‘Version one’ of the Staalhemel project opened on 1 October 2009 in arts centre STUK in Leuven (Belgium). This version was provided with an EEG brainwave analysis based on algorithms written by Wolfgang Eberle from IMEC. Algorithms developed for Staalhemel aim at detecting elevated levels of mental relaxation or distraction (alpha waves) and those of the inverse state: the absence of such alpha waves, which points to increased concentration or focusing levels.
These algorithms filter noise and interpret brainwave events. The brainwave data streams from all separate channels are also broadcast to the computer network that controls the installation, but are not used individually. More important is the stream of detected events which is broadcast on a different channel and controls the dynamics of the sound installation. These events are generated by the IMEC software algorithms and include: alpha waves dominant in either left, right or both hemispheres; absence of alpha waves in either hemisphere, and eye blinking and jaw movement artefacts. Alpha states are derived by averaging calculations on the alpha values from two or more available channels.
Although the prototype EEG headset contains 8 measurement electrodes and 2 reference electrodes, the EEG analysis of “version one” does not include access to alpha-beta interpretation per channel. That means that a comparison between states of mental activity per brain section (prefrontal, frontal, top and back of head) is not possible in this version. This limitation is imposed because we wanted to minimize the amount of preparation time for each visitor. One of the main reasons for the collaboration with IMEC in function of the artistic purpose was to develop an easy-to-use prototype that can be switched from user to user as fast, easy and reliable as possible. This to ensure the installation can be experienced by as many visitors as possible. Individual analysis per EEG channel would require a calibration of personal ranges and levels before a correct interpretation of alpha and beta waves would be possible.
In ‘version one’ a certain distribution of frequencies over the steel matrix is activated whenever the overall mindstate changes from or to a predominance of alpha waves. The total absence of alpha waves is interpreted as beta activity. This deduced beta activity is mapped to an increase in resonance and in the number of steel elements being struck. Elevated presence of alpha waves bring the sound to a slower pace with most steel plates dampened, resulting in a raindrop-like pattern. Within each mental state the individual pin frequency intervals will shift one segment to the left if detected values drop down, and shift to the plate on the right side when the values are rising. When a considerate amount of eye blinking is detected a frantic pattern will cut through whichever state is dominant at that moment. This feature can be switched on or off.
Each different state is subdivided in a number of regions, from ‘active’ over ‘moderate’ to ‘strong’. Whenever the event value crosses a threshold, the frequencies of all active pins are scaled to a broader or smaller range of milliseconds depending if the event value is decreasing or increasing. That means that if someone’s focus level reaches a higher degree, the pins will strike at a rate that is a multiple of their previous frequencies while keeping the distribution of rhythm intact. With 3 divisions of intensity per wave detection there is a simulation of a linear scale ranging – on a cognitive level – from extremely relaxed to extremely focused.