Monday, December 17, 2018

Long term sensory memories in TGD framework

There was a highly interesting popular article (see this) inspired by the recent findings about long term memory (see this) in conflict with the standard view about memories. Of course, also the memory feats of so called idiot savants known for decades are in sharp conflict with the standard view about memory.

The discussion of these findings in TGD framework led to a decisive improvement in the understanding of the proposed mechanism of sensory memory recall. Also a connection with the model of topological quantum computation realized axon-microtubule level emerged. Sensory memory would be realized as a topological quantum computer program running in reversed time direction in memory recall and generating the virtual sensory input from brain to sensory organs creating the original sensory experience.

The findings

The following gives a brief summary of the results of the experiment discussed here.

  1. A huge amount of storage capacity is required and it increases as more and more experiences are experienced.

    One can imagine an abstraction as a cure: store only essentials about the input. This is extremely powerful manner to store the relevant information. Picture about grandmother's house with all detail is replaced with world "grandmother's house". What is lost is detail. This storage mechanism is certainly used at higher levels of evolutionary hierarchy. Verbal memories are a good example.

    The experiment mentioned above however demonstrates that the memory storage is at least 1000 times more detailed than it could be, which suggest that a different very detailed storage mechanism usually unconscious to us is involved.

    Indeed, the memory feats of idiot savants show that sensory percepts can be stored in amazing detail. A possible TGD based explanation is that all of us have sensory memories - essentially re-experiences but at a lower level of personal self hierarchy, not as mental images represented as sub-selves but as sub-sub-..-selves not directly conscious to us. Temporal democracy would make it impossible to distinguish between recent and past and make it difficult to survive. Here would be the reason for why these persons are often called idiot savants.

    Sensory memories must be unconscious at our level of self hierarchy to allow the experience about living in definite moment of time and only cognitive (symbolic, verbal) memories involving a lot of abstraction satisfy this condition. If the percept is cognitive, it is about geometric past. If sensory, it is about "Now". Perceptive field effectively reduces from 4-D to 3-D (actually the duration of sensory chronon is about .1 seconds).

    Situation changes when temporal lobes are stimulated electrically as neuroscientists have known for decades but "forgotten". Perhaps animals do not conceptualize and have sensory memories.

  2. Proteins used for the storage in terms of modified synaptic contacts is slow by a factor 1000 slower than required to understand the above experiment. Memorizing would require a repeated stimulation but now the pictures were seen only once or twice.

  3. The lifetime of the proteins in synaptic contacts is only few weeks so that also long term memories would be unstable. Humans can remember for about 50 years, 1000 times longer than expected.

  4. The technical realization of the 3-D storage is also a problem. One should remember also the place, where the memory is stored, not only the memory itself! Here the association mechanism seems the only possibility but would allow only conditionings. In computer language LISP this idea is very concretely realized. Conditionings are however only pseudo-memories.

Wrong views about time andthe notion of memory as the basic problems

To sum up, the standard view about memories suffers from two fatal problems.

  1. The first fatal problem of the standard model of memory is the wrong view about the relationship between experienced and geometric time. The identification of these times forces to the notion of memory storage analogous to that in computer. The information about what happened must be stored again and again. This view has many problems already discussed.

  2. Second fatal problem is the conceptual flaw forced by behaviorism: memories are identified as conditiongs, habits, or behaviors - as you like. Genuine sensory memories are however re-experiences and would however correspond to re-experience to which is associated a synchronously firing neuron group: what neurons fire is not determined by synaptic contacts but by the sensory input mapped topographically to sensory area. This is very delicate and crucial difference.

TGD view about sensory memories

Could one realize memory as re-experience in TGD framework?

  1. In zero energy ontology (ZEO) of TGD no 3-D memory storage to the "brain now" is required. Memories are ideally where (in 4-D sense) the event occurred but memory recall creates further - usually less detailed and more abstracted copies - of the memory. To remember (in the genuine sense of the word) is to re-experience. Memory in this sense would be in the geometric past. Memory recall would be seeing in time: sending a signal to geometric past, where it is time-reflected back. Each memory recall could generate at least conceptual copy about the memory and in this manner the signal sent to the geometric past would have higher probability to generate the re-experience or at least secondary version of it. Learning, which is not mere conditioning, could rely on the generation of copies of the memory in 4-D perceptive field.

  2. Memories as re-experiences would involve synchronously firing neuron groups associated with quantum coherent units defined by magnetic bodies (MBs) of neurons and representing mental images. To understand this concretely, one needs besides the notion of MB also the hierarchy heff= n× h0, h=6×h0 of Planck constants. The synchronously firing neuron group (involving quantum coherent part of MB) in the geometric past is woken up by the time reversed signal to the geometric past and reflecting from it by providing energy (now negative). ZEO makes this possible.

  3. How the memory recall could realize this synchronous firing in the geometric past? This mechanism should be analogous to the reflection of negative energy signal in time direction from the brain of the geometric past. ZEO allows sending of a negative energy signal travelling to geometric past. It should somehow induce a transition generating the synchronous firing. The signal generating this transition should be very simple. It must induce the transition at correct location in the geometric past. Here the period of the carrier wave of the signal could be essential and large value of heff could make the signal energetic enough despite the period which could be measured in years so that energy for the ordinary value of Planck constant would be extremely small. Signal could also provide metabolic energy for the neurons, which should fire synchronously. Replicas of the memory help to achieve activation at the correct location.

  4. There must be a coding of the sensory input to the physical state of neuronal pathways coded by nerve pulse patterns representing the original sensory input from the sensory organs. If genuine sensory re-experience is required a signal generating the original sensory experience and thus the nerve pulse pattern from sensory organs creating it should be re-generated.

