I suddenly thought of a question while sleeping yesterday: Is the principle of dreaming similar to that of Midjourney's image generation?
Before falling asleep, the books I read, the movies I watched, the music I listened to, or the chaotic thoughts I had often get carried into my dreams. Throughout history, there have been many remarkable individuals studying the control of dreams, but precise control has always been elusive. One can only control the frequency of specific content appearing in dreams with a high probability.
This is similar to the images generated by Midjourney, where precise control over the generated images is not possible.
According to some studies, our brains engage in various activities during sleep, including neural activities related to memory, imagination, and emotional processing. These activities can manifest in our dreams, producing a wide range of scenes and plots.
Some researchers believe that we can influence the content of our dreams by inputting specific information into our brains before sleep. This method is known as "dream incubation."
For example, vague input instructions can be seen as a form of dream incubation. Some experiments in dream incubation have shown certain effects, but these effects are often unstable and may vary from person to person.
Dream incubation is not a method for accurately controlling dream content. Our dreams are autonomously generated by the brain and are influenced by many factors, including our emotions, experiences, and memories. Therefore, even if you engage in dream incubation, there is no guarantee that you will have dreams similar to those generated by Midjourney.
Both dream incubation and Midjourney's image generation influence outcomes by providing some vague instructions, but the final results are often uncertain because they both involve factors that are difficult to control precisely.
In Midjourney's image generation, we can provide some keywords or simple sketches to guide the drawing direction, but since the drawing process is autonomously completed by the machine, the final result may differ somewhat.
In dream incubation, we can provide some vague instructions to influence dream content, but the final dream content is also difficult to control precisely because it is autonomously generated by the brain and influenced by many factors.
Although the results of these methods are difficult to control precisely, they can still produce interesting outcomes and help us better understand how our brains and artificial intelligence work.
Additionally, I have found that when I am in a good mental state and have continuous good sleep, the dreams I have are easily recognizable, and I can quickly realize that I am dreaming.
The resolution of scenes in dreams is very low, similar to low-poly models in games. Usually, we cannot clearly see the faces of people in our dreams; perhaps this requires activating more brain neurons for detailed facial computations. Seeing the faces of people in dreams is not our goal in dreaming.
Low-power operation maintains brain activity frequency, integrates information from the day and past, deletes redundancy, and retains important information to convert it into long-term memory.
When our physical state is weak, we often cannot distinguish between dreams and reality, especially when experiencing nightmares, where there can be a sense of immersion, even to the point of being so scared that we cannot wake up.
When I can see at a glance that a dream is just a dream, I am in the dream, rather than the dream containing me. Thus, there is another method to determine whether the body is in a sub-healthy state.
In our dreams, we strengthen the processing and consolidation of memories we have learned and experienced while awake, which helps us better remember and apply this information.
Dreams can also help us process emotions and cope with stress. Some dreams can help us release stress and emotions, alleviating our negative emotional experiences while awake.
In dreams, our brains can freely explore a variety of imaginations and thoughts, which helps us better unleash our creativity and imagination while awake. This is similar to how Midjourney continuously accumulates user operation data to train its data model.
What if you frequently have the same dream? Mapped to the real world, perhaps this scene represents a challenge that needs to be addressed or an emotional issue that requires resolution. It’s like a line of fixed code that triggers a specific dream scene when encountering difficulties.
Dreams occupy half of our lives, yet we still know very little about the nature and function of dreams.