Journal: Nature Communications

Article Title: Enhanced role of the entorhinal cortex in adapting to increased working memory load

doi: 10.1038/s41467-025-60681-w

Figure Lengend Snippet: A Schematic illustration of the WM task and intracranial EEG (iEEG) recording sites in the entorhinal cortex (EC), hippocampus (Hipp), and lateral temporal cortex (LTC). B Under medium-to-high load conditions, decoding accuracy based on EC power features was higher than that derived from the hippocampus or LTC. C Cross-regional decoding, in which decoders trained on one region’s data were tested on another, revealed that EC-based decoders demonstrated the highest generalization under medium-to-high load conditions. D Residual decoding analysis showed that removing neural activity shared with the EC significantly reduced decoding accuracy in the hippocampus and LTC under medium-to-high load conditions. E Functional connectivity analysis indicated that the phase locking value (PLV) between the EC and other regions increased with enhanced WM load. The brain ( A , C ) was visualized by the BrainNet Viewer toolbox ( www.nitrc.org/projects/bnv/ ) .

Article Snippet: We utilized BrainNet Viewer in MATLAB (MathWorks) to visualize all the recording sites, as depicted in Fig. .

Techniques: Derivative Assay, Activity Assay, Functional Assay

Journal: Nature Communications

Article Title: Enhanced role of the entorhinal cortex in adapting to increased working memory load

doi: 10.1038/s41467-025-60681-w

Figure Lengend Snippet: A Each trial began with a 1 s fixation screen, followed by a 2 s presentation of four, six, or eight letters. After letters disappeared, there was a 3 s maintenance period with a black square shown. Participants responded whether a probe letter was part of the original set by pressing “IN” or “OUT”. B Channel locations of all participants included 91 channels in the hippocampus (Hipp; light red), 46 channels in the entorhinal cortex (EC; light blue), and 136 channels in the lateral temporal cortex (LTC; light green). The brain was visualized by the BrainNet Viewer toolbox ( www.nitrc.org/projects/bnv/ ) . C We conducted binary classification (load 4 vs load 6, load 6 vs load 8). Time-frequency analysis was performed on each trial to obtain power spectra in the hippocampus, EC, and LTC. For each classification task, 70% of the data was used for training and 30% for testing with a linear SVM classifier. D The decoding accuracy for load 4 vs load 6 did not show significant differences among the hippocampus, EC, and LTC across all cross-validations ( n = 100, two-sided permutation t test: EC vs hippocampus: p = 0.768; EC vs LTC: p = 0.379; LTC vs hippocampus: p = 0.690; see distribution with 150 iterations in Supplementary Fig. ). The EC exhibited the highest decoding accuracy for load 6 vs load 8 ( n = 100 cross-validations; two-sided permutation t test: all ps < 0.001). *** p < 0.001. E Differences in decoding accuracy between low-to-medium and medium-to-high load conditions were smallest in the EC ( n = 100 cross-validations; two-sided permutation t test: hippocampus vs EC: p < 0.001; LTC vs EC: p = 0.005; hippocampus vs LTC: p = 0.047). * p < 0.05, ** p < 0.01, *** p < 0.001. In the box plots shown in ( D , E ), the center line represents the median, and the edges of the box correspond to the lower and upper quartiles, respectively. The whiskers extend to the minimum and maximum data points at most 1.5 times the interquartile range. Source data are provided as a Source Data file.

Article Snippet: We utilized BrainNet Viewer in MATLAB (MathWorks) to visualize all the recording sites, as depicted in Fig. .

Techniques:

Journal: Nature Communications

Article Title: Enhanced role of the entorhinal cortex in adapting to increased working memory load

