Cognition Theory v0.7

A foreword to the theory

It is important to note that the theory that will be described below and used as a basis to construct the third-order logicoscientific model has not been published to any scientific paper and thus cannot be mentioned as a reference and will unfortunately have to be described in full within this paper.

Also, it is an algorithmic theory, without any reference to actual biological functions and parts of the brain that implement the proposed mechanisms and algorithms. This is because it has been build from scratch, without the help of any preexisting theory, as it is the author’s belief that none of the limited theories that he is aware of have significant relevance with the concepts and direction of his own theory. Therefore you will not find any references to other published work in a similar field, simply because there have been used none in the process of making. The author’s own brain has been deemed enough, for now, for analysis and testing of all parts of the theory.

The above also means that the description that will follow may just as well exist only on paper and have no tangible connection with the “true” algorithms and functions of the inner workings of the human brain. Having said that though, since these “true” algorithms have yet to be discovered, the author believes that the methods mentioned in the theory to follow are able to explain a lot of the everyday actions and thoughts of the human brain and it is in the reader’s discretion to judge that for himself as well.

This theory is a work-in-progress for the last 6 years for the author and it is evolving day-by-day to encompass and be able to explain an even greater variety of human interactions (driven by the brain). There are some parts of the algorithm that are known to be erroneous (but for lack of a better alternative they are still kept as parts of the theory) and thus there are many human interactions that cannot be fully interpreted and explained through the use of this theory. But, on the other hand, there are also a lot of diverse and complex human interactions that can be thoroughly explained in a very simple and revealing manner, by the use of this theory. The analysis conducted on this paper over the narrative methods is a great example of this, since we will be able to translate many of the intricate features of a narrative approach into algorithmic function of the brain, explained by the theory in question.

The author is an engineer with a computer science major, who has worked as a computer programmer for several years. This can easily explain why the theory has been constructed and described in an algorithmic manner and why all parts of the theory can be modeled/emulated relatively easily with a computer algorithm. This means that, when the theory is deemed complete, it will be easily transferrable and implemented to a computer algorithm/system/processing unit; this also means that all parts of the theory can be considered as strict logico-scientific rules, because only those types of rules and predicates can be implemented in a computer.

Finally, this theory is not to be disclosed to anyone, without the author’s prior permission, because it is yet a work-in-progress, it has not been sent for publication in any journal (after all, there are no references, so there is no hope in that) and it has been put in words especially for the “In search of Phronesis” course, for Mr. Tsoukas’ eyes only.

A theory of cognition

The goals

Every system has a reason for operating and the brain must have as well. We start our analysis from this topic, because everything else we will describe will come to fit into the “master plan” as a necessary component to achieve the mentioned goals.

· The goals

o Main goal: Maximization of pleasure

§ The goal, of course, is the maximization of the probability of survival. But because our body gives us pleasure when we do anything that increases this likelihood (thank evolution for that), the ultimate goal of brain function is the maximization of pleasure.

o Secondary goal: learn from previous experience and apply that knowledge readily, until it proves that it needs revision

§ Resources are limited and by occupying critical components of our brain “machine” all the time with the same calculations is costly, not elegant and wasteful of resources. Only the leanest survives in nature and thus, a secondary goal is to learn from previous experiences, store that knowledge in a way that is easily retrievable when the need arises again and use it readily ever after, until it proves that (because the situation has changed) a revision must be made into this “situational solution” because the situation has changed and the solution does not fit the circumstances any more (i.e. it does not guarantee maximization of pleasure)

o Tertiary goal: use priorities in the allocation of resources and deal each moment with the most important task at hand

§ The needs of the mind and body might be various and diverse at any one moment, and also conflicting. The brain should try to prioritize his needs and because it can only deal with only one situation at any time (will be elaborated further on), it must deal with the most important one, and also have the ability to constantly reevaluate importances and switch tasks if the priorities change.

The building blocks

Now that we have established what the system must achieve, we turn our attention to the constituents of the system. Please keep in mind that this is an algorithmic description of the building blocks of the system, without any knowledge or direct reference to actual, tangible parts of the brain that perform these roles. Thus, each building block of the algorithm might correspond to one discrete brain organ or more, or even the interaction of many that creates an emergent behavior which then corresponds to a building block of the theory. Where the actual organ that corresponds to each building block can be guessed with some accuracy, it is mentioned below, else it is left unassigned for future biological research if or when the theory proves itself.

