Yes, Perri Karyal really controlled Elden Ring with an EEG-based interface—but not by reading arbitrary thoughts. Her custom system classified repeatable brain-activity patterns linked to imagined movements, combined those signals with head-mounted motion sensing, and converted the results into ordinary virtual controller inputs.
That makes it a real, non-invasive brain-computer interface (BCI), but not plug-and-play “mind control.” It was a personalized project that required extensive training, calibration, software integration, and persistence.
As an Amazon Associate I earn from qualifying purchases.
The short answer: what Karyal actually built
Perri Karyal is a Twitch streamer, variety broadcaster, and psychology postgraduate who had previous experience working with EEG. She became interested in combining psychology, neuroscience, and gaming content using consumer hardware that could be used outside a hospital or university laboratory.
Free tools Windows power users keep installed
One-click scans. No signup required.
As described by IEEE Spectrum, her setup used an Emotiv EEG headset and accompanying software to recognize trained patterns in her brain activity. Those recognized patterns were mapped to game commands. Headset-mounted accelerometers supplied additional movement and camera controls.
#1 Best Overall
From Elden Ring’s perspective, the system did not look like a brain scanner. It received simulated keyboard, mouse, or controller inputs. The EEG software sat between the player and the game, turning selected signal classifications into button presses or directional commands.
How the EEG-to-controller pipeline worked
The system can be understood as a chain of six stages:
- Record: Scalp electrodes measure small voltage changes associated with electrical activity in the brain.
- Train: Karyal repeatedly performs or imagines selected mental tasks while the software collects labeled examples.
- Classify: A model identifies signal patterns associated with those tasks.
- Map: A recognized pattern is assigned to a selected game command, such as an attack.
- Combine: Motion sensors detect head movement for other functions, including character movement and camera-related controls.
- Input: Software sends virtual controller or keyboard events to the game.
A simplified version is:
Imagine a movement → EEG pattern → classifier → virtual input → head-motion input → game action
The exact classifier, thresholds, programming language, and input-emulation tools in Karyal’s private implementation are not established by the published profile, so they should not be presented as known technical details.
What EEG can—and cannot—read
Electroencephalography, or EEG, records electrical signals through electrodes placed on the scalp. Consumer headsets generally have fewer channels and noisier measurements than clinical or laboratory systems. Signals can be affected by electrode contact, blinking, facial and neck muscles, head movement, fatigue, and electrical interference.
EEG does not display the literal content of a thought. It does not understand an internal sentence such as “run forward,” infer a complete game plan, or decode any private idea on demand. Instead, a trained system attempts to distinguish statistical patterns in measured signals.
Karyal’s reported method used motor imagery: mentally rehearsing a physical movement without actually performing it. For one attack-related command, she reportedly imagined pushing a heavy object forward. The computer was not understanding the meaning of “attack”; it was recognizing a signal pattern that she had learned to reproduce and that the software had learned to associate with that command.
The European Data Protection Supervisor’s explanation makes the same essential distinction: imagined physical actions can produce patterns that trained software classifies and translates into commands.
Was she really playing hands-free?
According to the IEEE Spectrum profile, yes—with important qualifications. Earlier versions supported only a small number of inputs. Later versions recognized enough distinct EEG patterns to replace the required game inputs, while accelerometers in the headset handled movement and camera controls.
That qualifies as a hands-free controller for the demonstrated setup. It does not mean EEG alone controlled every action, that every stream used an identical configuration, or that Karyal could instantly control any game through thought.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Elden Ring was a powerful demonstration because its combat demands attacks, dodges, movement, camera control, and contextual actions in rapid sequences. Combining a limited number of EEG commands with head motion creates a larger control vocabulary. The game’s difficulty also makes delays, false positives, fatigue, and awkward command selection visible.
Rank #3
The published reporting provides a qualitative demonstration, not a controlled benchmark. It does not establish a conventional-controller-equivalent speed or accuracy percentage.
The training burden was the real story
Karyal said she trained the system 429 times to recognize one imagined forward-pushing activity. She also said she had spent more hours training the EEG system than playing the game. At the time of the IEEE Spectrum profile, she reported more than 250 hours in Elden Ring.
Those numbers describe her project, not a universal requirement. Training time varies with the user, headset, electrode contact, number of commands, classifier, signal quality, and game. Every additional command makes the classification problem harder and creates more opportunities for accidental inputs.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsCalibration is also personal. A model trained on one player’s signals and mental strategy generally cannot simply be transferred to another player. Even the same person may need to recalibrate when electrode placement changes, attention declines, or fatigue alters the signal.
Could an ordinary gamer build one?
A technically experienced reader could reproduce the general concept, but not as a weekend plug-in accessory. A realistic build path would be:
- Choose an EEG headset with real-time developer access and confirm that its operating system and SDK are supported.
- Fit the headset carefully and verify electrode contact, wireless connectivity, and data quality.
- Start with two or three commands plus a neutral or rest state.
- Select repeatable mental tasks, such as imagining distinct physical movements.
- Record many labeled examples of each task and the neutral state.
- Train or configure the headset’s mental-command classifier.
- Measure behavior outside the game before mapping commands to combat actions.
- Map only high-confidence detections to virtual keyboard or controller events.
- Add cooldowns, a neutral state, logging, and an emergency disable key.
- Combine EEG with head tracking, eye tracking, switches, or voice rather than trying to make EEG handle every input.
- Test first in menus or a low-stakes game.
- Recalibrate whenever signal quality or performance changes.
The exact menus, API calls, and software versions depend on the headset and implementation. Karyal’s full code and command mapping are not supplied in the cited report, so a responsible guide cannot promise a copy-and-paste recreation.
