By Daniel Oldis © 2017
In the not-too-distant future, an expert lucid dreamer will lie down to sleep in a magnetic resonance imaging (MRI) scanner. Non-metallic electroencephalogram (EEG) electrodes and near-infrared scanners (NIRS) will be attached to her scalp. Electrooculography (EOG) sensors will be gelled near her eyes; electromyography (EMG) sensors will be pasted to her arms, legs, wrists, ankles, elbows, knees and hips. EMG electrodes will be distributed around her mouth, chin and larynx; and our lucid dreamer will be primed to offer the world the first Magical Mystery Dream Tour—if she can fall asleep in such a contraption.
From a technological perspective, recording or reconstructing rudimentary dreams is a contemporary possibility with histories that go back to 1971. Separately, scientists and researchers around the world have reconstructed simple dream imagery (Horikawa 2013), decoded sub-vocal (dream) speech (Jorgensen 2006) and animated dream bodily movement (Oldis 2017). What these researchers have not yet done is come together (neuroscientists, speech professionals, kinesiologists, and computer programmers) to record a complete (though brief and blurry) dream.
“Ideally,
a lucid
dreamer
as the
research
subject
will take
us on a
brief, but
magical,
tour of a
simple
dream.”
There are several obstacles for such a “big dream” project. Funding for complex, interdisciplinary projects is a challenge, especially for projects with little short-term commercial, governmental or medical applications. While scientists, therapists and a curious public would likely welcome the accomplishment of an actual reconstructed dream (a dream movie), the cost and inconvenience of current technologies limit democratization of the procedure until hardware and expertise catch up to hypothetical medical and consumer applications of recorded dreams.
However, even greater than funding challenges, the primary obstacle to recorded dreams streaming on Netflix next year is training: in order to decode dream imagery, speech and motor behavior, the software needs to collect examples of visual, vocal and muscular neural patterns of the dreamer awake: watching, saying and doing normal-life stuff. These patterns or features of these patterns then become the set of possible dreamed elements to be matched to the dream experience and decoded.
And there’s the rub. “Set of possible dreamed elements,” of both “normal-life stuff” and non-normal stuff that dreams can contain is a pretty big set (though published neural feature patterns—color, shape, speech, facial, muscle, etc.—can help). The training cycle under MRI, NIRS and EMG recording is extensive and expensive. Example sets for visual imagery, voice patterns and muscle movements are large. And dream variability may further distort any trained example.
So, enter our lucid dreamer (call her Lucy). Lucy, an experienced lucid dreamer, has the ability to make dream action choices: walk to the tree in the distance; search for a lake; fly to a mountain; enter a house of a friend. Lucy, as an advanced oneironaut, can also, at times, manifest a specific dream scene on demand: an ocean beach, or a cabin in a woods, or a mirror in an attic. Lucy has learned how to make specific body movements in her dreams (move a leg or arm, make a fist) and make specific speech productions: “hello world,” “I’m flying,” “I see a river.”
