Article Figures & Data
Figures
- Figure 1.
Example of S1 layer 4 LFPs recorded during single- and two-digit stimulations. A, Location of the microelectrode array in S1. The blue dot represents the electrode at 0 mm. B–E, Single trials (gray) and average (colored thick line) LFP responses for all single digit stimulations (D2–D5) recorded from the electrode marked with the blue dot. F, Normalized mean area under the curve indicated as mean LFP response, according to recording electrode position (in millimeters). The colored transparent shapes indicate standard deviations. The spatial distribution of LFP responses was somatotopic, with each stimulation evoking a cortical activity profile gradually shifted along electrode position. G–I, Funneled cortical pattern of activity elicited by digit costimulation (thick solid lines), superimposed on that generated by single stimulation of the same digits (dashed lines).
- Figure 2.
Regression analysis of LFP. A, Example of mean LFP responses according to recording electrode position (see Fig. 1). Cortical pattern of activity evoked after D2, D3, and D4 single stimulation, D2D4 costimulation, and that obtained after the sum of D2 and D4 single-digit responses (D2 + D4). B–E, Linear regression between D2D4 activity pattern and each of the patterns shown in A. Note that the D2D4 costimulation-evoked activity pattern resembled that obtained after stimulation of the center digit, D3 (C). F, H, J, Mean R2 (ordinate axis) and proportion of significant regressions (diameter of shaded discs; F test, p < 0.05) obtained in the whole neuronal population for each of the costimulation patterns (D2D3, D2D4, and D2D5). G, I, K, Y-intercept distributions of the significant regressions between costimulation and single stimulation or addition activity patterns. A negative value indicates a greater activity for the costimulation (Wilcoxon test of the median versus 0; *p < 0.05; ***p < 0.001).
- Figure 3.
Single unit response properties and examples of recorded units. A, Number of spikes per trial generated by single- and codigit stimulation for all neurons (n = 162) and all trials recorded. B, Distribution of the number of spikes per trial and per unit. The median number of spikes is higher for the codigit than for the single-digit stimulation (Wilcoxon test, p = 0.000). C, Number of spikes per response, i.e., for every trial in which at least one spike was generated. D, Distributions of the mean number of spikes, per response per unit, for single-digit and codigit stimulation. The increased number of spikes per trial is not due to an increase in the number of spikes per response (Wilcoxon test, p = 0.54) but to an increased probability of response (B). E, F, Example of recorded units showing no additive or suppressive effect during costimulation. Left, mean wave form and neuron’s tuning curve represented as a mean response probability (± SD) for each single and codigit stimulation. Right, spike density function and raster plots for both single and combined stimulations (see color code matching digits in the diagrams). G, Unit that shows a distance-dependent suppression after D2 stimulation with an adjacent or nonadjacent digit. H, Unit responding to D2D3 and D2D4 costimulation as if D3, its preferred digit, were stimulated.
- Figure 4.
Population response modulation to costimulation. Proportion of units showing increased, decreased, or equal response probability to costimulation compared with that elicited by the stimulation of their preferred digit. Top row, edge units, i.e., with a maximal response probability to the stimulation of one of the costimulated digits. Bottom row, center units, i.e., with a maximal response probability to the center nonstimulated digit.
- Figure 5.
Cortical pattern reconstruction from the population responses. Mean normalized response probability (± standard error) to different stimulation conditions for neuronal populations sorted out according to the unit’s preferred digit, i.e., the digit whose stimulation elicited the highest probability of response. Population patterns for all neurons recorded after single-digit stimulation and D2D3 (A), D2D4 (B), or D2D5 (C) costimulation.
- Figure 6.
Unit response latency. A, Distribution of the first spike latency after single-digit and D2D4 stimulations for D3-preferring neurons. Note that these neurons responded to D2D4 as fast as to D3. B, Distributions of the first spike latency of the D2-, D3-, and D4-preferring neurons for D2D4 costimulation. There was no significant latency difference among these populations.