    As if one had in the geometric past a magnetic tape representing somehow the original experience. When played it would generate a signal to the sensory organs in turn generating the signal to the brain (including nerve pulses) giving rise to the original sensory experience. Note that ZEO indeed allows the sensory experience to be in geometric past. It is however communicate cognitive information about it to recent too.

TGD leads to a model for what could happen based on the idea that topological computation is realized in terms of the braiding of magnetic flux tubes connecting two subsystems (see this and this). This model leads to a model of memory representations as a kind of topological quantum computer program giving the original experience as an output while running.

Let us assume that second system is axonal membrane along which the nerve pulse patterns (and whatever else is needed) representing the sensory input flow. Second system would be naturally microtubules inside it.

  1. The flux tubes would connect the lipids of the axonal membrane to the tubulins (or units formed by them). Axonal membrane can be in liquid-crystal state meaning that the lipid are like liquid particles able to move. Nerve pulses would induce a 2-D liquid flow inducing the braiding of the flux tubes having second end fixed to (say) tubulin of the microtubule.

    There would be both time-like and space-like braiding. Dance metaphor is very helpful here. Consider dancers at the parquet with legs connected by threads (flux tubes) to a wall (microtubule). Time-like braiding would correspond to the dynamical dance pattern of lipids in time direction having a representation as a 2-D projection defined by the paths of dancers at the parquet. Time like braiding would be analogous to a running topological quantum computer program.

    Space-like braiding would be the outcome of the dance representing tangle of the flux tubes fixed to the wall and defining topological quantum computer program serving as a representation for the time like braiding and therefore also for the nerve pulse pattern (and whatever the signal involves) and the sensory input. Space-like braiding is analogous to the code representing the topological quantum computer program and should make possible to represent the program.

    If this space-like braiding can generate a signal serving as a virtual sensory input to the sensory organs, the sensory memory could be regenerated. The running of the topological quantum computer program would mean the opening/un-knotting of this braiding and would represent the time reversal of the sensory input, not yet sensory input, which could correspond to nerve pulse pattern from the sensory organs generating the sensory percept. It seems that the opening must generate a signal to sensory organs as virtual sensory input.

  2. Virtual sensory input brain indeed is the basic element of TGD inspired model of sensory perception as construction of artwork. The basic difference to the standard view is that the sensory qualia are at the level of sensory organs rather than in brain. Brain only gives names for the percepts and builds standard sensory mental images by using virtual sensory input from brain. The process is like pattern recognition by driving sensory input to a standard input near to the real input.

    In TGD framework however nerve pulse patterns would not carry the sensory information to the brain but would generate sensory input to MB as Josephson radiation from the cell membrane. The transmitters emitted at the synaptic contacts would generate bridges connecting axonal magnetic flux tubes to longer connected flux tubes and in this manner create the communication channels - kind of wave guides. Along thee dark photons (which can transform to bio-photons) could travel with light velocity.

    This communication mechanism is dramatically faster than the communication by nerve pulses and allows forth-and-back signalling involving virtual sensory input from brain to generate the standard percepts assignable to the synchronously firing neuron groups accompanied by magnetic bodies obtained by connecting neuronal magnetic bodies by flux tubes.

    The standard mental images would contain only the features relevant for survival or otherwise interesting. A still open question is whether the virtual sensory input corresponds to the time reversal of the ordinary sensory input (see this). The following consideration suggests that time reversal is indeed in question.

  3. If the virtual sensory input from brain is in time reversed time direction, one can think of very simple model for memory as re-experience. Big state function could occur meaning that the mental images associated with braiding generated by nerve pulse pattern and dark photon beam die and re-incarnate in opposite time direction. A time-reversed mental image is generated. This mental images is not conscious at our level of hierarchy living in opposite time direction.

    This mental image is not quite exact time reversal of the original and there is non-determinism of state function reduction involved. One can have however statistical determinism possible if large enough number of neurons are involved. Therefore the differences need not be too big. Also standardization comes in rescue: it would take care that the sensory mental is very nearly the counterpart of the original.

    The time-reversed signal from brain to the sensory organ should generate a nerve pulse pattern just as in the case of ordinary perception and the dark photon signal generating the sensory mental image defining the original sensory memory in good approximation.

  4. For the simplest alternative dark photons alone induce the flow of the lipids. Hitherto it has been assumed that the flow is induced by nerve pulse patterns. The most general option is that both are involved in the generation of the flow. One cannot exclude the possibility that the communication of data about nerve pulse pattern to MB generates a control signal which induces the liquid flow. There are many options to consider but the basic idea is clear and involves ZEO and MB in a crucial manner.

  5. An important open question is whether the virtual sensory input using dark photons propagates

    1. to the "sensory organs then" so that only cognitive memories would result as copies. In this case a person, who has lost eyesight during lifetime could have visual memories from time when she could see.

    2. or via the MB to the "sensory organs now" and stimulates sensory experience in "brain now". Person lost eyes during lifetime could not have visual sensory memories in this case.

    For the latter option one can ask whether the sensory experience is
    1. realized by the mere virtual sensory input to sensory organs. No copies of the sensory representation at the microtubule-axon level would be generated. If sensory organs are not intact, sensory memories would not be possible.

    2. or whether also a signal from sensory organs to brain involving nerve pulse pattern is needed to generate the experience. Each memory recall would create an almost exact copy of topological computer program giving rise to a genuine sensory memory while running.

    Various options might be tested by electric stimulation of the temporal lobes known to generate sensory memories.
See the chapter Sensory Perception and Motor Action as Time Reversals of Each Other: a Royal Road to the Understanding of Other Minds? or the article with the same title.

For a summary of earlier postings see Latest progress in TGD.

Articles and other material related to TGD.

No comments: