Seeing red

[Angrboda]

Okay, let's revise the word "cause" to "is."  Let's say that the brain, as a whole, coordinates input from various sources and collates them into an overall representation.  So when you say "isomorphic," you mean that the representation in the brain mirrors, at least approximately, the world as the organism (or robot or whatever) experiences it, constrained by the physical limits of the organism to collect information about the world. [In my view it models a reality that only 'partly' is real; things like thoughts, where the mind is, its ownership of the body, 'the self' are mere constructs, parts of a representation which don't represent external things but rather are simply themselves part of the furniture of this mental representation.]

My question is still about levels.  What states are representative, and which are just states?  It's hard to verbalize what I'm asking here, but let me try an analogy.  Let's say we are making a physical reprentation, i.e. a model, of the world.  We could start by building little trees, little cars, etc., each of those in turn composed of leaves or wheels, each of those composed of subparts.  We would know also that each subpart consisted of chemicals with specific structures, atoms with specific structures, etc.  Some of those (like the QM mechanics) we might be completely oblivious to.  At which level of organization do we start to accept that we are building part of a tree?  Only when we have the finished model?  Only when we have identifiable parts, like leaves and branches?  Right from the start, since we know the chemical compounds we are making are part of that final tree, despite having nothing of tree-ness in them?

I understand what you're asking and it's hard to verbalize an answer beyond that I don't think it works like that at all.   The states are representative if they participate in the feedback loop of the representational system in a significant way, meaning that they affect the outcome of output stages which then affect the content of input stages.  A lot of this is done by our mind subconsciously, by chunking input into symbolic entities.  Referring back to the robot on the road, states are figuratively said to refer; that's where you get the idea that I'm simply circularly referring to the representation as a representation because of its analog to human representation.   We think in terms of multiply embedded representations and some of those representations have the character of being maps or models of the external environment.  We recognize an analog in the robot's memory.  But forget that level of representation a moment.  There is a level of representation in which a sensor picks up the contours of a rock in the road, those contours are realized in some state, and, the state is such that it triggers a response to rock in the road-ness, which causes the vehicle to swerve, which destroys the state of rock in road (literally, the rock is no longer in front of the vehicle).  It is this feedback slice which constitutes representation.  So ultimately what matters is not any granularity of state, but granularity of feedback behaviors.   Whether that rock in the road is a rock, or an oil drum, or a person, all that is being represented is the trigger of a behavioral response.  It is holistic in two directions.  First, isolated states don't represent; the system represents, with some analogs to human processes that likewise feedback into the environment.  Second, what is actually represented is not a rock; that could be anything.  What is represented is a set of triggering conditions, and those are dependent upon what behavioral responses the robot has programmed into it.  Whether it represents at the 'tree' level or at the 'branch' level or at the 'leaf' level is not determined by the state by itself, but by what things each of those levels trigger.   I may 'see' at the tree level if I'm driving and trying to avoid obstacles; I may see at the branch level if I am trying to get up out of reach of a hungry tiger; I may see at the leaf level if I'm looking for fruit to eat.   The 'level' of the state is fixed by the behaviors.  In humans, because we have such complex representations, it may be hard to see the connection between perception and behavior, but I believe that is the ground floor at which all our models, maps, and representations operate.

Let me ask you this: would you say that our mental representation is composed of parts, and those of subparts?  If so, how many child nodes would you allow and still call an idea part of that representation?  At what scale of order would you say, "That's just stuff" ?

You'll have to be more specific about what kind of an idea you're talking about. The granularity of our perceptions is largely determined by the subconscious systems which perform that perception. My visual system tends to chunk things into images composed of objects, my linguistic system tends to chuck things into sentential 'images' (words and thoughts, propositional), my somatosensory systems tend to chunk things into sensations such as pain or touch, associated with a body location, and so on. It would help to give examples if you could be specific.

I think in terms of economies. If an input can form the coin desired by a representing/mapping operation, and that mapping can provide the coin recognized by an action system, then that action system can cash out that input in terms of behavior, leading to new input. What is accepted as coin by each stage of the representational loop is not determined globally by states of the mapping operation, but by what inputs/maps/actions/results can process the individual coin. It's all just 'stuff happening' but it is stuff happening in a certain, preordained, sustaining feedback loop. With humans we have loops within loops, but at the top level is a representation of a 'self', meaning the mind+body as a representation/map/model in a large feedback loop. Again, it's hard to put into words. I've resisted the urge to attempt to explain my model of mind because of this difficulty in the past, because I simply don't know how to explain it clearly.