- Figure 7.
Decoding results. Confusion matrices obtained with a k nearest neighbors classifier (k = 4). A, D2D4 population response classification. The classifier was trained with D2, D3, and D4 trials. Among the whole population, 94 neurons were chosen randomly, and 50 random subsets of 20 trials were used for each selected neuron. For every subset of 20 trials, all 20 different combinations of 19 + 1 were used to train and test the classifier. This procedure was repeated 100 times. The confusion matrices show the distribution of the 100,000 (100 * 50 *20) classifier choices. The most often chosen digit is D3, the center nonstimulated digit (39.98% > 33.33%, chance level). B, Confusion matrix for a classifier trained with D3 and D2 + D4 trials. Neurons with their maximal response probability for D2 and D4 stimulation were grouped, as for trials where these digits were stimulated alone. This merging creates an artificial condition in which responses of D2 population for D2, and responses of D4 population for D4, are added. This condition allows us to verify whether the classifier would chose D3 over the simple addition of D2 and D4 responses. Decoding results indicate that addition of D2 + D4 was chosen more often. C, D2D5 population response classification. The classifier was trained with D2, D3, D4, and D5 trials. D, Confusion matrix for a classifier trained with D2 + D5 and D3 + D4, which indicated an advantage for the addition of edge digits’ responses.
- Figure 8.
Voltage-sensitive dye imaging of cortical responses. A, Example of cortical activation dynamics evoked by digit stimulation. B, Example of mean DF/F over time for one pixel. C, Latency map with cold (blue) colors representing the shortest latencies. D, Latency distribution of all pixels of the image of activation shown in C. E, First activated pixels (FAP) for a single digit stimulation. F, Example of FAP contours for all single digits, showing a somatotopic organization of the distal phalange of finger representations. G, Percentage of overlap between FAP elicited by single-digit stimulations, for one animal. H, Mean percentage of FAP overlaps for the eight animals included in the study. I, Mean FAP areas for single stimulations, costimulations, and sums of single stimulations (paired Wilcoxon test, *p < 0.05).
- Figure 9.
Comparison of the first activated pixel (FAP) areas in response to costimulation and single stimulation. A, Left, example of FAP contours for D2D3 (gray), D2 (yellow), and D3 (pink) stimulations. Center, mean percentage of overlap between D2D3 and D2 and D3 FAP areas. Right, scatter plot of each animal overlap percentages between D2D3 and D2 or D3 versus overlaps between D2 and D3 or D3 and D2. B, Left, examples of FAP contours for D2D4 (gray), D2 (yellow), and D4 (green) and for D2D4 (gray) and D3 (pink) stimulations, respectively, in the top and the bottom. Center, mean percentage of overlap between D2D4 and D2, D3 and D4 FAP areas. Right, scatter plot of individual FAP overlap percentages between D2D4 and edge digits (D2 and D4) versus D2D4 and D3 (top); between D2D4 and edge digits versus D3 and edge digits (bottom). C, Left, examples of FAP contours for D2D5 (gray), D2 (yellow), and D5 (blue) and for D2D5 (gray) versus D3 (pink) and D4 (green) stimulations, respectively, in the top and the bottom. Center, mean percentage of overlap between D2D5 and D2, D3, D4, and D5 FAP areas. Right, scatter plot of individual overlaps percentages, between D2D5 and edge digits (D2 and D5) versus edge digits and center digits (D3 and D4; overlaps between adjacent digits; top); between center digits and edge digits versus D2D5 and center digits (bottom).
- Figure 10.
Location of costimulation-evoked first activated pixel (FAP) area along the somatotopy axis. A, Example of distributions of FAP areas projected onto the somatotopy axis. The axis was determined by performing a linear regression onto the single-digit FAP areas centroids. B, Single-digit and codigit stimulation–evoked centered activations. Points represent the mean of FAP distributions for the population, with bars indicating the mean and SD.
In this issue
