at-resonance stimulation Glx did not change significantly in both stimulation frequencies |
Lin et al. 2012 | 10 H | 7T 16-channel STEAM | OCC 20 × 20 × 20 3 time points: Rest, stimuli, recovery | GABA Glu NAA/water | jMRUI LCModel | A visual perceptual task | No | GABA and Glu modulations: GABA decreased under visual stimulation but did not reach significance and back to resting state in recovery Glu increased during the visual stimulation and back to rest state in recovery |
Lunghi et al. 2015 | 19 H | 7T 16-channel sLASER | OCC, PCC 20 × 20 × 20 2 time points: Pre- and post- monocular deprive | GABA/water/tNAA | LCModel | A binocular rivalry task (pre-, post-), 150 min monocular deprivation | (1) Mean duration (pre) (2) Mean duration (post) | GABA modulation: GABA decreased after monocular deprivation, in OCC not in PCC GABA correlation: (r = 0.62, p = 0.006) reduction of OCC GABA & increasement of the predominance of the dominant eye |
Mangia et al. 2007 | 12 H | 7T Head gradient coil STEAM | OCC 20 × 22 × 20 Continuously measure during visual stimulation on and off | Glu Lac GABA/water | LCModel | A radial red/black checkboard | No | GABA and Glu modulations: GABA did not change significantly during the visual stimulation Glu increased during the visual stimulation |
Mekle et al. 2017 | 20 H | 7T Quadrature coil SPECIAL | Right OCC 20 × 22 × 20 2 time points: Active stimulation Control stimulation | GABA Glx/no | LCModel | Half of a rotating checkerboard in the form of a torus | No | GABA modulation: GABA decreased in the active stimulation condition as compared to the control stimulation condition |
Schaller et al. 2013 | 6 H (active) 4 H (control) | 7T quadrature coil SPECIAL | OCC & POC 20 × 22 × 32 Continuously measure during visual stimulation on and off | GABA Glu Lac/no | FIDs LCModel | A black–grey checkerboard | No | GABA and Glu modulations: GABA did not modulate significantly Glu increases within the first two minutes of the primary visual activation |
Shibata et al. 2017 | 12 H (no-overlearning) 12 H (overlearning) | 3T 32-channel MEGA-PRESS PRESS | OCC 20 × 20 × 20 Overlearning (3) Pre-, post1, post2 Interference (2) pre-, between training 1 and 2 | Glu/GABA/Cr | LCModel | An orientation detection learning task | (1) Detection threshold | E/I modulation: The modulation of E/I was specific to learning and not to non-learning training E/I increased in non-overlearning and decreased in overlearning 30 min after training E/I returned close to baseline in both groups at 3.5 h after training E/I correlation: (r = 0.64, p = 0.024) E/I reduction and anterograde interference to training 2 |
van Vugt et al. 2020 | 25 H | 3T 32-channel PRESS | Left SM1 30 × 30 × 20 During task (16 min) | GABA lactate/tNAA | FID-A LCModel | An audio-motor mapping learning task | (1) Tracking error | GABA modulation: GABA increased during motor learning GABA correlation: (r = − 0.49, p = 0.014) GABA increase & audio-motor learning |
Ziminski et al. 2023 | 22 M | 3T 32-channel MEGA-PRESS | OCT & PPC 20 × 20 × 25 3 time points: Baseline, pre-, post-training | GABA Glu/water/NAA | MRspa LCModel | A radial or concentric Glass patterns learning task (SN task) | (1) Task accuracy | GABA modulation: It did not report GABA and Glu correlations: (r = − 0.60, p = 0.012) bigger reduction of OCT GABA and bigger behavioral progress (r = 0.36, p = 0.154) no correlation between PCC GABA changes & behavioral progress |
Motor performance and learning | Â | Â | Â | Â | Â | |||
Andrushko et al. 2023 | 15 H | 3T 32-channel MRSI | Left M1, right M1, Left SMA, right SMA 2 time points: Before and after the force matching task | GABA Glx/tCr | In-house scripts | A handgrip force matching task (right hand) A response task | (1) Reaction times | GABA modulation: GABA did not modulate significantly after the task GABA correlation: The faster measure (1) & larger decreased GABA levels in left M1, left SMA and right SMA respectively The faster measure (1) & larger increased Glx levels in right M1 |
Bell et al. 2023 | 9 H | 3T 32-channel MEGA-PRESS | left SM1 25 × 25 × 25 Continuously measure through baseline and 6 blocks about 30 min | GABA Glu Glx/tCr | Gannet LCModel | A serial reaction time task | (1) Reactions times | GABA and Glu modulations: No significant modulation in GABA and Glu in either task Glx correlation: higher Glx levels in Block 1 & worse progress in block 6, median progress and progress of the best block |
