Gamer in the scanner: imitation of human video gameplay and fMRI brain activity using artificial neural networks
Bibliographic record
Abstract
Artificial neural networks can successfully play video games, yet these AI agents have difficulty adapting to changes in the game environment, or transfer knowledge across different games. As human players can efficiently transfer skills across environments, the Courtois NeuroMod team is working to align the representations of artificial neural networks with human players [1]. We first designed and validated a fully MRI-compatible video game controller [2]. The data collected for this project are part of an extremely deep individual fMRI sample currently featuring up to 140 hours of fMRI per subject (N=6), made available for the community as part of the Courtois NeuroMod data bank (https://cneuromod.ca). We successfully trained artificial agents to imitate the actions of humans playing the game “Shinobi III: revenge of the ninja master” and found that the internal representations of the agents could be used to effectively predict individual brain activity measured with functional magnetic resonance imaging [3]. This work could open new avenues to train robust AI video game characters, and gain new insights in brain representations for active and complex stimuli. 1. Bellec, P., and Boyle, J. Bridging the gap between perception and action: the case for neuroimaging, AI and video games. Psyarxiv 2019. Link to Paper 2. Harel*, Y., Cyr*, A., Boyle, J., Pinsard, B., Jerbi, K. and Bellec, P. Gamer in the scanner: open design and validation of a video game controller for MRI and MEG. Psyarxiv 2022. Link to Paper * joint first authorship 3. Kemtur, A., Paugam, F., Pinsard, B., Sainath, P., Harel, Y., Le clei, M., Boyle, J., Jerbi, K. and Bellec, P. AI-based modeling of brain and behavior: Combining neuroimaging, imitation learning and video games. Proceedings of Computational Cognitive Neuroscience Conference 2022. Link to Paper
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".