Brain Physiology, Cognition and Consciousness group: topic
This is a public discussion board
From Neuron to Memory System: How Memory Might Work
Graeme Smith
Friday, 12 June 2009 04:34 UTC
How does memory work?
We think we know, based on thousands of years of research into it, but there are still indications that we might be completely wrong. Scientists created computers with the best scientific approach to memory based on current scientific theories, in the 1940’s, and they got it wrong. Most neuroscientists today will tell you that the computer is nowhere close to a good model of memory, yet many of the attitudes that informed the decisions on how to make computer memories remain unchallenged. Our popular theories of how memory works were laid down in the middle ages, by people who thought memory was a fluid, and that the brain was a reservoir. Somehow they thought, we opened valves, and let the memory flow from the brain to where it was needed. I call this a demand model of memory because you demand the memory by opening the valves. In computing terms we have replaced the fluid theory of electricity but retained the idea of current, and at one time tubes which were replaced with transistors were called Valves. So computer memories could be seen to be operating in a demand memory model as well.
As Jerry A. Fodor said in The Mind Doesn’t Work That Way!:The scope and limitations of Computational Psychology Neither Phenomenal approaches such as Neural Networks nor strict computational approaches such as Truth Preserving Functions, seem likely to achieve a suitable computational model of consciousness.
As a researcher in Artificial Consciousness, my main thrust is to eventually get support for my own Artificial Consciousness to be built. However to get there, I had to start with a model of memory. I chose to hedge my bets, to begin with a Neural Model, and add Functional support to that model, where it was needed. Further I wasn’t going to limit myself to Truth Preserving Functions, where Soft computing would be more effective. This type of approach to consciousness is called a Hybrid approach. To start off, however I wanted a model of memory, and one of the things I had to do, to get there was overcome Fodors assurance that no-one had yet developed a phenomenal version of a demand memory.
Those who are stuck on William James idea that phenomenal systems can’t be built based on what we know about the brain, have problems with my use of the word phenomenal in the way that Jerry Fodor used it in his book. They like to look for exotic things like QM entanglement to explain the fact that some things just seem indivisible. I however think that there are indivisible elements in the memory but that it doesn’t matter because the mind doesn’t try to divide them, it finesses the system, without that effect.
Fodor’s claim is not without merit, phenomenal systems are definitely easier to build out of neural networks, than demand memories. However, David Marr defined a type of memory he thought might help explain the cerebral neocortex back in 1970, and by understanding what his work uncovered, and brushing off some of the unfortunate assumptions of his day, I think I have undercovered one of the hidden biases that is keeping us from understanding the way the brain, or at least the memory works. And that bias is the assumption that explicit (demand Model) memory, is the natural state of the memory system, and that anything that doesn’t fit the model must be an add on to the basic explicit memory model. So we expect Implicit Memory to somehow be an add-on to the essential demand memory, and we are wrong.
What Marr described in 1970 is a type of memory more basic than explicit memory when implemented in neurons. To computer guys like me, this seems counterintuitive. How can anything be more basic than a good demand memory, we can implement dynamic ram with a capacitor, a resistor, and a transistor, how much more basic can you get? But what we keep forgetting is that neurons don’t work like transistors, there is no technological overlap. Within the logic that makes neural networks work, demand memory is much more expensive to build than a simple content addressable memory. That is what Marr claimed his 4 layer CODON was, was a self-classifying content addressable memory.
To understand why this might be, we need a little theory. Although in the 80’s and 90’s the connectionist school was over-run by the Parallel Distributed Processing guys from the A.I. discipline, and so we have to take some of its theory as being deliberately misleading to steer people away from trying to understand real neural systems, The basics are fairly equivalent. A neuron is first of all a cell, it has to survive like other cells, by absorbing nutrients and getting rid of wastes. However at some point in the evolution of animals, neurons gave up some of their survival functions to a helper cell called a glial cell, and converted those functions over into mechanisms for transferring information between cells.
The Parallel Distributed Processing guys figured what was important to this communication was the firing of the cell. Sorry that is a misconception, firing just speeds the process of transferring information up, it is not the only mode of transfer, nor is it the most important one for understanding natural systems based on nerve cells. However because the PDP boys wrote the manuals for the industry, they got to tell 20 years of modelers what to think. As a result of mistakes like this, they set the neural modeling of Natural Neural Networks back, to the point where Gary Marcus, has clearly stated that he feels it necessary to do a hostile takeover of the connectionist school.
