Bold claim to go from REM in sleep-like state to dreams and consciousness, and the original paper is not making that claim.
A good example is how they are able to recognise when certain prey is acting odd, deduce it’s injured and drop their stealthy approach for a more direct one. They’re also capable of remembering their environment and using indirect and often complex paths to sneak up on prey.
All of this seems rather possible even with basic learning mechanisms on molecular level. Not sure why you would claim that this need consciousness. But if you have a paper on this topic I would be more than interested to read them.