DIAGRAMMA

1. The stimuli endpoints

These are the input endpoints. All sensory information from the outer environment go through these points to be codified into neuron impulses and be forwarded into the brain.

2. The pattern matching mesh

This is a network of neuron interconnections that work similarly to a B-tree, i.e. they take codified information of the relatively few stimuli endpoints and spread it in the (relatively many more) pattern matching neurons by traversing the interconnected neuron paths in a preset and repeatable manner. The pattern matching network guarantees that:

o Similar stimuli will travel to similar destinations inside the brain

o Different stimuli will travel to different destinations

o Even stimuli that we are experiencing for the very first time will find their way into the mesh and will be pattern matched in the same way as every other stimulus, thus they will reach a destination close to the existing patterns that most closely match the new ones.

o The breadth of inputs is vastly increased from some thousands of stimuli endpoints to probably millions of neural network destinations, thus giving the brain repeatability (same inputs reach the same destination), pattern matching ability (similar inputs will “remind” the brain of the same things) and room to work with (because if the brain had only some thousands of inputs to work with, its computation capability would be severely limited; now this multiplication of inputs enhances the field of interaction)

3. The driving forces (DFs)

Our body actually commands the brain by manipulating our priorities through the driving forces. The driving forces represent the “needs” of our body and are activated in cases of hunger, sexual desire, pain, etc. The main job of the driving forces is to connect to pattern matching destinations that have proven related to the covering of the needs of each driving force and energize the driving pockets (explained below) in order to act upon the need.

4. The driving pockets

These are concentrations of energy («θύλακες ενεργότητας»), similar to the voltage of electrical circuits, that are being created with the help of the driving forces, either when the inputs coming from the pattern-matching mesh designate that there is potential pleasure (i.e. satisfaction of our active driving forces) in the outer environment or when the driving forces are activated by themselves due to an inner need of the organism (hunger for example, even when these is no visible food around us). The bigger the stimulus from the pattern-matching mesh, the bigger the activation/energy-concentration of the driving pocket. The bigger the current need of the organism (demonstrated by a very active driving force), the bigger the activation/energy-concentration of the driving pocket.

The driving pockets are reason for action of the brain. Since the brain consumes a lot of energy when working, every brain computation is an “investment” for a bigger probability of survival in the future. Thus, it must have a very good reason for acting, which is expressed by the activation of the driving pockets, which correspond to our inner needs (expressed by the driving forces) combined with the potential of the outer environment (expressed by the pattern-matching stimuli). I.e. the bigger the need (inner) and/or the bigger the potential (outer), the greater the reason for action, thus the greater the activation of the driving pocket.

In this way, the driving pockets help in the satisfaction of the tertiary goal of the system, the prioritization of pleasure opportunities, since they designate when there is much potential for pleasure and when not.

5. The Selector (in greek «Επιλογέας»)

The Selector constantly surveys all the active driving pockets and chooses to deal each time with the most active one. Thus, it is the subsystem responsible for the maximization of pleasure (primary goal), since it decides in each given moment which pleasure potential will be the most beneficial to allocate resources to and does just that. The way the selector works is very simple in terms of describing, but probably a bit complicated if you want to realize what effect its action have in the whole brain system. Thus, it will be described in greater detail later. For now, it is enough to say that the selector channels energy through the action mesh, trying to reach the action endpoints.

The selector is also our “inner eye” and our attention mechanism, because only the stimuli that go through the selector are “noticed” by the person’s brain and can be remembered. It is also our gateway to our feelings, emotions and pleasure, because only the stimuli that go through the selector can activate our emotions, give us pleasure and create feedback (strengthening of the neural connections). Finally, since feedback is the method of creating memory, only what goes through the selector can be remembered. You have probably realized up until now that the Selector is the orchestrator of the most important aspects of our brain function and the most critical part of the brain’s algorithm, as with the loss of it we lose consciousness and the ability to learn. It is also the “single point of failure” of the whole brain, because it is the only part of the brain that is unique and irreplaceable.

