parallel vs serial neural processing

[gib65]

This may seem like a really odd question, but I need to know the answer for a paper I’m writing. My question is also hard to articulate so I hope no one minds if I make it rather lengthy.

 

I’m wondering about how signals in the sensory cortexes of the brain, and particularly in the visual cortex, are relayed to the cognitive centers. Specifically, I’m wondering if the linear processing of information (from simple feature detectors to more complex features to abstract features to concepts and so on) is completely serial or is there some parallel processing, and if there is, how much? For example, the diagram below shows the a simplified model of how information is processed from the raw visual input to the cognitive centers:

 

 

I’ve used the example of a cup in this diagram. Even though this is general how it works, something doesn’t make sense to me. If the information was purely serial, as it is depicted here, then we would be able to gain knowledge of the fact that we are seeing a cup, but all the details of the cup (lines, colors, shapes) would be lost. In other words, we would be able to say “I see a cup” but we would not be able to say what color it was, what shape it was exactly, whether the handle was on the right side or left, etc. The reason it seems this way to me is that the only information that is relayed to the cognitive centers is the recognition of “cup”, but not the information of shapes, lines and colors you see in the previous steps of this serial process.

 

Therefore, what I’m wondering is how accurate this alternate diagram is (neurologically):

 

 

In this diagram, the recognition of a cup is still relayed to the cognitive centers, so we still see a cup, but all the information in all previous stages is also relayed directly to the cognitive centers in parallel. In this way, we can also say what color the cup is, and shape it takes, and so on.

 

So my question is: Has there been any research to show that the second diagram is more accurate than the first in a neurological sense?

[Douglas]

It seems to me that your latter diagram is series/parallel processing. How about if there were four arrows from the cup to the circles and four arrows from the circles to the square??

[sunspot]

Although I am no expert on neurobiology maybe this example might help. If you went for a hike in the woods, in another state, you would notice a bunch of trees that were new. Without much familiarity with these trees, you would probably just see a bunch of unusual trees. If someone familar with the trees told you which were which, you would begin to see the very same trees differently. In your mind, a certain tree would change from a generic image of a tree into say a swamp oak. At that point, you would also notice distinctions from other trees and well as variation from swamp oak to swamp oak that were hidden before.

 

The sensory signals into the eye did not change, only the way the data was processed within the imagination. This analogy almost indicates the same data input projecting onto an image/concept within the mind. The better the image/concept the more data that is conscious. In other words, each day of hiking would add more and more layers of details to the inner image allowing more and more sensory data to become conscious. The scientist sees the most subtle distinctions due to his or her inner image being highly refined.

[bascule]

The second diagram is how Dennett's multiple drafts model operates. Information generated by the primary sensor centers in the cortex is passed around in parallel and selectively combined by the cortex to form a more coherent picture (what we experience is the selective recombination of all sensory data within the cortex) Once a processing "mistake" has been discovered (i.e. "That's not a flashlight, it's a fire extinguisher!") the "correct" version will immediately replace the mistaken versions.

 

If you want to read all the experiments used to justify this model please read the An Empirical Theory of Mind section in Daniel Dennett's excellent book Consciousness Explained.

 

I should also note that much of the sensory preprocessing is in serial. These parts of the brain can't make "mistakes" because they're merely applying transforms, not attempting to look for patterns in sensory data.

[gib65]

It seems to me that your latter diagram is series/parallel processing. How about if there were four arrows from the cup to the circles and four arrows from the circles to the square??

 

So you mean no arrows between the circles? Each feature of the cup gets processed individually and directly?

 

Well, you may be right about the four arrows going directly from the cup to each of the circles, but based on what I know about the neurology of vision, the arrows linking each circle to its neighboring circle are correct. Information coming into the primary visual cortex first gets processes as points, and then that information is passed on to "line detectors" and then certain collections of lines are recognized as shapes, and then this shape information is passed on to "object recognizer" that associate the shape with a label like "cup" or "tree" and so on. However, I don't think the visual system can recognize the raw incoming data as a "cup" or "tree" right away - that is, without some lower level processing of the lines, colors, shapes, and so on. I think what the neurolgical sciences tell us is that the brain identifies the objects it sees based on its components like lines, shapes, colors, and so on.

