… if you read the linked paper, they literally just have a neural network generate and optimize the image that it thinks the ‘patient’ is seeing. I fail to see how this is actually us seeing what we dream, it’s just a roundabout way of estimating what we’re thinking about via fMRI.