6. The action mesh

This is a network a neurons, heavily interconnected between them, with two directions for the flow of the electrical impulses. The one direction is to the action endpoints, in order to translate the stimuli and the needs of the organism to meaningful action that will alter the environment and provide pleasure. The other direction is an inner loop back to the destinations of the pattern-matching mesh and the area of the driving pockets. Thus, when the Selectors directs a flow of energy/electricity towards the action mesh, some of the current flows towards the action endpoints and some loops back towards the pattern-matching destinations. The amount of energy that will be directed in each path depends on the interconnections that have already been built inside the action mesh in the previous appearances of similar patterns. This dual direction creates the inner-outer loops that will be described later and are of the essence for the emergence of thought and the learning process.

7. The action endpoints

These are the neuron endpoints that transport electrical impulses to the body muscles, in order to act on our opportunities for pleasure.

The mechanisms

Now that we know the main constituents of the algorithm, we can discuss the dynamics of their interaction, in order to understand how noesis emerges.

· How the traversal of neural paths works: A signal that crosses from neuron to neuron, at an intersection to multiple paths will follow these rules:

o The stronger the signal that tries to pass through, the more energy will pass through each alternative path (similar to conservation of energy)

o The stronger the bond between the two neurons (i.e. the thicker their interconnection) the more signal will go through this specific interconnection.

o The more recent the activation of an interconnection between two neurons, the easier it is for a new signal to go through it, thus more of the signal will flow through this direction.

§ Think of the activation of the neuron as a tuning fork that is agitated once. It vibrates for quite some time while dissipating the remaining energy. The same probably holds true for the neurons as well. The more “vibration” the neuron has left from a previous activation, the easier it is for a new signal to pass through.

· The axe effect: When you try to chop down a tree with an axe, you can’t do it in one blow. The force needed is immense. But if you strike multiple, repeated blows to the same spot, each blow goes further inside the trunk, by taking advantage the fact that the precious blow cleared some room from the inside of the trunk. The same thing happens to the neurons and it is a vital part of the way we think! Now that we have explained how the traversal of the neural paths works, we can explain it:

o When the first pass of a signal through the neurons tries to traverse our neural network, if there is no strong preexisting path (thick interconnections that form a path) to guide it through the other end (we’ll explain what this is later), then it will be distributed to many diverse connections and will dissipate quickly, unable to make enough “hops” from node to node.

o If some mechanism (call me Selector) continues to send signals through that direction, then these paths will be already activated by the previous signal and, as we already know, an activated path is easier to traverse and thus the next signal will dissipate a little further on where the previous signal dissipated.

o This “little further on” might be enough to guide it into “familiar territory” where the paths to the other side are strong now and it can flow from one end to the other without dissipating in the middle.

§ This is why when we think of something unfamiliar, we can come up with a solution some of the times. If the situation at hand is not vastly unfamiliar to us, then some traversal iterations, fueled by our Selector, can bring the unfamiliar signals close to familiar ones and help us proceed to action. Notice also that we need the help of the Selector to think and make progress on our thinking, and we already know that the selector only deals with one thing at a time (the most important one), and that is why the brain can process only one thing at a time (more on this later).

· Feedback and the creation of memories: Feedback is the strengthening of the connections between neurons. Memories in the brain are strengthened neural paths. It is a very difficult concept to grasp, and an even more difficult to describe, but I am very confident that it is very close to the truth, and thus it is used throughout the theory. The bottom line is that feedback creates memories. And the reason to create memories is twofold:

o Remember the good, in order to do it again in the future

o Remember the bad, in order to avoid it in the future

These two goals are translated into two mechanisms of feedback:

o Feedback in the pattern-matching mesh, when driving forces are activated. This will help us remember all things that have to do with an activation of a need of our organism and directly relate to this need when we see all kinds of similar patterns, both good and bad. I.e. when we feel pain and the driving force of pain activates, we will associate the pattern of the needle that just stung us with this driving force and will immediately recall this pain sensation (and thus create a driving pocket out of it) when all patterns similar to a needle fall into our attention. Moreover, the greater the pain, the greater the feedback, the greater the association/memory.

o Feedback in the action mesh, when driving forces are deactivated. When a driving force (such as that of hunger) deactivates, it means that we have just done something beneficial to our survival. As we will describe soon, the action mesh is the tool with which our brain creates thoughts and actions, thus any thought or action that resulted in the deactivation of a driving force must be remembered and the neural interconnections that were used must be strengthened. This helps us learn thoughts and actions that enhance our survival.

