Source: Deep Learning on Medium
This is a nice statement on imagination and dreams. A nice motto I encountered on a just-boarded up Motel, along US 101, near San Francisco in the early 1990s was “Worry is a Waste of Imagination”. Which I thought was brilliant.
That said, as a computational cognitive neuroscientist, you might be interested to know that machine learning AI guys have been playing around with dreaming a lot. Most notably the Geoffrey Hinton of “Deep Learning” fame runs his neural nets to imagine the image of a dog after they have learned to see one (using dog as an example here). These machine imagined images are quite cool to look at because they are dreamlike, and such. The literature on this subject is getting quite large and I am sure is accessible with google scholar or such.
This imagining is also key to understanding how humans work with truth and lies in natural language and memes and such. Oddly, imagination is inherent in all our communications in judging truth and lies, and bears a threat to our very existence when we cannot use our brains to see into each others’ brains through dialogue.