One approach that the PDP guys thought was dangerous was David Marr’s attempt to use probability Mathematics to define a circuit that was made up of heterogeneous neurons. Despite the fact that they did not have the ability to model heterogeneous groups of neurons, a direct attack on his claiming that the heterogeneous group he called a codon was a content addressable memory, was made pointing out that the model did not exibit wave-forms similar to the real cortex.
This attack, assumed that Marr’s circuit had to have frequency artifacts similar to the real cortex in order to suggest the role of content addressable memory for the cerebral neocortex. Well Marr was a pioneer in the field and he got some things wrong, and others even more wrong, but it was a seminal stage in the science, and what he got right is more important.
Marr was able to use probability mathematics to analyze heterogeneous networks of neurons and predict their function. Nobody before that even tried, and after the way he was treated few after him had the nerve to try again.
I am not going to start spouting probability equations, if only because I don’t follow his math. But the PDP guys called a neuron a processor that had input, process and output capabilities. I however look at it from a different perspective I see a neuron that has storage, processing, and transfer capabilities. And when I look at the 4 layer cortex model that Marr worked with, I also see a content addressable memory, but I think that perhaps he expected too much equivalency between neurons in the connections department, and was sadly dissapointed before his death in 1980. So I wonder at how much self-classification the system is doing.
If you look at neurons from my viewpoint you begin to see that the shape of the neuron has a logic, that memory neurons have lots of synapses, and opportunistic process growth, that processing neurons tend to have complex and often bushy or mossy dendrites, and that transport neurons tend to have virtually no division at the dendrite level and form long thin neurons.
If you look at the type of organs that are produced, Memory Neurons tend to form tissues, Processing Neurons tend to form globular organs, and transport neurons tend to form fibrous bundles. In other words the shape of the neuron, indicates it’s function, and the shape of the organ might indicate to some small extent what type of function it performs.
Looking at Marr’s 4 layer model, we see that the first layer is processing neurons, the second and third layer are memory neurons, and the fourth layer is dominated by processing neurons again. The whole structure is probably a memory role because it is a tissue.
Stimuli coming in at the first layer, are processed affecting their storage in the second and third layers, and the 4th layer probably is mostly control neurons, and in some cases inputs of senses that have been directed through the thalamus. These latter signals condition the general inhibitive environment caused by the Martinotti cells and thus reduce the inhibition at the first layer encouraging the layer 2/3 neurons to fire.
This simple memory circuit can be seen to be designed to respond to either signals from layer 4 or layer 1, and the first layer is used to train the second and third layers to respond to patterns of stimulus. Given a stimulus each pyramidal cell in layer 2/3 decides whether or not to fire, and this type of system is exactly what is needed for a content addressable memory.
It is also exactly wrong for an explicit memory suggesting that it must be the explicit memory that is the add on. Indeed when we look at the micro-architecture we see that a type of tissue called Allocortical tissue makes up the tissues at the bottom of the Sulcys and Divides of the brain. and since Sulcys and Divides are a later architectural configuration there might be an indication that the six layer tissue called isocortical tissue, that pushes the gyruses up away from the sulcys and divides, was a specialization of the brain.
-
Replies
Jump to resultsResults
-
Otmar Pokorny said ‘A Given Memory is seldom a ’true version’ of a past event.’
While this is true, I don’t see it’s application to the subject. He goes on to say
‘The Original happening is experienced from a different perspective on the part of each involved.’
Also no doubt true, and well covered by the idea of perspective. He continues the statement with:
‘So the events implicaitons and basic meanings differ according to the fucus of each participant.’
While this interpretation does not stretch incredulity too much, he has missed the point of my work already. The fact is, that in a content addressable memory, each event is decomposed into multiple “points of view” according to which sense modality is reporting, and as a result, the content of each POV, is stored separately, and responds separately according to the stimulus that is currently being interpreted, at this level there is no such thing as an implication or a basic meaning just a reflexive reaction to a stimulus. Stimulus goes in, reaction comes out as a cloud of separately activated confirmations of content. Think of it as a whole crowd of groupies saying ‘pick me’ to a rock star. You can’t hear the individual signal for the noise of the crowd.
-
Dear Graeme,
I wonder what your starting point is: human memory or computer memory?
Both are very different.
Also I wonder what a model would be good for. The human memory is a natural phenomenon. Thus it can be studied. Theorizing maybe challenging, interesting but also a waste of time.
I advocate studying memory. In a computer, correct me if I am wrong, we know “everything” because we designed it and built it. The information in a computer has to enter it via completely known ways of input. Also the structure of information via bits is known and also the structure of the memory is known.In a human being, many of these aspects are not known or not well enough known. This starts with sensory input, its representation and processing in the brain until some part of it “enters” our consciousness. We also very importantly have to take into account, that human beings have a rich inner life with thoughts, emotions, imaginations and likes, dislikes, preferences, judgements etc. that also add to and modify our experience of the world around us.
