Initial experiments for the History of Art project with the term “neurosis” in the prompt returned results with brain-like imagery:

the depiction of neurosis in the history of artthe depiction of neurosis in the history of artthe depiction of neurosis in the history of artThe image with the deflated brain (or is it a bunch of small brains piled like pillows?) was intriguing, the overall effect was not what I was going for. I then redid the prompt adding text weight modifiers to downweight or remove any brain-related output.

the depiction of neurosis in the history of art::1 brain::-0.5the depiction of neurosis in the history of art::1 brain::-0.5the depiction of neurosis in the history of art::1 brain::-0.5While the engineers who collect the data, design the model, and train it are responsible for what text-to-image generators can produce, the user still has a duty to accept, reject, or modify the results of their prompt.