The experimental pilot study led to several results. Most importantly, it was shown that a successful sleep communication is possible using the Sleep Communication Framework and the Sleeator 2 together with the Zeo as its implementation. The successful sleep communication consisted of a math problem with random numbers and operators which was sent into the dream, solved inside the dream without awakening, and sent back to the wake world. The successful sleep communication was recorded by EEG and dream report. Additionally, further (partly successful) attempts during the self-experiments and during the non-self-experiments showed some very promising results, too, as for example a dream of subject 5 in which she correctly identified stimuli that had been incorporated into her dream, but unfortunately woke up before she could answer a complete math problem.
Since the resources (especially the time resources) have been limited in this master thesis project, only a pilot study with a relatively low number of participants (five from the non-self-experiments plus one from the self-experiments) and a low number of nights recorded (eleven plus two, not counting the early self-experiments during the development process of the Sleeator 2) could be conducted. Thus, an advanced statistical analysis, on how well the ten requirements for a successful sleep communication as given by the Theory of Sleep Communication can be fulfilled using the Sleep Communication Framework and the Sleeator 2 implementation, does not seem legitimate. However, these ten points shall at least informally be commented on, based on the experiences of this pilot study:
(1) The person to sleep communicate with has to be sleeping and dreaming, and this has to be detected by a dream sleep detector or by a body signal detector and (2) Playback of a stimulus containing a message, possibly encoded using a coding scheme (here: a random math problem): About 71 % of the awakenings of subjects of the non-self-experiments following stimulus presentation took place during dream sleep. This number, however, has to be treated with caution, as it strongly depends on the device and the REM detection algorithms used. As stated above, further studies focusing on this point are necessary. The stimulus generation (1000 Hz sinus tones) was performed automatically by the Sleeator 2, which led in some cases to the result of too loud stimuli. From this experience, it can be concluded that a stimulus control (on, off, intensity) from within the dream is necessary. (3) Incorporation of the stimulus into the dream: All of the in total six subjects incorporated the stimulus into at least one dream, which shows that even this very simple stimulus of 1000 Hz sinus tones can be used for sleep communication. On average, incorporations could be found in 40 % of the reported dreams. (4) Lucidity of the dreamer: This point was the bottleneck of the experiments. Only two of the five subjects of the non-self-experiments were able to produce lucid dreams (one of which only had a one-second lucid dream); moreover, maintaining the lucid state without waking up long enough to fulfill the task was not possible to the independent subjects. Thus, there was only one lucid dream during the self-experiments long and stable enough to attempt a complete sleep communication, which succeeded. It could be found out, that if the sleeper has a lucid dream, it is also possible to actively look out for incorporations, as has been shown in this one lucid dream in which the dreamer looked for something that could beep, found it and then was able to sleep communicate. Increasing the chances to obtain and maintain lucidity thus seems to be a crucial point for future sleep communication. (5) correct detection of the incorporated stimulus by the dreamer: There were no subjects who have falsely identified something in their dream as an incorporated stimulus. Whenever the subjects reported an incorporation, there was a stimulus played before. However, this might also be due to the low number of subjects and nights. (6) correct decoding of the stimulus message if a coding scheme is used: there are mixed findings concerning this point. In the completely successful sleep communication, five of five incorporated signs have been decoded correctly, however, there were also some numbers wrongly decoded in the first nights of the experiment due to mixing up long and short Morse signals. An easier to decode message coding scheme or more training before sleep communicating might help overcome this problem. (7) comprehension of the message transported by the incorporated stimulus by the dreamer: If there was a message decoded, there was never a problem like: “4+4. What does this mean?” On the contrary, the two subjects having decoded incorporated numbers during their dreams stated to have known perfectly well what these messages were good for. (8) heeding the incorporated message by the dreamer and thinking about a response: In the dream report of the one completely successful sleep communication, no problem was stated with calculating the solution to 4+4. However, the message at hand was not very complex, further research on more complex messages is needed. (9) sending the response back to the wake world by encoding it into a body signal, and (10) correct detection and decoding of the body signal by hand or machine (here: an eye movement signal): Again, apart from the lucidity signal, which was successfully sent several times by both (partly) sleep communicating subjects, only a few answer numbers have been sent by encoding them into body signals. However, nearly all of these body signals could be decoded by hand, even using only the Zeo EEG and the eye movement artifacts within the signal.