1. The first level of consciousness should be simple instructions for actions to be performed (also triggers for the actions, and rewards and perils), no feedback loops of any kind, and cost functions will not evaluated. We will need to introduce chaos, not in the general meaning of the term, but randomness, so that better mutations survive
  a. If we talk about survival, then we’ll need to define perils too, but I write this in a side note, for it has to be a property of the environment rather than the organism itself, also the rewards and triggers as well

2. The second level must include feedback loops, such that in the process of doing a task, due to the randomness, if a reward is received with a lower cost function, the event is recorded and new associations surrounding the reward must be built, on sufficient reinforcement, this should become the default way of doing the task

3. The third level will need to have VM’s that will be stripped down versions of the organism itself, embedded inside the organism, that will imagine a limited amount of possibilities, and evaluate the cost functions beforehand

4. The fourth level will evaluate the efficiency of the VM’s, and create better VM’s, stop when further introspection leads to the same results, or consumes resources in excess