[reaanb2]

Consciousness for me is tied to internal modeling of the world and oneself in it. An agent without an internal model is limited to reacting to immediate inputs, and constrained in terms of the complexity of memory it can have. It could sense pain, for example, it could react based on the location of the pain, it could have thresholds based on experiences of pain. But it could not have a concept of itself being harmed, or objectify the source of the pain, or reason how to minimize the pain. It could only react, with some mix of direction or randomness.

With an internal model of the world, an agent can objectify patterns of experience, view the world as consisting of things, recognize things, associate behavior with and predict things, react based on a more comprehensive understanding of the structure of the world rather than just what it can immediately experience. For example, without looking, I know there's a closed door behind me. I have an internal model of my environment, built from simpler object models such as wall and door and window and floor. I am aware of my circumstances and of my position within it. I can interpret sensory experiences in terms of that model - the sound of the door opening, or a gust of wind on my neck. This is consciousness in my view - experiencing the world in terms of models which divide it into recognizable and predictable parts.

Different agents' modeling abilities can vary, depending on their available memory, pattern recognition abilities, and so on. Also important is the ability to hypothesize - manipulate models not related to immediate experience - and communicate models. Humans tend to be pretty good at both these things.

Free will, for me, is not an inherent ability of an agent, but rather a judgment of an agent's decision-making abilities from the perspective of another agent. If one agent can fully control or predict the behavior of another, the latter has no free will. To the extent that an agent is uncontrollable or unpredictable, it's said to have free will. Therefore I view free will as an arms race between competitive agents. Memory and modeling prowess are the basic weapons in this war, but they demand a high amount of energy. Evolution has equipped conscious creatures with a few other tricks, such as forgetting, which helps to increase unpredictability and keep down energy demands.

[sdwr]

That point about competitiveness resolves the consciousness "paradox" pretty much completely, and explains all the moving goalposts.

Judging something as conscious means believing in its agency