**README** —————————— Pfeffer et al. (2021) Circuit mechanisms for the chemical modulation of cortex-wide network interactions and behavioral variability. Science Advances. —————————— **RE-CREATING MAIN FIGURES** —————————— In order to re-create the main figures of the paper, all information you need can be found in the designated folders. Each folder contains a MATLAB file (e.g. fig1.m) which should take care of loading the needed data as well as the code (both of which can be fold in the designated subfolders). The functional connectivity matrices (for all subjects, drug conditions, behavioral contexts and frequencies) can be found in ‘fig1/data/fc_cortex.mat’. Functional connectivity was defined as the (orthogonalized) power envelope correlation between all pairs of nodes, resulting in a 400x400x28x3x2x17 matrix (averaged across two recording blocks). The format of the file is: n_vox x n_vox x n_subj x n_drugs x n_contexts x n_freqs, with n_drugs being placebo (1), atomoxetine (2) or donepezil (3) and n_contexts rest (1) or task (2). n_freqs were the 17 logarithmically spaced center frequencies, ranging from 4 Hz to 64 hz (2.^[2:0.25:6]). Due to the large number of output files (1M+ files), results from the whole-cortex model simulations have only been uploaded in a condensed version. The code in order to simulate all models can be found in the repository. If something is not working right or you need help/data, please drop me an email: thms.pfffr@gmail.com —————————— **DATA AVAILABILITY** —————————— The following data is available in the repository: - All FC matrices (see above) (in folder: fig1) - Behavioral data from bistability task (in folder: fig3) - Behavioral data from foraging task (in folder: fig4) - Simulated data can be created from the model code that is available —————————— **CODE AVAILABILITY** —————————— The following code is available in the repository (see dedicated folder 'code'). Note that almost all code is written such that it can be processed in a distributed manner (w/o using MATLAB's tools for distributed computing) on the HPC cluster of the Department of Neurophysiology. No effort has been made to optimize the code to run on other systems. **WC_model:** Contains code for Wilson-Cowan model simulations - pmod_wc_analytical.m (2-node model) - pmod_wc_wholebrain.m (whole cortex model) - pmod_wc_wholebrain_detosc.m (find oscillatory regime; in essence the same as *wholebrain.m, but without noise) - pmod_wc_wholebrain_task.m (heterogeneous task related inputs) - pmod_wc_wholebrain_detosc_task.m (find oscillatory regime, heterogeneous task inputs) **LIF_model(s):** Contains code for leaky integrate and fire simulations - run_model.py (run LIF model with stimulation) - integration_circuit_mod.py (needed for run_model.py) - plotting.m (plots the results from run_model.py) - run_decision_circuit.py (run "bistability" circuit) - integration_circuit.py (needed for run_decision_circuit.py) - plot_decision_network.m (plot results) - All code requires Python 2.7 and Brian 1.4.4 (and numpy, scipy etc.) - **MEG analysis:** Code to compute FC statistics - pupmod_compute_altered_correlations.m (computes fraction of altered correlations from FC matrices) - pupmod_src_powcorr_permtest.m (run and save permutation distribution) - pupmod_src_powcorr_getstatistics.m (obtain p-values from permutation distribution)