 

Although I am no expert on neurobiology maybe this example might help. If you went for a hike in the woods' date=' in another state, you would notice a bunch of trees that were new. Without much familiarity with these trees, you would probably just see a bunch of unusual trees. If someone familar with the trees told you which were which, you would begin to see the very same trees differently. In your mind, a certain tree would change from a generic image of a tree into say a swamp oak. At that point, you would also notice distinctions from other trees and well as variation from swamp oak to swamp oak that were hidden before.

 

The sensory signals into the eye did not change, only the way the data was processed within the imagination. This analogy almost indicates the same data input projecting onto an image/concept within the mind. The better the image/concept the more data that is conscious. In other words, each day of hiking would add more and more layers of details to the inner image allowing more and more sensory data to become conscious. The scientist sees the most subtle distinctions due to his or her inner image being highly refined.[/quote']

 

What you seem to be talking about here, sunspot, is information from passed memories and stored knowledge effecting perception. In other words, we could diagram what you're saying by putting arrows from the "knowledge and memory" box to the circles - that is, in the reverse direction of the arrows that are already there. I think this actually does happen to a certain degree. Studies have shown that past experience and knowledge effect how we perceive things.

 

Nevertheless, I may not be able to identify a new tree as an oak or birch or whatever, and I may not see all its details at first, but if I was to look for its most detailed features, such as color, size, orientation of lines, location in space, and so on, I wouldn't have any trouble doing this.

 

I appreciate the feedback, but I'd still like to know which of my two original diagrams is more accurate. Do each of the circles send signals directly to the cognitive centers of brain (2nd diagram) or is it completely serial (1st diagram)?

[Douglas]

So you mean no arrows between the circles? Each feature of the cup gets processed individually and directly?

How about if we replace the "cup" with the word "T I M E".

 

It seems to me if you're doing serial processing, and if the 2nd set of data that represents the "I" has malfunctioned, the processing will stop, and the brain will just recognize "T ? ? ?"

 

If the brain is doing parallel processing, and the data containing the "I" has malfunctioned, the brain will recognize "T ? M E". Since the brain is capable of filling in missing data, it will know that the word is "T I M E".

[ashennell]

If we look at the substrate for this kind of precessing, namely the visual cortical areas of the 'what' pathway, then there is a hierarchical organisation althought it is not by any means completely sequential.

 

Has there been any research to show that the second diagram is more accurate than the first in a neurological sense?

 

Perhaps the most important addition to your first diagram would be a seqeunce of connections going in the opposite direction. The visual system contains both bottom-up and top-down flows of information. As a rule of thumb, bottom-up information flows level-by-level up the hierarchy while top-down information can often be passed down a number of steps at a time. For example the highest level in the hierarchy could projet to 2 or 3 levels below it.

 

This is a more accurate abstraction of the organisation.

 

If the information was purely serial, as it is depicted here, then we would be able to gain knowledge of the fact that we are seeing a cup, but all the details of the cup (lines, colors, shapes) would be lost.

 

It would take quite a lot to explain why this really the wrong way to look at how the brain is organised. The feedback projections are important. For example if wish to see if our cup has a square handle or rounded handle then this top down 'question' could alter way the bottom-up information is processed so the specific details of the object are focused on instead of the whole. Perceptual processing is context sensitive. Selective attention is probably our subjective experience of this process in action.

 

Conscious awareness of the cup probably works a little differently - we don't necessarily only become aware of the cup at the top of the perceptual hierarchy. I'm not sure if this is relevant to your question.

 

Part of your point about losing the details is certainly true. In some cases we find it very hard to not perceive the top-level concept and only see the detail. I notice you have memory at the top of the hierarchy- I'm not sure that this is completely accurate but anyway imagine trying to store the details of a cup without the concept of the cup into memory. The reason why this dosn't happen is not because we can't (because of connectiviety) but because the individual parts of an object have no behavioural significance on their own.

 

I hope this is useful. Maybe it is a little confusing?