Rank #4
- Used Book in Good Condition
What can go wrong?
| Problem | Likely response |
|---|---|
| No commands are detected | Check electrode moisture and contact, headset fit, receiver or Bluetooth connection, and software permissions. |
| Too many accidental inputs | Reduce sensitivity, enlarge the neutral state, add cooldowns, and simplify the command set. |
| Performance deteriorates during a session | Stop and recalibrate. Fatigue and changing electrode contact can alter the signal. |
| Head movement causes unwanted actions | Separate motion and EEG mappings, reduce motion sensitivity, or use another input for movement. |
| The game ignores virtual inputs | Check operating-system permissions, input-emulation focus, game-window focus, and the game’s selected input mode. |
| Discomfort, skin irritation, headache, or unusual symptoms | Stop using the device and follow the manufacturer’s safety guidance. Consumer EEG is not a substitute for clinical evaluation. |
Current hardware: what the commercial options offer
Karyal’s original 2023 setup should not automatically be equated with every specification of the current product line. Emotiv’s present U.S. product page promotes the EPOC X in connection with the gaming demonstration and lists 14 EEG channels, saline-based felt sensors, built-in nine-axis motion sensing, wireless connectivity, configurable 128 or 256 samples-per-second sampling, and battery life of up to nine hours with a USB receiver or up to six hours over Bluetooth Low Energy.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →The same ecosystem advertises Cortex API/SDK access and integrations for Python, MATLAB, Unity, and other development workflows. Hardware pricing and software access can vary by region, storefront, promotion, and subscription tier. The cited pages showed an EPOC X price signal of $1,199 on one U.S. page and $999 on another storefront, so neither should be treated as a universal price.
- Emotiv EPOC X: The closest commercial match for a multi-channel, motion-assisted experiment. The listed U.S. price was $1,199 at the cited page, but buyers should verify current regional pricing and software terms.
- Emotiv Insight: A lower-cost five-channel option listed at $499 on the cited catalog. Fewer channels make it a different platform, not an equivalent EPOC X replacement.
- Emotiv MN8: EEG headphones with two channels, listed at $399 on the cited catalog. Its form factor is approachable, but it is not a direct substitute for a multi-command research-style setup.
- OpenBCI Ultracortex: An open, maker-oriented alternative whose Complete Headset listing showed selectable eight- or 16-channel configurations at $2,999. It offers more openness but also more integration responsibility and a higher listed price.
Before buying, check developer access, supported operating systems, channel placement, motion sensors, electrode preparation, battery and connectivity, software fees, comfort, support and returns, and whether raw EEG is processed locally or sent to a vendor service.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.EEG is not automatically the best accessibility controller
EEG’s potential advantage is hands-free control when other voluntary movements are unavailable or exhausting. But it is not automatically faster, cheaper, or more reliable than established accessibility interfaces.
| Interface | Typical strength | Trade-off |
|---|---|---|
| Adaptive controller | Direct, configurable game input | May still require usable hand, arm, or switch access |
| Eye tracking | Fast selection and cursor control for some users | Requires reliable eye control and careful setup |
| Head tracking | Useful for camera or pointer movement | Can cause fatigue and may not supply enough discrete inputs alone |
| Switch scanning | Works with very limited voluntary movement | Sequential selection can be slow |
| Sip-and-puff | Practical discrete control for some users | Requires suitable breath control and equipment |
| Voice control | Natural for menu and command input | Noise, privacy, and recognition issues can matter |
| EEG | Potentially hands-free and silent | Training, calibration, signal noise, false positives, and low command bandwidth |
For many players, a switch, eye tracker, head tracker, adaptive controller, or combination system will be easier to maintain and more predictable than EEG. The right choice depends on the person’s available movement, fatigue, environment, and preferred games—not on how futuristic the interface looks.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why this is not a clinical brain-computer interface
Karyal’s system is wearable and non-invasive: it requires no surgery and uses electrical activity measured at the scalp. That makes it more accessible than invasive BCIs, but non-invasive EEG generally offers lower signal resolution and lower command bandwidth than systems using implanted sensors.
Best Value
The project demonstrates an experimental consumer interface, not a clinically validated assistive device. It should not be used to make claims about restoring movement, treating a condition, or matching medical BCI performance.
Neurodata and privacy
EEG data can be sensitive even when it cannot decode arbitrary thoughts. As the European Data Protection Supervisor discusses, an EEG control system may involve collection, preprocessing, classification, and storage or transmission of neurodata.
Before using a commercial headset, ask:
- Does raw EEG leave the computer?
- Is cloud processing required?
- Who stores the signal recordings and trained mental-command profiles?
- Can those profiles and recordings be deleted?
- Can gaming data be reused for analytics, advertising, or profiling?
- What do the vendor’s current terms, privacy policy, and return policy say?
A gaming experiment is still a data-collection decision. Buyers should review those policies before creating a long-term personal signal history.
Recommended Free Tools
The bottom line on “mind-controlled” Elden Ring
Perri Karyal’s demonstration was real and technically impressive. It showed that a consumer EEG headset can be trained to recognize a small set of repeatable mental-task patterns and turn them into game commands. It also showed why hybrid control matters: head motion supplied part of the complete hands-free interface.
What it did not show was unrestricted thought-reading or a universal controller that anyone can buy and use immediately. The result was a custom, personalized BCI built around extensive training, noisy signals, calibration, and software work. For accessibility, conventional adaptive inputs may still be the better solution. For makers and researchers, however, the project is a compelling example of how EEG can become one component in a practical—if demanding—game-control system.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
