Article Figures & Data
Figures
- Figure 1.
Task design, behavioral, and modeling results and EEG. a, Schematic representation of the experimental paradigm. After a variable delay (2–4 s), two stimuli (snack items) were presented on the screen for 1.25 s, and participants had to indicate their preferred item by pressing a button. The central fixation dimmed briefly when a response was registered. Snack stimuli shown here are for illustration purposes only. Participants viewed real branded items during the experiments. b, Behavioral performance (red circles) and modeling results (black crosses). Participants’ average (N = 21) reaction time (RT) and accuracy (top and bottom, respectively) improved as the value difference (VD) between the alternatives increased. An SSM that assumes a noisy moment-by-moment accumulation of the Vd signal fit the behavioral data well. c, Average (N = 21) model predicted evidence accumulation (EA; black) and EEG activity (red) in the time window leading up to the response (on average, 600–100 ms before the response), arising from a centroparietal electrode cluster (darker circles in the inset) that exhibited significant correlation between the two signals. Shaded error bars represent standard error across participants (reproduced with permission from Pisauro et al., 2017).
- Figure 2.
EEG energy of electrodes is a good correlate of energy consumption across voxels for modest levels of dipole coherences. a, Effect of dipole coherences on the maximum of the error percentage (‘z’) in approximating EEG energy as a sum of voxel BOLD values among 63 EEG channels averaged across simulations. b, Same as a but for the average of ‘z’ rather than its maximum across EEG electrodes. c, The histogram of the maximum of ‘z’ among EEG channels for the highest coherence level simulated (coherence = 0.2).
- Figure 3.
GC shows a significant positive correlation with EEG energy during food choice (GLMs 1–2). Group-average activation map (t stats) for the “EEG energy” regressor in GLM2 (step-wise regression on the residuals of original GLM done by Pisauro et al., 2017) showing activity in the bilateral insular, opercular and inferior somatosensory cortices; p < 0.05, cluster-corrected (right cluster = 293, left cluster = 218 > threshold = 136); a, axial and multiple coronal views, b, lateral and medial views on the inflated cortex. See Extended Data Figure 3-1 for the illustration of the regressors used in the single-subject GLMs. See Extended Data Figure 3-2 for the activation maps regarding to the nuisance regressors.
- Figure 4.
Right GC activation is robust to the addition of visual stimulus offset as a nuisance regressor (GLMs 3–9). Group-average activation map (t stats) for the “EEG energy” regressor in GLM4 showing activity in the right insular, opercular and inferior somatosensory cortices (p < 0.05, cluster-corrected (cluster = 214 > threshold = 121); a, axial and multiple coronal views, b, lateral and medial views on the inflated cortex. See Extended Data Figure 4-1 to compare the effect of addition of “vstim-off” regressor on the activation map for the raw EEG regressor. See Extended Data Figure 4-2 for a similar result for EEG energy when the raw EEG and EEG energy regressors are simultaneously used in the second-step GLM. See Extended Data Figure 4-3, for activations revealed by the raw EEG and EEG energy if instead of vstim-off, one uses a boxcar function for the duration of visual stimulus.
- Figure 5.
Right GC activation is robust to normalization of EEG-driven regressors (GLMs 10–11). Group-average activation map (t stats) for the normalized “EEG energy” regressor in GLM11 showing activity in the insular, opercular and inferior somatosensory cortices (p < 0.05, cluster-corrected (cluster = 567 > threshold = 295); a, axial and multiple coronal views, b, lateral and medial views on the inflated cortex. See Extended Data Figure 5-1 for variability of average energy of EEG over a total run between subjects.
- Figure 6.
Higher order EEG powers 3 and 4 do not reveal significant positive clusters of brain activation during food choice (GLMs 12–13). Group-average activation map (t stats) for higher powers of EEG in GLM12 and GLM13. a, Negative correlations with EEG pow3 (p < 0.05, cluster-corrected; clusters > threshold = 121). b, No significant correlations with EEG pow4 (p < 0.05, cluster-corrected; cluster threshold = 71).