[gib65]

If we look at the substrate for this kind of precessing' date=' namely the visual cortical areas of the 'what' pathway, then there is a hierarchical organisation althought it is not by any means completely sequential.

 

 

 

Perhaps the most important addition to your first diagram would be a seqeunce of connections going in the opposite direction. The visual system contains both bottom-up and top-down flows of information. As a rule of thumb, bottom-up information flows level-by-level up the hierarchy while top-down information can often be passed down a number of steps at a time. For example the highest level in the hierarchy could projet to 2 or 3 levels below it.

 

This is a more accurate abstraction of the organisation.

 

 

 

It would take quite a lot to explain why this really the wrong way to look at how the brain is organised. The feedback projections are important. For example if wish to see if our cup has a square handle or rounded handle then this top down 'question' could alter way the bottom-up information is processed so the specific details of the object are focused on instead of the whole. Perceptual processing is context sensitive. Selective attention is probably our subjective experience of this process in action.

 

Conscious awareness of the cup probably works a little differently - we don't necessarily only become aware of the cup at the top of the perceptual hierarchy. I'm not sure if this is relevant to your question.

 

Part of your point about losing the details is certainly true. In some cases we find it very hard to not perceive the top-level concept and only see the detail. I notice you have memory at the top of the hierarchy- I'm not sure that this is completely accurate but anyway imagine trying to store the details of a cup without the concept of the cup into memory. The reason why this dosn't happen is not because we can't (because of connectiviety) but because the individual parts of an object have no behavioural significance on their own.

 

I hope this is useful. Maybe it is a little confusing?[/quote']

 

I think it's always going to be confusing, but I think I understand your point. It sounds like what you're saying is that the second of my diagrams (the one with some parallel processing) is more accurate, but it still needs some modifications. If I understand you correctly, the modifications I'd make is to make the parallel arrows bidirectional (or maybe even unidirectional but in the opposite direction from what I have drawn). This would imply that once all the information about the cup has been processed along the serial path, I become conscious of seeing a cup, but if I wanted to know what the cup's exact shape or color or whatnot is, my brain would send signals back to the appropriate circle and probe it for information, sort of like a request for the details concerning that feature. This actually makes more sense out of the way my perceptions of things feels than my second diagram. That is, when I look at a cup, I do indeed see a cup but I'm not necessarily paying attention to its details. Nevertheless, I could easily focus attention on its details if I wanted to, and I would have no trouble perceiving them.

[insane_alien]

wouldn't object recognition come from knowledge and memory or am i off on a tangent here.

[sunspot]

I would have to say the second diagram. The reason I say this is that if the first was correct one could not see just dots if there was a line present. I believe the memory is arranged in layers but in a parallel fashion which allows us to jump between layers. The best example of this are abstract drawings with hiddened objects. Many people are often stuck on the dots or lines until they see the object, then that is all they see. The imagination must start at the top layer of dots and work downward. The optic nerve has plenty of bandwidth for all these parallel layers to flow into the brain at the same time.

[gib65]

wouldn't object recognition come from knowledge and memory or am i off on a tangent here.

 

Yes, but this is a different kind of memory than what I intended to convey in my diagram. "Object Recognition" is a type of memory for what things are. The box labelled "knowledge and memory", however, means to convey the storage into memory of specific events or instances, such that I can say "Yesterday, around 4:00, I was looking at my blue coffee cup."

[sunspot]

If we look at these two distinctions, object recognition and knowledge and memory, knowledge and memory is done within the imagination. Object recognition requires sensory input in conjunction with imagination. The sensory imput can only consciously overlap what is in the conscious imagination. If we consciously separate from the sensory input, i.e., staring at the object but our consciousness drifts into the mind's eye, so we can think about it within our imagination, new relationships will appear. This allows the same sensory input to overlap the new more advanced image we created in the imagination. The unconscious mind stores all the data from the sensory organs. Consciousness filters within the imagination.

 

What is sort of interesting is that the link between the conscious and unconscious mind can have an impact on what we see. For example, in the woods at night, trees can become shadowy figures. The sensory data overlaps our conscious imagination, which is also being overlapped with unconscious content. The ego will make sensory comparisons within the aberated imagination to create scarey distortion of sensory reality.