From Brain Radio to Mental Block: A History
(All this and more is available for perusal at http://mentalblock.net)
(This Flickr feed has the A-Z photo documentation for the process)
The project began with an interest in using EEG equipment to control media. There are two consumer level EEG hookups on the market, the MindFlex and the Star Wars Force Trainer. Both use chips from NeuroSky. The MindFlex only let us listen to LED states, whereas the Force Trainer gave us direct serial data. Wonderful. Here’s us playing with the Mindflex, and then hacking the Force Trainer.
After some playing around we settled on listening to people’s brainwaves made audible, and broadcasting your own. It was called Brain Radio. In the original scenario, a wearer would put on a brain-reading headset to start broadcasting brain state over the air. The wearer could tune into sounds generated by other headset-wearers via a dial. Brain Radio sought to create a kind of perceived telekinesis, in which your thoughts — alpha and beta waves reduced to “attention” and “meditation” by the EEG headsets we hacked — supply the source for sounds broadcast to any other headset-wearers who chose to tune in. The sound is designed to reflect and represent on the broadcaster’s state of mind. However, who, exactly, you’ve tuned into is unknown. Instead, it’s left up to your own inference and interpretation of other’s physical behavior relative to the sound.
We started off by building three of the headsets, but the system was designed to scale to any number of users, growing the number of potential connections quadratically. To gain access to the brainwaves we hacked Star Wars Force Trainers produced by Uncle Milton Toys. The headset of the toy contained a chip made by NeuroSky which takes in transcutaneous electrical signals from the brain and turns it into serial data. There was another toy we tried, the MindFlex, but there was no way to hack the serial. The Force Trainer transmits the serial data over bluetooth to a base station, which when used in the “proper” way maps the values to the speed of a fan and thus the height of a ball. Our online research only took us so far. We found a diagram outlining the header pins on the bluetooth receiver within the Force Trainer base station, but could not coax serial values. Emails to the hackers we fruitless. To get around this issue we tapped the Tx, Rx, power and ground pins on the NeuroSky chip itself, with positive results– we got a steady stream of 32 data packets from the Tx pin, and the Rx pin proved to be useless. (It was about here that we must have fried the NeuroSky chip because for a while we thought the sensors were dubious, giving us poor connectivity denying us the packets we had fought for). Fortunately, NeuroSky wants people to do strange and wonderful things with their chips and published the content of the packets with good documentation. The data from the Tx pin went into the Arduino which Once we wrote the parser in Processing and made a graph to illustrate the aforementioned brain states of “attention” and “meditation,” the next hurdle was sending the data wirelessly to the computer. We used Xbee Series 2 radios to send the data directly into a digital pin on the Arduino, which processed the data, and spat it back out through its own TX pin to the Xbee, and then wirelessly to the computer. The idea was that once the central computer received steady streams of brain state data from all three headsets, the data could be used to control a sound patch in Pure Data, which could then be broadcast over walkie-talkies back to the wearers. The Pure Data patch was created to mirror binaural beats, which are said to be able to influence the level of meditation or attention of the listener. The wearers would hear the binaural beats through bone-conducting headphones, a way of feeling sound more than hearing it– frequencies produced by the body would be funneled back into the body. Through this multi-channel closed feedback loop, it is conceivable that multiple wearers could passively induce change in other’s mental states, or acquire the state of another.
But that’s not exactly how it went. Hacking the walkie-talkies was harder than we expected, and getting the values into Pure Data and back out again worked inconsistently. Going from the bread board to perf-board took longer than expected and was buggy. We also suffered from the fabled “feature creep”. We tried adding a seven segment LED display that would flash floating numbers in front of the wearer’s face like a HUD. The numbers reflected off the Lexan face shield beautifully, and functioned to show the wearer which brain he or she was tuned to. We added a mini vibrating motor to the channel changing dial so that when a wearer turned the dial it would vibrate beneath their touch at each channel. Everything worked on the breadboard, but did not work on the perf-board.
The next step was to dismantle the entire idea and start from scratch. What did we want to control, and how did we want to shape a wearer’s mediated experience of their own brain, or someone else’s? We went back to an idea we had mulled over but discarded in favor of audio: visual obfuscation versus clarification. Enter: Mental Block.
Tags: arduino, EEG, MentalBlock, PhysComp