In which sleep state is sleep communication possible?
The sleeping person has to be dreaming, and this needs to be detected from the wake world, in order to be able to incorporate stimuli into a dream. Moreover, lucidity of the dreamer is necessary to ensure voluntary and conscious communication. It seems reasonable to use REM sleep as the preferred sleep communication sleep stage. Even though dreams can occur in all sleep stages, most dreams (and especially most lucid dreams) take place during REM sleep (see chapter 2). Thus, it seems reasonable to concentrate on REM sleep for sleep communication purposes in order to increase the probability for successful sleep communication. Presenting stimuli the whole night independently of the sleep stage does not seem promising, since there is a high chance of waking up the sleeping person, especially in the N1 and N2 sleep stages (compare the experiences from the experimental part of this thesis, chapter 5).
Can you use any sort of stimulus?
The stimulus being used has to be able to contain (possibly encoded) information. It is not absolutely necessary to use words as a stimulus in order to have an information-containing stimulus. By using a coding scheme as for example the Morse code, also light flashes, sinus tones or even tactile stimuli can easily transport arbitrary information. Disadvantage of such a coding scheme is that the dreamer has to learn the scheme if he is not familiar with it.
Next, decoding the message from the stimulus has to be simple, so that the sleeper does not have to concentrate too hard on the decoding. Unstable dreams or awakening might be the result.
How do I determine a good stimulus intensity?
Furthermore, the stimulus has to be of the right intensity. For example, auditory stimuli must not be too loud in order to not wake up the sleeper, but also not too quiet, because they might not be incorporated into the dream world then. A method to find a suitable intensity could be to start with a low value and increase the stimulus intensity step by step, until the sleeper either lucidly signals a dream incorporation or wakes up (in which case the intensity shortly before waking up can be taken as a threshold value).
Finally, the stimulus has to be easily detectable, even when being modified by the incorporation. A suggestion would be to use a stimulus of an exceptional, unusual kind so that the dreamer has no difficulties to keep track of the message. An exceptional stimulus might help the dreamer to become lucid, too.