Fig 1.
Main schemes of thalamocortical interaction.
A) Functionally distinct parts of the thalamus including the dorsal thalamus with its anterior, lateral, and medial regions, the reticular nucleus as the main component of the ventral thalamus in green, and the intralaminar nuclei in grey. B) First-order relay scheme in which the subcortical/peripheral afferences are relayed to the cortex generally to layers 3 and 4. C) High-order relay scheme in which afferences from a cortical region are relayed to another. Coronal slices were acquired from the Big Brain project [17] using ebrains platform.
Fig 2.
Simulations for two levels of thalamic noise (low noise in A, B, C; high noise in E, F, G), three implementations of thalamocortical SC (in colours), and exploring the parameter space for coupling factor (g) divided into prebifurcation (g < 7 shadowed regions in B and F) and postbifurcation (g > 7). B, F shows group averaged rPLV(α) and KSD(α) metrics. In the margins, boxplots show maximum values for those metrics per subject in prebifurcation (A, E) and postbifurcation (C, G). D shows the averaged bifurcation diagrams consisting of the maximum and minimum voltages per simulation for the cortex (thick line) and the thalamus (dashed line).
Table 1.
Network features for the thalamus and the cerebellum.
Fig 3.
Cortico-cerebellar control experiment.
Comparing the simulations implementing high noise in thalamus and in cerebellum with three versions of their structure. Left, boxplot showing maximum rPLV(α) values per subject in the prebifurcation space for both experiments. Right, lineplots conveying averaged values of rPLV(α) per value of coupling factor, and SC version. Shadowed areas covering the prebifurcation range.
Fig 4.
Simulation samples for one subject with parceled thalamus.
In columns, simulation samples combining low/high thalamic noise with coupling values from pre/post- bifurcation. We selected g = 4 to simulate prebifurcation and g = 36 to simulate postbifurcation. First two rows show samples from the simulated signals (A), and their corresponding spectra (B). Last two rows showing PLV(α) (C, D) and dFC(α) (E, F). C and E showing empirical references for PLV(α) and dFC(α), respectively.
Fig 5.
Parameter space explorations for pth with high thalamic noise.
Sets of three simulations averaged for subject one and parceled thalamus. Heatmaps showing different metrics from the same each simulation: the empirical-simulated correlation of PLV in alpha band (rPLV(α)), the simulated mean and std of the PLV(α) values, the frequency peak of the nodes’ averaged spectra (FFT peak), the signal to noise ratio in thalamic nodes computed as the amplitude of simulated signals divided by the standard deviation of the Gaussian noise used for the thalamus (SNR(th)), and the bifurcation of cortical signals using the averaged maximum-minimum signals’ voltage.
Fig 6.
Parameter space explorations over ηth to balance SNR in the thalamus.
Sets of three simulations averaged for subject one and parceled thalamus. Heatmaps showing different metrics from the same each simulation: the empirical-simulated correlation of PLV in alpha band (rPLV(α)), the simulated mean and std of the PLV(α) values, the frequency peak of the nodes’ averaged spectra (FFT peak), the signal to noise ratio in thalamic nodes computed as the amplitude of simulated signals divided by the standard deviation of the Gaussian noise used for the thalamus (SNR(th)), and the bifurcation of cortical signals using the averaged maximum-minimum signals’ voltage. Vertical dashed lines define the optimal SNR range.
Fig 7.
Complementary simulation samples.
Simulation samples derived from the parameter explorations with parceled thalamus and cortical nodes in prebifurcation (g = 2). In columns, simulation samples in which the thalamus drives cortical activity with different levels of noise (ηth = [0.022, 0.09, 0.5]) and at two different points of the thalamic bifurcation determined by pth = [0.12, 0.15], the former corresponds to the slow limit cycle of JR. First two rows show samples from the simulated signals (A), and their corresponding spectra (B). Last two rows showing PLV(α) (C, D) and dFC(α) (E, F). C and E showing empirical references for PLV(α) and dFC(α), respectively.
Fig 8.
First line shows the SC versions for the thalamocortical experiment: woTh, Th, pTh. The second line shows the SC versions for the cortico-cerebellar control experiment. Dashed lines representing driver connections.
Fig 9.
JR model of a cortical column.
a) Block diagram depicting JR operators and modules where each color is associated with a different neural population: pyramidal (cyan), excitatory interneurons (green) and inhibitory interneuron (red). b) Histological contextualization of the cortical layers. Modified from [62, 90].
Table 2.
JR parameters used in simulations.
Fig 10.
The bifurcation separates two states.
In the center, a bifurcation diagram shows the minimum and maximum voltage for each value of the bifurcation parameter. At the critical point (dashed line), the bifurcation occurs and separates two states of the system: a) damped oscillator, whose activity tends to decay to a fixed point; and b) limit cycle oscillator, whose activity is a self-sustained oscillation.