Intracranial recordings from the human brain are rare due to the invasiveness of implanting electrodes in epilepsy patients suffering from drug-resistant seizures. While patients await seizures to localize their onset zone in the brain, implanted and connected to electrophysiological recording systems in the hospital's epilepsy monitoring unit, there is a unique opportunity to perform computer-based tasks with them to study memory and cognitive functions.

This invasive seizure monitoring and localization is typically performed with several standard clinical macro-contact electrodes implanted around the areas suspected to generate seizures. The same electrode leads can additionally be equipped with micro-contacts to record brain activities at highest resolution of action potentials fired by individual neurons or high frequency oscillations generated by neuronal assemblies within a single cortical column.There are no extra electrode implantations - just more contacts along the same electrode leads that are implanted for clinical purposes of localizing and treating medically refractory seizures. As the patients engage in performing the tasks, low and high frequency oscillations, a.k.a the brain waves, as well as neuronal action potential or so-called ‘spikes’ are recorded from distinct electrode contact locations when words or objects are encoded and then recalled in the mind.

In our previous dataset release (Cimbalnik et al. 2022) we published only the standard macro-contact recordings from a cohort of Mayo Clinic patients performing a battery of cognitive tasks with concurrent gaze-tracking and pupillometry. Now, we have extended this original dataset to two new cohorts of patients from the Czech Republic and Poland, performing the same battery of tasks in Czech, Slovak and Polish languages. These new cohorts include micro-contact recordings from at least ten contacts in any one patient sampled at 32 kHz to enable detection and analysis of single neuron spikes. We have also complemented the dataset with automated tools for importing, viewing and preprocessing the data stored in a standard BIDS (Brain Imaging Dataset Structure) format to meet the international requirements for sharing and reproducibility.

Cracking the code of the human mind…

One of the greatest challenges in neuroscience is to understand how abstract processes of the mind are generated by electrophysiological activities of the brain. Our dataset provides a range of possible studies addressing how word concepts are encoded, maintained and recalled in a range of cognitive paradigms using the same pool of common nouns in English, Czech, Slovak and Polish languages. These universal words like “table”, “car” or “hill” presented in a visual or auditory form as written names or generated speech, respectively, should induce comparable concepts in the human mind across various cultural backgrounds. The paradigms assess how these are encoded as lists or word pairs, and recalled freely or with a cue of the paired word. There are also additional gaze-tracking tasks for smooth pursuit and anti-saccades to control for responses to non-verbal stimuli. All of the tasks are performed together in a battery, where the same words are encoded and recalled repeatedly in various tasks of the same session or on subsequent days.

For the purpose of dataset validation, we have included preliminary analysis of the electrophysiological and pupil size responses to confirm accurate synchronization with the behavioral events in the tasks. Consistent patterns of pupil constriction during word encoding and pupil dilation during recall on successful memory trials, or bursts of high frequency oscillations preceding vocalization of the recalled word, are two examples of the validation analyses demonstrated in the paper to inspire possible studies. The latter example is perhaps the most comprehensive because the bursts of high frequency oscillations are coincidentally detected across multiple macro- and micro contacts a few hundred milliseconds before the recalled words are started to be spoken, suggesting a link with emergence of implicit or explicit concepts heralding the behavioral response (Prathapagiri et al. 2025). These global brain electrophysiological discharges are associated with consistent pupil size dynamics. Hence, cognitive processes occurring in the mind can be investigated on all levels, ranging from the systemic and general pupil responses to a word through to micro-scale spiking of individual neurons reflecting specific concepts.

Dataset contents from inside and outside of the skull

Our dataset comprises not only electrophysiological, gaze-tracking and pupillometric signals but also audio recordings of the behavioral responses (recalled words), annotations of the task events (e.g. word presentations), and 3D brain coordinates and anatomical labels of electrode contact localization. All these data are organized separately for each patient and structured into particular sessions and runs of the task battery, according to the BIDS format. We have also provided links to the open-access library of tools from the Brain & Mind Electrophysiology lab. The datasets themselves are available on EBRAINS - the official portal of the EU’s flagship Human Brain Project.

Key advantages:

• International, multi-center data from English, Czech, Slovak and Polish speaking patients performing the same battery of cognitive tasks
• Multiple tasks using the same word stimuli encoded and recalled within one experimental session and repeated on subsequent days
• Robust and reproducible organization of the data into BIDS format
• Complete metadata including annotations of every task event and behavioral response, anatomical labels and electrode localizations
• Open-access libraries of freely available tools for data analysis
• Examples of preliminary analyses and published studies

From tracking engrams to Brain-Computer Interfaces

We invite anyone interested in the analysis and modelling of these signals for research purposes. Our goal is to encourage diverse neuroscience and biomedical engineering studies. The datasets are suitable for a wide range of applications, ranging from (1) tracking engrams of the remembered words over subsequent task phases, paradigms and sessions, (2) ‘mind-reading’ of specific word concepts before they are recalled, (3) closed-loop brain stimulation to restore memory performance, all the way to (4) developing brain-computer interface for direct extra-sensual communication.

1. Cimbalnik, J. et al. (2022) Intracranial electrophysiological recordings from the human brain during memory tasks with pupillometry. Scientific Data 9(1):6
2. Prathapagiri, S. et al (2025) Coincident bursts of high frequency oscillations across the human cortex coordinate large-scale memory processing. bioRXiv