> 20 look at my reaction to world

who is the "my" you are referring to? you have an 'a priori' conflict.

By "my" I guess he's referring to an internal state that has been built up over previous interactions with the external world? So more like:

  10 receive information from world
  20 do operations on information
  25 update state based on operation result
  30 goto 10

I think it's more like:

10 receive information about the world, body and reward signals

20 evaluate the current situation and possible actions

30 act

40 goto 10

Emotion results in step 20 when we judge situations and actions in the context as good or bad. In this step two subsystems cooperate:

1. a system for fast reaction - works best when there is no time to reason, or when the action is repetitive, or when available information is uncertain

2. a system for slow, reasoning based reaction - works when we can build a mental model and imagine possible outcomes, is especially necessary when we encounter novel situations and have the concepts necessary to reason about it

System 1 is based on instinct and system 2 is learned. They are both essential as they are specialised in different situations. Using system 2 all the time would be too expensive and probably impossible, we need to rely on instinct which in turn relies on evolution to be fine tuned.

Learning happens through the reward signal. We reevaluate situations and actions based on outcomes. Emotion is just a synonym for the value we assign to our current state with regard to our goals and needs.

Our goals include adapting to the environment in order to assure the integrity and necessities of life - the primary goal, then as secondary goals - being part of a social group, learning, mastery, conceiving children, curiosity and a few other instincts. We are born with this goal-program which is in turn evolved.

Yeah, this "perception-action cycle" has been well known and taught in neuroscience for a long time. What's new, at least since I studied it, is the anti-correlated tick-tock of these two key networks that seems to be happening. Amazing how similar to a game engine consciousness seems to be panning out.

> The default mode network (DMN) is an internally directed system that correlates with consciousness of self, and the dorsal attention network (DAT) is an externally directed system that correlates with consciousness of the environment

DMN : self awareness :: DAT : non-self awareness. The distinction is at least partly hard wired. If there's a self, there's a "my".

Another name for consciousness is self-awareness, which requires a self. And what else could a "self" be but a neural construct? This article is a theory of its construction.

I would say the kalman filter is quite appropriate where the state is our model of the world, the 'page flip' as a state update based on observation. Also the same as a Bayesian model update:

> state = kalman(state, observation)

with a kalman filter the model always stays the same and is used to pick which sensor is reporting most accurately. changing the model would change everything, maybe that's why changing your worldview has such an impact on the way you "see" things.

Taking this further, I suppose you could consider persistence of vision [1] as analogous to consciousness.

It's an artifact of the limitations of the system.

[1] https://en.wikipedia.org/wiki/Persistence_of_vision

Persistence effect has to exist in one way or another to get real-time self awareness; if it were not an input alternation, it would exist at the reasoning level.

Except without the frames and colored lines. Lines 10 and 20 don't provide the experiences. They're just behavior. Somehow all the sensations have to be added in when looking at the world and looking at one's reaction to the world.

let experience$ = INKEY$