The famous French novelist, Marcel Proust once wrote that “…one cannot properly describe human life unless one bathes it in the sleep into which it plunges night after night and which sweeps round it as a promontory is encircled by the sea…” 

A team of researchers at the famous Massachusetts Institute of Technology’s (MIT) Dream Lab are working on an open-source wearable device - a glove - that can track and interact with dreams in an attempt to give us the opportunity to ‘bathe in our dreams’ and hopefully have control over the content of our dreams. “Think the ‘Inception’ in other words, but with a Nintendo Power Glove,” writes Futurism. 

The glove called Dormio was conceptualised by Adam Horowitz, a PhD student at MIT Media Lab’s Fluid Interfaces Group and a Dream Lab researcher. The Dormio, using sensors wrapped around the user’s wrist and fingers, tracks muscle tone, heart rate and skin conductance to identify when the user is asleep. 

Once they enter hypnagogia, an audio cue plays, mostly one word. For the experiment, Horowitz tested the audio cue ‘Tiger” on 50 people; a tiger appeared in many of their dreams. “What is so exciting about hypnagogic dreams is not the comparative emotionality or narrativity, it is the semi-lucidity: As we are not fully asleep, we can survey the changes in our mind during sleep onset, watch where dreams go, feel our executive function fail and watch our thoughts take on a life of their own, and report them at rates above 90% when we’re woken up,” writes Horowitz in this PhD thesis. 

The Dream Lab team was inspired by the age-old ’Steel Ball’ technique which was used by the likes of inventor Thomas Edison and surrealist Salvador Dali to get answers and be inspired by their dreams. They would nap with a heavy object in hand, and when muscle tone lessened and the object dropped to the floor below, awakening during sleep onset to capture ongoing hypnagogic cognition. Edison purposefully napped while musing over an unfinished problem, and when it did come up in his hypnagogic dream, called this twilight zone of semi-lucid hypnagogia his ‘genius gap’ and found it fertile for creative solutions. Dormio, a serial incubation device is a closed-loop, sleep state aware system enabling presentation of dream direction stimuli specifically during hypnagogia. 

The team also built an Artificial Intelligence (AI) algorithm for hypnagogia detection into the MUSE EEG dataset which tracks real-time feedback on brain activity to detect the end of hypnagogia and beginning of unconsciousness to look for wider frequency windows of the hypnagogic state. The preliminary algorithm uses a variety of existing techniques from the automatic sleep spindle detection literature, particularly Molle and Devuyst to detect the number of times a person slips into hypnagogia state to trigger the audio cue multiple times, to make the most of it. 

Dormio is also informed by past sleep neuroscience work and aims to democratise the technology to make the science and technology relevant to more people. The team replaced polysomnography, an expensive and cumbersome sleep-tracking technique with sensors which allowed them to make cheap, wearable, simple tracking and intervention systems. A step-by-step guide was posted online with biosignal tracking software available on Github, allowing everybody to theoretically make their own Dormio glove. The creators believe that Dormio’s breakthrough can help conduct more research into sleep and its relationship to memory, learning and creativity. “The applications of dream themed human-computer-interaction (HCI) devices range from nightmare alleviation in PTSD to creativity augmentation, anywhere that dream content has a marked effect,” writes Horowitz. 

A similar device built by Dream Lab researcher and PhD candidate Judith Amores relies on smell rather than an audio cue. A preset scent is released by a device when the user reaches the N3 stage of sleep, a regenerative period when the body heals itself and consolidates memory. The idea is to strengthen this consolidation using scents.