doi: 10.1038/s41467-025-60681-w

Figure Lengend Snippet: A Schematic of cross-regional decoding analysis. Using the entorhinal cortex (EC) as an example, we trained classifiers using power features from EC for each trial and predicted the load using power from the hippocampus (Hipp) for both low-to-medium and medium-to-high load conditions. The specific decoding steps were the same as shown in Fig. . For all brain regions, models were trained using their own power features and tested on data from the other two brain regions. The generalization of each brain region was determined by averaging its accuracy when tested on data from the other two brain regions (hippocampus: light red; EC: light blue; lateral temporal cortex: LTC, light green). The brain was visualized by the BrainNet Viewer toolbox ( www.nitrc.org/projects/bnv/ ) . B Accuracy matrix of cross-regional decoding on low-to-medium load (left) and medium-to-high load (right). The rows of the matrix represented the regions used for training, while the columns denoted the regions employed for testing, with the values representing the average accuracy. C The averaged cross-regional decoding accuracy for load 4 vs load 6 did not differ significantly among hippocampus, EC, and LTC across all cross-validations ( n = 100; two-sided permutation t tests: EC vs hippocampus: p = 0.637; EC vs LTC: p = 0.128; LTC vs hippocampus: p = 0.141). The EC showed the highest cross-regional decoding accuracy for load 6 vs load 8 across all cross-validations ( n = 100; two-sided permutation t tests: EC vs hippocampus: p < 0.001; EC vs LTC: p = 0.002; LTC vs hippocampus: p < 0.001). ** p < 0.01, *** p < 0.001. The center line represents the median, and the edges of the box correspond to the lower and upper quartiles, respectively. The whiskers extend to the minimum and maximum data points at most 1.5 times the interquartile range. Source data are provided as a Source Data file.

Article Snippet: We utilized BrainNet Viewer in MATLAB (MathWorks) to visualize all the recording sites, as depicted in Fig. .

Techniques:

Journal: Nature Communications

Article Title: Intracortical recordings reveal vision-to-action cortical gradients driving human exogenous attention

doi: 10.1038/s41467-024-46013-4

Figure Lengend Snippet: A Left: Illustration of the localization in normalized space (MNI152) of the contacts included in the analysis (black circles; n = 1403) in the left hemisphere (LH; n = 671) and in the right hemisphere (RH, n = 732), pooled across patients. Each localization is the mean coordinates of the two contacts composing the contact’s bipolar montage. To reveal prototypical temporal patterns simultaneously across all conditions, the trajectories across the 8 condition dimensions of the mean high-frequency broadband (HFBB) target-locked activity of 664 significantly responsive contacts (significant time-point-by-time-point t -test for at least 100 ms in one of the experimental conditions compared to baseline), were clustered using a custom-made trajectory K -means approach. Right: Example of target-locked mean normalized HFBB responses of one contact in the right angular gyrus in Congruent (full lines) and Incongruent (dashed lines) trials, at short-SOA (blue) and long-SOA (red), with targets contralateral or ipsilateral to the contact. Dashed vertical lines represent onsets of the target (black), short-SOA (blue), and long-SOA (red) cues. Shaded areas represent SEM across trials. Brain visualization was done using BrainNet Viewer Matlab toolbox (Xia M, Wang J, He Y (2013) BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics. PLoS ONE 8(7): e68910. doi:10.1371/journal.pone.0068910). B Prototypical temporal profiles of contact clusters across conditions: Trimmed-mean target-locked activity profiles of three contact clusters across the 8 conditions (Congruent/Incongruent Trial × short-SOA/long-SOA × Ipsilateral target (Ipsi)/contralateral target (Contra)). Cluster 1 (yellow) shows contralateral fast responses, with cue-target activity segregation at both SOAs; Cluster 2 (red) shows bilateral slower responses with spatial sensitivity, with cue-target activity segregation at long-SOA but response integration in short-SOA; and Cluster 3 (green) shows bilateral slowest responses with stimulus-type sensitivity, with cue-target activity segregation at long-SOA but response integration at short-SOA. Dashed vertical lines represent target onset (black) and cue onset at short-SOA (blue) and long-SOA (red). C Temporal gradient of target-locked activity (trimmed-mean) of the three clusters. The Black dashed line depicts the target onset. D Scatter plot of peak times of mean target-locked activity of contacts of Cluster 1 (yellow circles), Cluster 2 (red circles), and Cluster 3 (green circles), in Congruent (x-axis) and Incongruent ( y -axis) conditions, showing a significant temporal gradient (Mixed 2-way ANOVA, Cluster main effect p < 0.001, η 2 = 0.378; linear polynomial contrast: p ≤ 0.001). Squares represent mean peak time; the Dotted gray line denotes the equity line; Shaded areas represent peak time distributions.