Apparently part of conscious experience as well as other content enter memory of different kinds. Roughly long term and short term memory.Memory is an essential part of a healthy human being. How it exactly works or does not work so well in certain cases, is of great importance for this forum.
Many thanks for starting an investigation of this matter.
Yours friendly
Hans -
Ottmar you are still thinking in terms of complete memories. That is part of the whole memory system, not the area I am illuminating right now. Marr was not modelling the whole memory system, and did not do a good job of modelling the cerebral neocortex which he was claiming to model, but he did do a fairly good model of Allocortical Tissues which are it is true part of the memory system.
What Layer 2/3 store, are memories of course, but they are a special type of memory, they are recognitions of content. Thus the output of a Allocortical memory is the cloud containing all recognitions of content that were triggered by the stimuli currently being experienced. They cannot be separated into someones picture of a porch. Because there is no structure to their presentation. The stimuli from the outside world come in via Layer 4 and combine with the stimuli from inside the brain at layer 1, where the classification of the stimuli is probably done, to code it into the layer 2/3 Memory Neurons. A new stimulus, not only triggers the layer 1 neurons but it triggers the layer 2/3 neurons, which reinforce their own involvement with that stimulus. The primary difference between layer 2 and Layer 3 seems to be the depth within the brain that they reach Layer 3 connects deeper into the brain. In your next post you talk about probability matrixes.
I don’t know what you mean by probability matrixes, but the classification mechanism is probably a soft mechanism rather than a hard one. This would mean that it is a function of the network of all Layer 1 processing which layer 2/3 pyramidal neurons would be trained to self-select for which stimuli.
Consistency in knowledge at the neuron level cannot be maintained with just synaptic connections. Relatively new evidence is for parallel threads of biochemistry that implement different strategies for protecting the information depending on which thread of biochemistry is activated. Evidence that short term memory may not remember the same elements as long-term memory suggest that some of these threads can be disrupted during their processing by new information being added at the wrong stage of the process.
Essentially however the only way to stabilize memory at all, is to trade the relative instability of a chemical storage option for the more permanent solution of growing new fibrils to connect between the neurons. There is evidence that these new fibrils can be reabsorbed as part of the process of growth, especially in adult brains. Eric Kandel pioneered the biochemical work being done today, when he traced fibril growth to the expression of DNA triggered by the chemical cascade reaction started by the NMDA synapse in the hippocampus.
Recent work in Anesthesia has shown that there is a connection between 40 hertz SSR activity and Implicit Long-Term Memory, that suggests that there might need to be Thalamic involvement before memories are converted from synaptic storage to fibril growth, but it is early days for that suggestion.
Hans, I started off with a background in computing. However I disagree with you that studying a model of the human mind is useless. I think it has a lot to teach us about the nature of the human brain. In fact one of the problems I have, is that my theories about how the human brain work, are running so far in front of my ability to confirm them, that I am getting stressed out about how I am going to ever get even a portion of them proven. But that is my problem. I am reading a book on scientific revolution right now, and it seems to me that the main problem with science is how difficult it is to change directions when new information is available. One of my roles as modeler, is to suggest a new paradigm that is more demonstrably up-to-date and allows scientists that adopt it, to make informed decisions on what to study, that they couldn’t with the current mishmash of bits and pieces.
Articulating the whole model is a daunting task, and I don’t really want to spend the rest of my life squabbling over memory when I have already extended it in my mind to a theory of mind. But, we gotta walk before we can run, and I just this month made a breakthrough into some experimental protocols that can check some of the hypotheticals that my model is based on.
-
Consideration of cortical layers and associated banding (brought out using dyes) we find a general-to-particular focus (banding gets increasingly diffuse as we move from surface to core and so through cortical layers) indicates a grounding in recursive activities.
The banding identied at the surface level of the cortex covers all sensory/motor areas as it does abstraction areas (associations etc) and if we drill down we find the same banding present in the amygdala manifesting the presence of the fight/flight dichotomy after recursion (any ‘interdigitation’ of the elements of these dichotomies indicates the recursion of such.)
In the context of memory development, using a recursive process, there is a property that brings out a parallel format that allows for rich meaning generation spanning the whole network and at the same time allowing for local, specialist meanings.
The process reflects a mechanistic form of recursion out of which emerges an organic form that sets the foundations for meaning generation through a rich associative memory system – BUT it takes depth to achieve tbis.