Something that was stated between the lines and must be expressed clearly is that, when feedback occurs, ALL at the time active neural connections are strengthened, with a multiplication factor analogous to the amount of activation each interconnection still had at that moment (…remember the tuning fork example, a traversal of a neural path leaves the path on a gradual downgrading “vibration”) and the amount of the activation/deactivation of the driving force (so a big deactivation will also lead to strong feedback and strong memories of the fact). The fact that all active connections are strengthened is very simple for the brain to implement, because on any other way the brain would have to judge which connections are the good ones (and thus must be strengthened) and which are irrelevant and happened to be active at the time. Since this is a tough call to make, all active connections are strengthened, and this for example also explains why we have associated some songs with some persons or places; we had a good time in the past being in that place and hearing that song and ALL of those patterns were strengthened at once, thus when we hear that song again we have strong connections to the other patterns that were active at that time as well and all these come back to our memory as well.

· Mechanism for actions and learning – the inner/outer loops – the “snail”: This is the most important mechanism of the human brain. It describes the way we act, the way we receive pleasure and the way we learn. And it all revolves around the Selector and its only task of “getting the signal to the other side”. Hang tight, because this is a difficult concept to explain in words:

o As we have already said, when an action pocket is created, there is a reason for the brain to react in order to gain pleasure (or to reduce discomfort, if it is a driving force such as pain; nevertheless) and increase our chances of survival.

o When we have no previous knowledge of what to do to satisfy our driving pocket (think of a hungry infant trying to feed), the selector “fires” constant bolts to the action mesh, hoping to reach the other side (the action endpoints). Some of them do reach the other side (probably with the help of the axe effect), some loop back into the pattern-matching destination points.

o With a lot of effort, with chance and with the help of our surroundings (e.g. parents), we manage to satisfy our DFs. This is create feedback throughtout the action mesh and strengthen all the connections that were employed at the time of receiving pleasure.

o This feedback will also make it more likely the next time that we will have the same need and the same driving pocket that we will use the existing connections and reach our target sooner than before. The paths are paved better now, the signals find less resistance and manage to travel to the action endpoints (and be translated into movement) easier. The Selector does not need to be occupied fully into this task; some of the signals will flow directly from the driving pocket to the action endpoints.

o Furthermore, because there is the “inner loop”, except from the paths to the action endpoints that will be strengthened, some other paths from the action mesh to the destinations of the pattern-matching mesh will be strengthened as well.

o This strengthening of the inner loop is, in essence, an expectation of the pattern to come when we fire those signals towards the action endpoints. I.e. we embark onto making an action, but because we have made the same action several other times and it was good and it created feedback and it was binded with the patterns that followed that action… we have now come to expect what the next pattern that will in the pattern matching will be! We do something and we almost see how our action will change our surroundings and what new patterns it will generate.

o If the pattern that we expect to arrive from the outer environment through the pattern matching as a consequence of our action does happen, then everything continues as normal, the signal is again transmitted through the action mesh, generating another action that will alter our environment a bit more, generating new (but expected) patterns that will drive new actions etc.

o In this way we can execute a series of actions where one triggers the other (by generating the patterns that will lead us to the next action), because we have learned to execute them in sequence. And if all these paths that are contained in the “outer loop” are strong enough (because we have executed the same again and again with success), then the Selector does not need to be involved at all! He is needed only at the beginning, to recognize that this action pocket is the most active at the time and start the loop, and then the whole sequence of actions is self-fueled, IF the patterns generated by the inner loop always match the actual patterns that arrive from the outer loop as the consequences of our actions.

o When these two sets of patterns don’t match, we will either stop executing or ask the help of the Selector again, to deal with the uncertainty, via its known mechanism.

o All these inner and outer loops that continue ad infinitum and the one fuels the other, I call them “the snail”.

