The self-experiments took place at home (first seven nights) and in the sleep laboratory of the NeuroBioPsychology department at the University of Osnabrück (last two nights).

The sleep communication was conducted according to the Sleep Communication Framework with the Sleeator 2 as its implementation.

The hardware device used for sleep recording and eye movement recording during the first seven nights was the Zeo (like in the early self-experiments). During the last two nights, when sleeping in the sleep laboratory, polysomnographic (PSG) recordings were conducted additionally to the Zeo recordings using a NeuroScan amplifier and software system. This system recorded 19 channel EEG (according to the 10-20-EEG system as depicted in Appendix D), horizontal and vertical EOG and chin EMG. Impedance was kept below 5 kΩ. Data was sampled at 500 Hz.

REM sleep was detected by two systems: by the Zeo’s intern sleep stage classification on the one hand, and on the other hand simultaneously by a threshold REM detector optimized for the sleep of the subject. This self-programmed classifier took three frequency bands of the EEG of one second (9-14 Hz, 15-30 Hz, and 41-45 Hz) into consideration and if the sum of the frequency values in each band was below (9-14 Hz, 41-45 Hz) or above (15-30 Hz) a given threshold, it classified the second accordingly as REM or NREM. The frequency bands and thresholds were determined by analyzing the course of different frequencies for the subject over several nights, and identifying different patterns in the frequency bands of the subject’s sleep. One of these patterns was assumed to correspond to REM sleep (basically because the frequency values corresponded to the AASM manual’s rules (Iber & American Academy of Sleep Medicine, 2007) and the Zeo’s intern classification showed REM sleep, too). Both REM sleep detection systems were used simultaneously, i.e. if at least one of the two systems detected REM, the Sleeator 2 assumed that the subject was in REM sleep.

Stimuli were played if out of the last 300 seconds at least 50% were classified by at least one system as REM sleep.

Computer generated sinus tones with a frequency of 1000 Hz played by Logitech S-120 speakers were chosen as stimuli. A simple random math problem (two random one-digit numbers had randomly to be added or subtracted) was encoded into the sinus tones using Morse code. Examples can be found in Figure and as an *.mp3 file on the DVD. The stimuli were played using the auto increment feature of Sleeator 2 for the stimulus intensity, increasing the loudness of the tones form 10% to 100%. 100% were set to match approximately twice the loudness of surrounding noise (i.e. at home noise of the computer and the street outside, and in the sleep laboratory noise of the computer, the ventilation system and the polysomnographic recording system).

The eye movement chosen to signal lucidity was “look five times to the left (and in between back to center)”. This signal is different to the left-right-left-right signal normally used in lucid dreaming research but was chosen in order to make it easier to detect in the Zeo EEG.

Example for an auditory stimulus (1000 Hz sinus tone ON or OFF) containing a Morse coded message: 5 short tones ("5"), two long tones ("M", minus), three short and two long tones ("3"), resulting in the math problem "5-3"; and accordingly the second math problem, "0+9".