Thus rote, mindless, XOR, dichotomised, forms of information processing will lead, over time, to the emerging rich associative memory – the scope of which is covered in such as development of savant skills. This dynamic covers a memory as the result of constructive and destructive wave interference patterns (understandable given the neuron’s focus on frequencies, wavelengths, and amplitudes).
As QM shows us, any experiments that work on recursing a dichotomy, such as left/right slit/down-converters etc, will introduce these patterns – and as my work has shown, the patterns are standard across ALL scales where information is dichotomised (see the diagrams etc in my IDM summary ) and our brains focus on dealing with complex/new information through oscillation across the full brain bring out a focus on recursion of the “what/where” dichotomy (more precisely what-who-which/where-when-how).
The layering of the brain, or more so the development of increasingly specialist ‘brains’ from our reptilian/amphibian background, means the same general to particular pattern of brain development as we seen in cortical layers development.
-
Hi Chris welcome to this thread.
I think you are missing the point of this thread, and overgeneralizing at the same time. While no doubt staining studies of one type or another might show activity, it depends on the staining study, what patterns will emerge.
For instance a classical serotonin stain will stain a solid stripe at the Striate Cortex, and in fact the striate cortex was named because the first study that named it was a serotonin study.
Since the striate cortex is deeper in the brain than the cerebral cortex, your overgeneralization about general to specific, may be reversed for serotonin stains. Staining studies are important, but without the information on which aspect of the cell you were staining for, they can’t be interpreted the way you just did.
Further, the Micro-Anatomy of the brain does not evenly distribute memory cells throughout the layers of cortex, they show up in distinct laminae, and just saying that the studies where more specific at layer 1 and more diffuse at layer 6, might also indicate that there are fewer Pyramidal Cells in Layer 6 than there are in layer 2 and 3.
I can’t comment on your assertion that the banding is also noticeable in the Amygdala, because I am only peripherally aware of that organ, and what it might do. Currently we are studying the memory, and the amygdala is considered part of the limbic system which influences but is not considered part of the memory system. When we get a little further into my theory, I will present information on what I think the connection might be, but for now, I am dealing with the first 4 layers of the cortex, and especially the first 4 layers of Allocortical Tissue that seldom has any more layers than that.
-
“Further, the Micro-Anatomy of the brain does not evenly distribute memory cells throughout the layers of cortex, they show up in distinct laminae, and just saying that the studies where more specific at layer 1 and more diffuse at layer 6, might also indicate that there are fewer Pyramidal Cells in Layer 6 than there are in layer 2 and 3.”
The banding patterns in, for example, frontal lobe regions (see such invasive work as Goldman-Rakic’s 80s/90s studies) brought out what appears to be the customisation by environment of genetically-determined interdigitations where such reflect expression of recursion at work.
e.g. the genetics can give (from general to particular)
LLLLLLLL-RRRRRRRR (l= left hemi link, R = right hemi link)
LLLLRRRR-LLLLRRRR
LLRRLLRR-LLRRLLRR
LRLRLRLR-LRLRLRLR ( interdigitation from IDEAL recursion and enviroment where frontal lobe regions show links from ipsi/contra-lateral hemispheres – links via CC.)Actual nature of last level is more such as:
LLRRRRLR-LLLRLRLR where environment that does not elicit responses allows for recruitment of neurons to serve in increasing the bandwidth of other neurons nearby.
This SAME form of pattern is manifest in the fight/flight dichotomy recursion as it is in the occipital lobe covering left and right fields of vision.
“I can’t comment on your assertion that the banding is also noticeable in the Amygdala, because I am only peripherally aware of that organ, and what it might do. Currently we are studying the memory, and the amygdala is considered part of the limbic system which influences but is not considered part of the memory system. When we get a little further into my theory, I will present information on what I think the connection might be, but for now, I am dealing with the first 4 layers of the cortex, and especially the first 4 layers of Allocortical Tissue that seldom has any more layers than that.”
I am surprised at this; emotion is a fundamental in colouring experiences and so encoding of such – to ignore limbic system feedback etc when considering encoding/decoding of memory is errornous IMHO. The development of the brain indicates local memory areas for each level of development and the extending of such to more refined frontal areas.
The invasive work on eliciting fight/flight behaviours through stimulation ‘along’ the path of interdigitation of the dichotomy in the amygdala is covered in papers in such as:
Gainotti, G., and Caltagirone, C., (eds) (1989) “Emotions and the Dual Brain” Springer-Verlag
…and I am sure you are aware of links of ‘old’ and ‘mid’ brain to neocortex frontal areas. In the development of the brain nothing is ‘thrown away’, refinements extend ‘over’ previous works and so continue to be exploited in aiding in information processing.