This process can gradually bring us from 0% knowledge on doing something to 100% knowledge on that. At first, we have the full, undivided attention of the Selector, which tries to push through the signals to the other side (and by the mechanism of thought that we will discuss in a few moments). When we succeed even just a bit, we get some pleasure and create some feedback. The next time, it will require a little less attention from the Selector to get the job done. This is looped until we are fully capable of executing the task without the selector helping almost at all.

This explains why, for example, we can drive to work without even realizing how we got there and not remember a single bit of the route till there (remember that only through the selector we create memories; if it is not involved, we do not create new memories). But if someone tries to collide with us, the unexpected patterns will break the loop, create a strong driving pocket (by the activation of the pain/fear driving force) which will be greater in activation than the driving force that was creating this loop up until know, and now the Selector will focus immediately on this new need and try to handle it the best way possible.

The mechanism of learning can also be said to resemble the way an apprentice at a craft learns gradually from the master. At first the apprentice will focus on doing the simplest tasks possible, the ones that he already more or less knows. He will also need a lot of attention from the master craftsman (the Selector in our case). Gradually, as he learns from his rights and his wrongs (feedback), we will require less attention from the master, until some day he knows just about everything and needs no attention at all.

This is the way our brain handles multitasking! We cannot deal with more than one unknown situations at a single moment. We have only one Selector. But if the situation is familiar, we need only decide on the action (have our selector deal with the situation for a split-second to initiate the loop) and then the process continues by itself, allowing us to deal with other driving pockets of lesser importance. This is multitasking, and the most common way we use it is to combine action with thoughts…

· The mechanics of thinking: Thought is actually very similar to action; it uses the same components and produces new patterns in the pattern-matching mesh, but by using only the inner loop, instead of producing the new patterns via the action that manipulates our environment and changes the patterns that we experience. Let us elaborate more on this:

o As we have already said, the inner loop is a way for the Selector to feed the results of its activity back into the pattern-matching area. This is effect is like making us see new patterns. The new patterns are activated through the current that flows via the inner loop and we experience it as if they were activated from the environment.

o In essence, the inner loop brings into our attention (because they all end up in the selector, which the attention directing tool of our brain) new patterns: images, sounds, memories, etc. It is not much different in the way we experience it, it is only of lower intensity.

o Thus, as you can already imagine, the inner loop is actually our method of thinking! This translates to the following facts:

§ Patterns activated by thinking are experienced in the same way as patterns activated by the outside environment. In other words, we “live” our thoughts: they are no different from the real world.

§ Thoughts can activate driving pockets, just as well as patterns from the environment do. This means that, while we are doing something, the inner loop might activate some pattern that will “remind” us of something more important, activate a driving pocket, “steal” the attention of the Selector and now the Selector will deal with our new thought.

§ Since thoughts are weaker signals than signals from the environment, they create less active driving pockets. This has a twofold consequence: when something important is going on in the real world (e.g. danger for our lives), we cannot think, because the driving pockets created by the real patterns are much stronger than the “weak” driving pockets that are created by our inner loop thinking process. It also means that these driving pockets get deactivated easily enough as well (fleeting thoughts!); if the selector does not deal with them right away, we might have just as well forgotten it the next moment!

§ But this easy deactivation of the driving pockets of thought has an added advantage: it enables us to make associations («συνειρμούς»). The mechanics are like this: in an non externally demanding environment, a thought creates a driving pocket, the selector has nothing better to do, so starts dealing with this weak pocket, sending signals into the action mesh and back to the pattern-matching through the inner loop. And while this weak driving pocket gets deactivated quickly, a new pattern (thought) that was the result of the previous loop activates a new driving pocket and since it is somewhat more active than the previous one, the selector now drops the previous one and deals with the new one. This fuels the thought process even more and enables us to move from one pattern/thought to the next, by creating weak driving pockets that “stay alive” just enough to drive us into a new path of thinking.

o When these inner loops of the thought process, along with the “axe effect” of the Selector, happen to reach familiar patterns with strong connections to the action endpoints, the inner loop ends and is transformed into an outer loop! In this view, thoughts and actions are complementary. Thoughts are a preparation for the actions, some kind of refinement, before actually enacting on your needs. They are also very important for the learning process, because as we have already said learning can occur even with thoughts, since they produce similar patterns with the outside environment. Nevertheless, they are (from a survival point of view) an intermediary in the process of acting, because only action in the end will ensure our survival. But action well-thought of will increase our survival probabilities even further, and this is why it has been developed as a process throughout evolution.