Article Snippet: Contact localizations in standard MNI152 space were visualized with the BrainNet Viewer Matlab toolbox ( http://www.nitrc.org/projects/bnv/bnv/ ; Matlab R2016b and R2020a, The MathWorks, Inc.).

Techniques: Activity Assay

Journal: Nature Communications

Article Title: Intracortical recordings reveal vision-to-action cortical gradients driving human exogenous attention

doi: 10.1038/s41467-024-46013-4

Figure Lengend Snippet: A Clusters’ spatial profile. Illustration of the localization of the contacts composing each cluster: Cluster 1 (yellow), Cluster 2 (red), Cluster 3 (green). For each cluster, dots represent contacts’ localization in dorsal (middle), lateral (top), and medial (bottom) views of the right hemisphere (RH; right) and of the left hemisphere (LH; left). B Core–Periphery gradient: Clusters’ anatomical localization follows core–periphery gradients , where Cluster 1’s contacts are the most peripheral, and Cluster 3’s contacts are closest to core regions. C Left: Scatter plot of contacts localization along core–periphery gradients (Cluster 1—yellow circles, n = 62 independent contacts; Cluster 2—red circles, n = 97 independent contacts; Cluster 3—green circles, n = 67 independent contacts; rectangles represent clusters’ mean). Right: Violin plots of contacts localization along Core-Periphery gradients for Cluster 1 (yellow), Cluster 2 (red) and Cluster 3 (green), showing a significant core-periphery gradient (Gradient 1: 1-way ANOVA, p < 0.001, η 2 = 0.06; linear polynomial contrast: p ≤ 0.001; Gradient 2: 1-way ANOVA, p < 0.001, η 2 = 0.28; linear polynomial contrast: p ≤ 0.001; n = 232 independent contacts in total). The box centerlines depict the medians, the bounds of the box depict the 75%/25% quartiles, and the whiskers depict the top & bottom 25% percentiles. D Cluster contacts are structurally connected: Corrected tractography t-maps, showing the significant white matter voxels, which connect pre and post-rolandic contacts within each cluster (Cluster 1—yellow; Cluster 2—red, Cluster 3—green), derived from a fiber tracking analysis of 176 healthy individuals. E Contacts’ receptive windows lengthen along the cluster gradient: Raincloud plots of individual contacts’ receptive window length (circles), showing a significant linear lengthening from Cluster 1 (yellow, n = 62 independent contacts), to Cluster 2 (red, n = 97 independent contacts), to Cluster 3 (green, n = 67 independent contacts; 1-way ANOVA : p < 0.001, η 2 = 0.11; linear polynomial contrast: p ≤ 0.001; n = 232 independent contacts in total). The box centerlines depict the medians, the bounds of the box depict the 75%/25% quartiles, and the whiskers depict the top & bottom 25% percentiles. Brain visualization was done using BrainNet Viewer Matlab toolbox (Xia M, Wang J, He Y (2013) BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics. PLoS ONE 8(7): e68910. doi:10.1371/journal.pone.0068910).

Article Snippet: Contact localizations in standard MNI152 space were visualized with the BrainNet Viewer Matlab toolbox ( http://www.nitrc.org/projects/bnv/bnv/ ; Matlab R2016b and R2020a, The MathWorks, Inc.).

Techniques: Derivative Assay

Journal: Nature Communications

Article Title: Intracortical recordings reveal vision-to-action cortical gradients driving human exogenous attention

doi: 10.1038/s41467-024-46013-4

Figure Lengend Snippet: Mean target-locked long-SOA activity in Cluster 1 (yellow), Cluster 2 (red), and Cluster 3 (green), was computed over trials pooled across all cluster contacts for Congruent trials (full lines) and Incongruent trials (dashed lines). A In Cluster 1, no significant Congruence effect was observed in a 3-way ANOVA with Holm multiple comparisons correction. B In Cluster 2 activity in Congruent and Incongruent trials (IOR-related) differed significantly in a 3-way ANOVA with Holm multiple comparisons correction at 0.24–0.3 s post target (shaded red areas; Congruence main effect: largest p = 0.002), and a significant hemispheric difference between IOR-related responses was observed at 0.14–0.022 s post target (shaded brown area; Hemisphere × Congruence interaction: largest p = 0.03; Diagonally striped areas represent significant Congruence × Hemisphere post hoc comparisons ( p < 0.05)). C In Cluster 3, activity in Congruent and Incongruent trials differed significantly in a 3-way ANOVA with Holm multiple comparisons correction at 0.66–0.68 s post target (green shaded area; Congruence main effect: largest p = 0.003). A-C. Shaded areas around traces depict SEM; Dashed vertical lines represent target onset (black) and cue onset (red) at the long-SOA Condition. D Representative examples of HFBB power IOR-related activity in the Congruent (full line) & Incongruent (dashed line) long-SOA conditions of individual contacts of Cluster 2, shaded areas around traces depict SEM. p Values are Holm corrected. Brain visualization was done using BrainNet Viewer Matlab toolbox (Xia M, Wang J, He Y (2013) BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics. PLoS ONE 8(7): e68910. doi:10.1371/journal.pone.0068910).