The organic element in recursion acts to link all of the elements derived from that recursion to give us a ‘regular network’ format (all is connected), and so potentials, that, when customised by local context elicit patterns we equate with ‘small world network’ development. These dynamics thus allow for local, concentrated, memory (close to/on the data entry points (dendrites etc) where such allows context to PUSH the life form and so conserve energy) but also a non-local links, some actualised, some not, across other areas of the brain. The benefits of the non-local links is in being able to access data by bypassing local damage etc and so work around the block etc.
The cortical layers cover 1 for the RAS, 3 for the limbic, 4 for the cingulate cortex and 6 for the neocortex. Focusing on the neocortex can cover specialist, precise, memories and the column dynamics with the thalamus but the other levels supply support as data is managed general to particular.
A common problem with too high precision in analysis is in missing the forest for the trees and so missing the feedback/feedforward dynamics spanning the particular/general. Thus focusing on, stimulating, the last level of interdigitations of fight/flight elicits, for example, fightXORflight but that is not the behaviour we see, we see a range of emotional responses and that indicates the columns in the fight/flight ‘matrix’ contribute as general levels setdown foundations for precision levels to function, it is not all mapped into the one level/layer.
If the genetics sets down an ideal network from recursion and environment then customises this network through actualisation so we get patterns that are in singular form are unique to the individual but also more general encodings of the patterns that cover memory management at the particular and general levels.
-
Dear Graeme,
I do not at all deny the usefulness of models or discussing them. I think that they are different than the “original” if like in the case of human memory there is an original.
In that case a model has to be evaluated how well it grasps the reality of the natural phenomenon it claims to be a model of.It is understandable that working on a model of memory you can move on to the mind, because there are links.
I encourage you to elaborate on that as well, hopefully in a different thread. I think memory is crucial and we need better understanding of it.
Yours friendly
Hans -
Dear Graeme,
you wrote:“Further, the Micro-Anatomy of the brain does not evenly distribute memory cells throughout the layers of cortex, they show up in distinct laminae, and just saying that the studies where more specific at layer 1 and more diffuse at layer 6, might also indicate that there are fewer Pyramidal Cells in Layer 6 than there are in layer 2 and 3.”
It would be great, if you would give references for this opinion. What are in your opinion “memory cells”?
Yours friendly
Hans -
“I think you are missing the point of this thread, and overgeneralizing at the same time. While no doubt staining studies of one type or another might show activity, it depends on the staining study, what patterns will emerge.”
your focus is too ‘trees’ – focus on what is implied by the banding/interdigitations. These patterns are derived from recursion and our whole being is grounded in a single cell that then goes through recursion to develop. Given that, so the properties of recursion will be manifest in the life form to varying degrees. My point is one of those properties covers the emergence or an organic format from a mechanistic format where the organic format covers linking of all elements of the recursion with each other.
The differentiating of data covers the categorisation of such into ‘wholes’ or ’
parts’ or ‘static/dynamic’ relationships and these with emotional colourings. This level of information is SINGLE context in that it covers an ‘emotional world’ – all is expressed as secondary+ harmonics of senses.Symbolisation is a level above this such that there is a hierarchy of meanings present and so of memories present; there is a slot for the emotional/cognitive experience and on into the symbolic levels of meaning. This indicates at least two levels of memory given the hierarchy of the neurology.
Thus base level memory covers sensations of wholeness, partness, etc and composites. Move up a level and we move from concrete experience to abstract experience and so REFINEMENTS of a memory where the generic universals then get tied to some unique context not the memory ‘in toto’. IOW the column contributations allow for differentiating across rows in a memory matrix.
-
Hi Chris, I think you have missed the trees for the forext. By that I mean, that you, like many here, think that memory is everything that remembers, and ignore the fine details like the fact that neurons have a particular shape if they take on a particular role. Or the fact that a particular memory neuron doesn’t just remember, it also processes the information it is remembering. What I am describing right now in this Thread is a specific arrangement of cells, and what that means to the way that memory works. While the layer 4 cells are probably hooked up to the limbic and attention systems they really have little effect on the way that Allocortical Tissues Remember, the Cloud of unformatted data that they produce.
We can argue about this if you want, but what is important to me, as the originator of this thread is to work through the stages of the memory slowly so that the structure can be understood. This isn’t the time to bring in the limbic system, that is all.
Results
-
