jamest wrote:GrahamH wrote:jamest wrote:I'm hoping to have more time tomorrow - especially in the evening - to deal with previous posts from the past couple of days. But I have a nagging question and wondered if there were any reasonable answers:
If brains states (NNs) are responses to the external environment, then how does the brain distinguish between those NNs that refer to its own body and those that refer to the environment beyond that body? That is, how does the brain distinguish between 'the self' (body) and the non-self (world)?
In answering the question, we have to remember that 'the body' is actually external to the brain. So, there will be NNs associated with the body in which the brain is housed and NNs associated with the environment beyond that body. The question is, how does the brain know which NNs refer to which?
By the connections, James. NN that are in a functional chain from neurons in the retina are responding to things 'seen in the world'.
But, one's body is "seen in the world", also. It's part of what is external to the brain.
So? One's body is seen at the centre of the perceptual POV. That makes it distinct from other objects that are not always present. I think your question has been answered. The differences in perceptual pathways define distinct classes <my body> and <the world around me>.
jamest wrote:
Relative location of objects can be inferred from an image,
Here's another problem: if the brain is privy to nothing other than its own brain states, then how does the brain know that specific NNs refer to 'an image'? To use the lego analogy again (of being aware of nothing other than complex lego structures - commensurate with NNs), how would you know that specific groups of lego structures referred to 'an
image'? You couldn't ever possibly know that unless you had been
told that they were.
James, give up on this Lego analogy. It is a major failure of thinking. The brain is fundamentally not a processor of abstract symbols. Since you like toy analogies try a different one. A child's peg board with various shapes.

The board 'recognises' certain shapes in certain locations. There is no need for the board to assign meaning, the peg fits or it doesn't. In fitting we might suppose it activates some mechanism. If such a system could form new holes when it encounters new shapes of pegs, and those shapes were produced by, say, sensory input from a tree, then it would learn to respond to trees without needing to 'know what a tree is'. When the tree is present the peg fits and activates the mechanisms that respond in ways appropriate to trees.
NNs form in response to input (learning) and their activation drives output (decision/action). The presence of the stimulus pattern causes the response. There is no need of a knowing homunculus interpreting the patterns. The patterns drive the mechanism.
jamest wrote:Another interesting problem that arises, here, is about 'space'. That is, how could NNs accurately account for the space between objects? You say that NNs are responses to objects and events external to the brain. But, how can NNs ever account for the void [of space] between objects? Can there be an NN that accurately represents a void of material influence? Moreover, how can the brain respond to 'a void'?
Show me 'void'.
'Space between objects' is perception of objects correlated with relative position of the observer. This positional information is not mysterious. I look to the left I see a lamp. I look to the right, I see a door. The 'separation' is the angle I must move my head, and the binocular convergence of my eyes, to see one or the other. NNs can respond to such triangulating information. We don't perceive a void, we perceive objects.
jamest wrote:Of course, I'm of the opinion that space (and time) are [absolutely] constructed by the self. But your model cannot embrace this idea, for obvious reasons. Therefore, I'd like to hear your responses to my questions about 'space', here.
Relative location of objects can be inferred from an image
Sure, if you are looking
directly at that image. But please remember that the brain is privy to nothing other than [relatively] static neurons. That is, there are no images of motion, per se, just change [of states of those neurons]. At the very least, this implies that the change of states of neurons just gives the
impression of motion - something that is discerned, rather than is actual. Again, concepts such as 'consideration' and 'meaning' rear their ugly heads.
There are responses to objects at differing positions relative to short term memory, that is perceiving motion. Flick books tube TVs and movies on film demonstrate that we perceieve motion from instants of perception of objects at different locations. Object recognition, location recognition and short term memory account for motion perception.
jamest wrote:
NNs receiving input from tactile and pain neurons in the body 'feel the body'.
Again, my point is that the brain shouldn't be aware of the fact that it is receiving "input from tactile and pain neurons in the body". In fact, the brain shouldn't be aware of anything, other than its own internal states. Upon this realisation, should all brain models be constructed.
The brain isn;t 'aware of own internal states'. The 'awareness' is those states. The states are what causes the responses. If I shout 'BOO!' in your ear it is your brain states that propagate from that event that make you jump. It isn't because you become aware of the noise and choose to jump. There is no homunculus inside hearing what you hear and deciding to tell the brain to make the body jump.
jamest wrote:Whenever I address the details of your model, as here, I see your failure to realise the limitations that your endeavour places upon you. I'm hoping, eventually, that you actually understand the purpose behind my posts. Then, you might see the value of them.
It would be a great help to this discussion if you could try to think constructive about the model and try to answer some of your own questions. Is that something you would conseder, or are you only here to attempt to destroy an idea you don't like.