Chalavi et al. 2018 | 32 Y 28 O | 3T 32-channel MEGA-PRESS | OCC and left SM1 4 time points: Day 1-pre Day 1-post Day 3-pre Day 3-post | GABA/water | Gannet | A bimanual coordination task | (1) Initial error rate (2) Retention error | GABA modulation: OCC GABA decreased in random and increased in blocked group from pre- to post-training in combined YA and OA. The difference of GABA modulation following two schedules is significant in OA (p = 0.018, corrected), not in YA (p = 0.22, corrected) OCC GABA levels decreased in the blocked group and increased in the random group from Day 1 to Day 3 in OA (p = 0.008, corrected) but not in YA (p > 0.1) GABA correlation: (t = 0.26, p = 0.021) measure (1) & baseline SM1 GABA levels |
Chen et al. 2017 | 16 H | 7T 32-channel MM-eliminated MEGA-sLASER | Left SM1 15 × 20 × 30 Continuously measure during the 20-min task | GABA Glx NAA/tCr | In-house scripts | A hand clenching task | No | GABA modulation: GABA levels slowly decreased (− 15 ± 2%) to a stable level Glx levels increased (8 ± 2%) rapidly during the 10-min clenching period Both of them recovered towards baseline during the rest period after clenching |
Coxon et al. 2018 | 10 H | 3T 32-channel MEGA-PRESS | SM1, DLPFC 2 time points: Pre-, post-exercise | GABA Glx /water,Cr,NAA | LCModel | High-intensity interval exercise | No | GABA modulation: SM1 GABA increased post-exercise compared with the pre-measure DLPFC GABA did not change significantly Glx levels did not change significantly in both SM1 and DLPFC |
Floyer-Lea et al. 2006 | 13 H (learn) 10 H (move) 7 H (rest) | 3T coil: NS MEGA-PRESS | Left SM1 20 × 20 × 20 Continuously measure during 30-min task and 20 min after task | GABA Glx, Cr NAA/Cr | LCModel | A force tracing task | (1) Tracking errors | GABA modulation: Left SM1 GABA levels decreased (18%) during sequence learning and did not change in movement and rest condition. Right SM1 GABA levels did not change significantly Left SM1 GABA levels recovered 40% of the maximum decrease during 20 min after stop learning |
Kolasinski et al. 2019 | 11 H (learn) 12 H (move) 11 H (rest) | 7T 32-channel sLASER | Left SM1 20 × 20 × 20 Continuously measure through 6 blocks about 40 min | GABA/tCr | LCModel | A visually cued serial reaction time task | (1) Learning gain (RT) (2) Best performance | GABA modulation: In the learning group, SM1 GABA levels decreased significantly in the block 6 compared with the block 1, not in movement group and rest group GABA correlation: (r = 0.515, p = 0.049) measure (1) & block 1 GABA levels (r = 0.613, p = 0.015) measure (2) & block 1 GABA levels |
King et al. 2020 | 36 O | 3T 32-channel MEGA-PRESS | Right SM1 30 × 30 × 30 2 time points: Pre-, post-learning | GABA/water | Gannet | A motor sequence task, anodal (1 mA, 15 min), sham tDCS | (1) Reaction times | GABA modulation: GABA did not change significantly GABA correlation: SM1 GABA change & motor learning magnitude (r = − 0.41, p = 0.013), age (r = 0.45, p = 0.06), and baseline GABA (r = − 0.51, p = 0.001) |
Maddock et al. 2016 | 38 H | 3T 32-channel MEGA-PRESS PRESS | OCC, ACC 30 × 25 × 25 2 time points: Pre-, post-exercise | GABA Glu/NAA/water/Cr | jMRUI LCModel | Exercise on a bicycle ergometer | No | GABA and Glu modulations: OCC GABA, OCC Glu and ACC Glu increased following exercise |
Maruyama et al. 2021 | 25 H (learning) 12 H (control) | 7T 32-channel STEAM | Right M1 20 × 20 × 20 3 time points: Pre-, during, post-task | GABA Glu/tCr | LCModel | Motor sequence learning task | (1) Reaction times | GABA modulation: GABA did not change significantly during these 3 time points GABA correlation: Bigger GABA/Glu ratio reduction is associated with better performance improvement in the learning group |
Maes et al. 2022 | 28 Y 29 O | 3T 32-channel MEGA-PRESS | Left SM1 3 time points: Pre-, during, post-task | GABA/water | Gannet | Multidigit task | (1) Accuracy | GABA modulation: GABA decreased significantly during the task and returned to pre-levels in the post measurement GABA correlation: A bigger reduction of GABA was significantly related to better performance in older but poorer performance in young adults |