Tables
GLM
indexSignal to
regressRegressors GLM1 BOLD vstim – VD – rt – EEG GLM2 Residual of GLM1 EEG energy GLM3 BOLD vstim – VD – rt – vstim
off – EEGGLM4 Residual of GLM3 EEG energy GLM5 BOLD vstim – VD – rt – vstim off GLM6 Residual of GLM5 EEG, EEG energy GLM7 BOLD vstim (onset) – VD – rt
– vstim boxcarGLM8 Residual of GLM7 EEG GLM9 Residual of GLM7 EEG energy GLM10 Residual of GLM5 EEG (normalized) GLM11 Residual of GLM5 EEG energy (normalized) GLM12 Residual of GLM3 EEG pow3 GLM13 Residual of GLM4 EEG pow 4 - Table 2
Number of voxels and location of peak activity for both positive and negative clusters
Region/cluster #(voxels) Hemisphere Peak X Peak Y Peak Z BA Energy, GLM2 (+): Inferior somatosensory (gustatory cortex) 293 Right −56 8 14 43/6 Insula (gustatory cortex) 213 Left 4 12 20 13 Energy, GLM4 (+): Inferior somatosensory (gustatory cortex) 214 Right −54 8 14 43/6 Energy, GLM6 (+): Inferior somatosensory (gustatory cortex) 213 Right −54 8 14 43/6 Energy, GLM9 (+): Inferior somatosensory (gustatory cortex) 219 Right −54 8 14 43/6 Energy, GLM11 (+): Insula (gustatory cortex) 567 Right −44 10 20 13 Energy, GLM4 (–): Superior frontal gyrus 10945 Right −18 −56 30 9 Inferior parietal lobe 3351 Right −48 54 56 40 Supramarginal gyrus 2157 Left 58 54 36 40 Superior temporal gyrus 2048 Left 30 −14 −34 38 Middle temporal gyrus 1668 Right −52 22 −12 21 Cingulate gyrus 928 Left 12 44 30 31 Cuenus 872 Left 28 90 22 19 Cingulate gyrus 791 Right −4 44 34 31 Middle occipital gyrus 705 Left 38 64 −2 37 Declive (cerebellum) 683 Right −34 68 −26 — putamen 603 Right −22 −12 0 — Superior temporal gyrus 528 Right −44 −10 −20 38 Inferior frontal gyrus 488 Left 28 −10 −18 47 Caudate head 257 Left 8 −4 6 — Thalamus 150 Right −14 12 12 — Culmen (cerebellum) 148 Right −30 46 −32 — EEG pow3, GLM12 (–): Middle frontal gyrus 275 Left 40 −50 −10 11 Inferior parietal lobe 140 Right −46 68 48 40 Inferior parietal lobe 135 Left 40 66 48 40 BA, Broadman Area.
Extended Data Figure 3-1
The regressors used in fMRI analyses. a, The four nuisance regressors; the visual onset, the value difference, the reaction time, and the visual offset regressors. b, The regressors of interest; the raw EEG and the EEG energy regressors. Download Figure 3-1, TIF file.
Extended Data Figure 3-2
Activation maps for the nuisance regressors (GLM 3). Group-average activation map (t stats) for the nuisance regressors all with p < 0.01 and cluster-corrected: (a) visual onset (vstim) regressor, (b) visual offset (vstim-off) regressor, (c) value difference (Vd) regressor, and (d) reaction time (rt) regressor. Download Figure 3-2, TIF file.
Extended Data Figure 4-1
Comparing the activity map for the raw EEG regressor with and without considering stimulus offset as a nuisance regressor (GLMs 1, 3) Group-average activation map for the raw EEG regressor (a) without considering “vstim-off” nuisance regressor as in GLM1 showing the activity in pMFC and premotor cortex (p < 0.05, cluster = 1281> threshold = 911) and in (b) with considering “vstim-off” nuisance regressor, GLM3 (no significant activity with p < 0.05). Download Figure 4-1, TIF file.
Extended Data Figure 4-2
Robustness of the activity in the right GC for simultaneous regression with raw EEG and EEG energy regressors in the second-step GLM (GLMs 5, 6). Group-average activation map (t stats) for the “EEG energy” regressor in GLM6 showing activity in the insular, opercular, and inferior somatosensory cortices (p < 0.05, cluster-corrected (cluster = 213 > threshold = 146). a, Axial and multiple coronal views. b, Lateral and medial views on the inflated cortex. Download Figure 4-2, TIF file.
Extended Data Figure 4-3
Activations revealed by the raw EEG and EEG energy when using a boxcar function for the duration of visual stimulus (GLMs 7–9). Group-average activation map (t stats) for (a) the “EEG energy” regressor in GLM9 (p < 0.05, cluster-corrected, cluster = 219 > threshold = 138) and (b) the raw EEG regressor in GLM8 with no significant (p < 0.05) activity. Download Figure 4-3, TIF file.
Extended Data Figure 5-1
Between subject variability of average energy of the “raw EEG” over a whole run. a, Average energy of the electric potential over all decision periods in an experimental run corresponding to the best electrode of each subject. b, Histogram of average energy of subjects’ best electrode EEG Download Figure 5-1, TIF file.
In this issue
