Dataset for: Neural basis of compositional control

This dataset contains the complete preprocessed behavioral and neural recordings collected during the prey-pursuit task described in Chericoni et al. (2026).The dataset includes two cohorts:Intracranial cohort: 19 participants with drug-resistant epilepsy who performed the prey-pursuit task while undergoing intracranial monitoring for epilepsy surgery.Healthy cohort: 50 healthy participants who performed the same prey-pursuit task.Behavioral data include all task variables used in the study, including positions, velocities, accelerations, reward values, and agent identities. Behavioral signals were sampled at 60 Hz.Neural data consist of preprocessed single-neuron recordings from all isolated neurons included in the study. To facilitate alignment with behavior, neural activity was resampled to 60 Hz and synchronized to the behavioral time base. The dataset includes all recorded neurons and is not restricted to neurons selected for specific analyses.Participant identifiers have been anonymized. Raw clinical recordings are not included; however, all preprocessed behavioral and neural data necessary to reproduce the analyses reported in the manuscript are provided.Data are organized as follow and further instructions can be found in the uploaded readme.docx- events_info: table with behavioral infos. Each row is a trial; each column represents an event (typo) during the trial, times are in ms (Fs = 60Hz). Chase start: the player does the first move; chase end: the player catches the prey. Reward_val = 0 means the player did not catch any prey.- neuronData: table with time binned behavioral data and associated spiking data (if available), sampled at 60 Hz. Each row is a trial, while the columns are timeseries arrays organized as follow: time, task/session time in ms, each row is a 16ms bin. X and Y: coordinates of self-position on the screen. X and Y prey: coordinates of prey position on the screen; if size(Xprey, 2) = 2, the first column represents X of the prey that is defined as prey1 in events_info (same for Y). Spikes: spikes matrix where each row is a 16ms bin (again Fs = 60Hz) and each column a neuron, the matrix is filled with the number of spikes happening in each time bin. Brain_region contains brain regions labels for each neuron. NOTE: data from healthy participants also include a neuronData table, this is just a naming convention and in the case of healthy subjects neuron data does not contain any spikes matrix, but only XY for self and prey across time.