Nettekoven et al. 2022 | 17 H | 3T 32-channel MRSI Semi-LASER | Bilateral cerebellum 4 Time points: Baseline, early task, late task, post-task | GABA Glu /tCr | In-house scripts | A visuomotor task with no adaptation A visuomotor task with adaptation | (1) Error | GABA modulation: Learning-driven GABA levels in the left cerebellar nuclei, as measured by the difference between these two tasks, decreased, but not in the right cerebellar nuclei GABA correlation: Greater decreased GABA levels in the right cerebellar nuclei in the early training phase & better motor adaptation |
Sampaio-Baptista et al. 2015 | 22 H (high) 18 H (low) 20 H (control) | 3T 12-channel SPECIAL | Left SM1 20 × 20 × 20 3 time points: baseline, post 1: after 6-week train, post 2: after subsequent 4-week | GABA/Cr | LCModel | A three-ball juggling task | (1) Daily scores (2) learning rates | GABA modulation: GABA levels decreased significantly in Post 1 compared with the baseline measure and increased significantly in Post 2 compared with Post 1 in the low-intensity group GABA levels did not change significantly among three measures in the high-intensity group |
Streeter et al. 2010 | 34 H | 4T MEGA-PRESS | Left thalamus 3 time points: Baseline, after 12w but before 60Â m intervention, after 60Â s intervention | GABA/Cr | LCModel | Yoga Walking | (1) Mood and anxiety scales | GABA modulation: GABA did not change significantly in either group GABA correlation: Bigger GABA changes from scan 2 to scan 1 was related to better mood and lower anxiety |
Cognitive performance and learning | Â | Â | Â | Â | Â | Â | ||
Bezalel et al. 2019 | 31 H | 3T 12-channel MEGA-PRESS | dACC 40 × 25 × 10 3 times points: Baseline, during task, final rest | GABA Glx/water | In-house scripts | A reinforcement learning task | (1) Gain options/10 times (2) Overall gain options/condition (3) Learning rates | GABA and Glx modulations: dACC GABA and Glx did no change significantly between the initial and final rest Compared to the discrimination condition, dACC GABA levels increased and Glx/GABA levels decreased in the uncertainty condition (p < 0.05) GABA correlation: (r = − 0.592, p = 0.001) measure (1) & dACC GABA levels in NULL condition |
Frank et al. 2021 | 20 H | 3T MEGA-PRESS PRESS | Right PIVC 25 × 25 × 25 2 time points: Low attention load High attention load | GABA Glx/water | LCModel | Visual tracking task (low attention load and high attention load) | (1) Accuracy | GABA and Glx modulations: GABA did not change significantly in these 2 states Glx decreased from low to high visual attentional load |
Koolschijn et al. 2021 | 19 H | 7T 32-channel fMRI-fMRS sequence | OCC 20 × 20 × 20 Continuously measure during the task | GABA Glu/tCr | MRspa LCModel | Three stage interference tasks | (1) Accuracy | GABA and Glu/GABA modulations: GABA as well Glu/GABA ratio decreased in the right inferences relative to erroneous answers |
Kühn et al. 2016 | 19 H | 3T 32-channel SPECIAL | ACC 25 × 35 × 20 3 time points: Pre-, during, post-task | GABA Glu Gln tCr/water | LCModel | A color/word Stroop task | (1) Reaction times (2) errors | GABA and Glu modulations: GABA and Glu levels increased during the task and decreased at the post-task measure |
Michels et al. 2012 | 16 H | 3T MEGA-PRESS | Left DLPFC 25 × 40 × 30 Baseline and continuously measure during the task | GABA Glx/water | LCModel | Working memory task | (1) Accuracy | GABA and Glx modulations: GABA increased significantly initially and then decreased in the following runs Glx did not change significantly across the conditions |
Namgung et al. 2021 | 19 F | 3T 32-channel MEGA-PRESS | PFC 30 × 30 × 30 2 time points: Baseline, after training program | GABA/water | Gannet | 10-day relaxation training program | (1) Stress | GABA modulation: GABA decreased significantly after 10 days training program as compared to baseline |
Spurny et al. 2020 | 20 H | 3T 64-channel 3D-MRSI MEGA-LASER | (1) Hippocampus, insula, thalamus (2) 2 Time points: pre-, post-21-day | GABA Glx tNAA/tCr | In-house scripts | An associative facial learning task | (1) IRS (2) FRS (3) RI | GABA and Glx modulations: GABA or Glx did not change significantly in the three tested regions from the pre- to post-training GABA correlation: (r = 0.69, p = 0.013) measure (1) & hippocampal GABA levels no correlation between GABA in thalamus or insula and measure (1), (2) or (3) no correlation between Glx before or after learning and measure (1), (2) or (3) |
Vijayakumari et al. 2018 | 10 H | 3T 32-channel PRESS | Left DLPFC 20 × 20 × 20 3 time points: Baseline, during and post-task | GABA Glu Glx/no | LCModel | A working memory task | (1) Accuracy | GABA modulation: GABA did not change significantly during this task |