· Escaping the vicious circle of self-amplification: Have you ever thought why everything that brings us pleasure every time, will bring us less pleasure next time, until it finally stops to entertain us and it virtually goes unnoticed? This is a defense mechanism of our brain, to avoid a vicious circle of self-amplification, and we now have the tools to explain it:

o When a driving pocket is formed, the Selector (which is also the gateway to pleasure, as we have already said) handles it and, if successful, an action will be executed that will deactivate the driving force, bring pleasure and strengthen all the interconnections from the driving pocket till the action endpoints.

o The next time the same driving pocket activates, the road to pleasure will be easier, as the interconnections are better paved now. If this was the case, then we would get easier and more pleasure, forcing us into a spiral of increased willingness to execute such an activity to get pleasure and increased aptitude in execute that activity and actually getting pleasure.

o Sooner rather than later, this activity would be the only thing that we would be willing to do (all other things would promise less pleasure and thus would be rejected by the selector). I.e. the self-amplification of the interconnections relating to this activity would lock us in into this activity and never release the grip, until we die of starvation or exhaustion.

o Fortunately our brains have a self-defense mechanism to such situations. Since the Selector is the only gateway to pleasure (and feedback and the strengthening of neural connections), when we increase our aptitude in an activity we have already mentioned that the Selector is called for help less and less; thus the pleasure we can extract for an activity is gradually decreased, the more we exercise it.

o In fact, since the selector is also our attention mechanism, the patterns that this activity energizes are redirected less to the selector, which means that it gradually starts to go unnoticed! Thus, our defense mechanism to self-destruction is also a curse of noticing and liking less and less the things that have pleased us the most…

· Communion/Consonance/Empathy («ταύτιση») as an extension of our motives: Since all patterns that give us pleasure eventually result in the strengthening of the bonds between our driving forces and the patterns, this provides a way for a person to extend his motives/patterns that energize his driving pockets, with this mechanism:

o The first patterns that will provide pleasure and bind with the driving forces will create an “inner layer” of patterns that the person most closely relates to his need and to the probability of pleasure.

o Other patterns that are similar to the first ones, but not entirely identical can now utilize this strong path and reach the driving force without loss of signal strength, and thus agitate the driving force as well and create driving pockets. These driving pockets will cause us to act in order to deactivate them and will probably produce some pleasure as well, thus creating a second layer on top of the previous.

o By continuing this process, a person builds throughout his life a series of layers of patterns that relate (other more other less) to the satisfaction of his needs (which are expressed by the driving forces).

o And since the first patterns that a person will see when he satisfies his needs are closely related to his own self, we arrive at the conclusion that in the inner layer of motives are thing relating to oneself and gradually as you move to the outer layers, you can find patterns/motives that are close to the previous one pattern-matching-wise. Thus a person gradually cares in his life about things that are related to him and considers them an extension of himself (because this outer patterns are connected to the inner/self patterns that are directly connected with the driving forces).

o The better the pattern matching ability of a brain (bigger brain, better structured, more interconnection capabilities), the more diverse the array of patterns that can be connected to our inner motives, thus the more empathy we can feel for other things but ourselves.

o On the other hand, the more layers a signal has to traverse from an outer layer till the inner layers and the driving force, the bigger the chance that the signal will dissipate and fail to reach its destination. This means that there is also a limit to our growing empathy.

On a side note, many of the things that a modern human thinks and acts upon are based on this concept of empathy. Since all the major (bodily) driving forces are usually kept satisfied in the modern world (like hunger and usually sexual desire), the things that actually do agitate the driving forces and make us act are usually related to motives/patterns that reside on the outer layers and they owe our empathy to them on our exceptional pattern-matching ability as humans and the pleasure that these (or similar) patterns have given us in the past.

Thus, a foreign child suffering can cause us to act as if this has been our child, if this child relates to previous patterns that have given us pleasure in the past (e.g. if it is same-colored as our child or looks a lot like it; that’s why in advertisements depicting pictures they try to put persons that look a lot like their target audience, in order to increase empathy and the motive to act).