Article Snippet: Contact localizations in standard MNI152 space were visualized with the BrainNet Viewer Matlab toolbox ( http://www.nitrc.org/projects/bnv/bnv/ ; Matlab R2016b and R2020a, The MathWorks, Inc.).

Techniques: Activity Assay

Journal: Nature Communications

Article Title: Intracortical recordings reveal vision-to-action cortical gradients driving human exogenous attention

doi: 10.1038/s41467-024-46013-4

Figure Lengend Snippet: A RT modulates target-locked neural activity, pooled across conditions and color-coded from fastest (Magenta) to slowest (yellow) RT bin. A dashed vertical black line represents target onset; Color-coded dots at the top of each panel represent mean RT for each bin (pink—fastest RT to yellow—slowest RT); 1-way repeated measures ANOVA, Holm multiple comparisons correction. Top: Late RT modulation of activity in Cluster 1 (yellow): Main effect of RT bin at 0.5–0.54 and 0.56–0.68 s post-target (shaded yellow area; largest p = 0.002). Middle: RT modulation of neural response offset in Cluster 2 (red): Main effect of RT bin at 0.3–0.56 s post target (shaded red area; largest p = 0.028). Bottom: RT modulation of response in Cluster 3 (green): Main effect of RT bin at 0.28–0.3 and 0.4–0.42 s post target (shaded green area; largest p = 0.007). B Examples of single contact neural activity in the fastest (pink) and slowest (yellow) thirds of trials for the three target-locked clusters. Vertical dashed black lines represent target onset; Vertical full lines denote mean RT for fastest (magenta) and slowest (yellow) trials, shaded areas around traces depict SEM. C Visual modulation of response-locked neural activity pooled across conditions, color-coded from fastest (Magenta) to the slowest (yellow) bin. The dashed vertical gray line represents RT; color-coded dots at the top of each panel represent the mean target onset time for each bin (pink—earliest onset to yellow—latest onset); 1-way repeated measures ANOVA, Holm multiple comparisons correction. Top: target onset time modulates activity in the RT-Cluster 1 (yellow): Main effect of RT-bin at 0.12–0.10 s pre-response (shaded yellow area; largest p = 0.04). Target onset time modulates activity in the RT-Cluster 2a (orange): Main effect of RT bin at 0.70–0.68 s, 0.52–0.50 s, and 0.30–0.20 s pre-response (shaded orange area; largest p = 0.004). No significant modulation in RT-Cluster 2b (turquoise) and RT-Cluster 3 (green). Arrows between panels A and C denote the contingency between target-locked and response-locked clusters (see Fig. ). D Examples of single contact neural activity in the fastest (pink) and slowest (yellow) thirds of trials for RT-Cluster 1 and RT-Cluster 2a. Vertical dashed gray lines represent RT; Vertical full lines denote the mean target onset time for the fastest (magenta) and slowest (yellow) trials, shaded areas around traces depict SEM. p Values are Holm corrected. Brain visualization was done using BrainNet Viewer Matlab toolbox (Xia M, Wang J, He Y (2013) BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics. PLoS ONE 8(7): e68910. doi:10.1371/journal.pone.0068910).

Article Snippet: Contact localizations in standard MNI152 space were visualized with the BrainNet Viewer Matlab toolbox ( http://www.nitrc.org/projects/bnv/bnv/ ; Matlab R2016b and R2020a, The MathWorks, Inc.).

Techniques: Activity Assay