Seamus McGrenery:
currently living and working in Dublin Ireland

Preconscious thought and the emergence of mind

How the body’s cells needs are met by circuits for preconscious thought, and how these circuits combine to create a conscious mind—one which can be mechanically modelled

The question of why brain processes should give rise to consciousness feelings has long troubled philosophers and scientists. But what if instead of looking at phenomenal consciousness as a hard problem we see at it as one key to understanding how our brains work. And I assume that, just as hearts evolved because body cells need oxygen, our brains are there to ensure that the cells that make up our bodies get what they need.

Science has shown that many of the features we are conscious of only represent reality they are not exactly what is physically present in reality. There are waves in energy and we experience colour, pressure waves move the air and we hear sound, plants have chemical composition yet we experience taste. We now understand that there is a microscopic world, invisible to us. The world we are conscious of is scaled, both in size and time to match our bodies’ capabilities. In short we know that consciousness is nothing like looking out through window-like eyes.

There are many suggestions about what consciousness is. What has so far been lacking, I believe, is a plausible description of how preconscious thought is logically organised, and why there is a need for conscious experiences. Also so far lacking is a description of a mechanism capable of experiencing what we recognise as consciousness. This account attempts to provide those answers by looking at the goals of data processing in the brain and the types of data processed.

I do not offer a computer analogy, for evolution did not start with a computer-like brain. Even if one had been available off the shelf it would be hampered by the frame and halting problems. We should go back to early evolution for clues to understanding how our brains work.

Differences between complex animals and other organisms without brains are worth examining. One striking difference is the way they take in nutrients. Most animal bodies are basically tubes for taking in bulk chemicals, pre-processed by other species. This allows more efficient collection of nutrients than say, plants which take in small amounts of raw materials along much of their surface.

Animals’ greater efficiency at feeding can result in them needing to move as immediately available nutrients are exhausted. Efficient eating, and movement, comes with the risk that fatal amounts of toxins can also be ingested quickly. To cope with this danger early animal evolution must have included a simple response to stop toxic chemicals getting into the body.

Reflex reactions for opening or closing the mouth, as well as moving towards or away from objects would have needed communication between cells that detect nutrients, as well as those that detect toxins, and the cells that can expand or contract to create movement. That simple circuits for this communication were the beginnings for animal brains seems to be evidenced by the brains development close to the mouth.

Very early animal brains likely had circuits for acquiring food and avoiding poison. Sharing data between these circuits does not require consciousness. It is all of one class; direct detection of a chemical triggering muscle movement.

As early animals began to move about an ability to detect food at a distance evolved. Using patterns of light falling on the skin, for example, as a source of information can greatly increase survival chances. But light contains no direct information on the chemical content of its source, so patterns of association have to be used. For example we can associate a red circular pattern of light on our retina with the body acquiring sugars. This is a preconscious part of our thought about an apple.

It is difficult for brain cells to work out that light means there is an apple. The system that detects the apple must be able to find the shape and colour at different angles and distances. Having decided there is an apple it would be wasteful to constantly repeat the task of finding. It is easier to store its position and keep checking that it is still there.

The colour and shape of light detected can change greatly as the animal moves, as shadows are cast on the apple or as a cloud crosses the sun. Using a stored icon or canonical shape for an apple can make both apple detection and checking of the stored position easier, provided that it is sufficiently flexible to account for variations in shape and colour. The stored icon for an apple can also be used to bind information about it being a source of sugars. The association between the shape and the sugar can be stored by part of the food circuit.

Thinking about parts of brain function as a logical circuit allows us to grasp how connections between very different physical activities—a pattern of light falling on some cells, a series of cells following patterns of contracting and expansion, some cells releasing enzymes while others absorb chemicals—result in complex actions. Toad prey-catching behaviour; a series of reflex moves in response to a visual stimulus is an example of the food circuit in action.

Toads have reflex behaviours for approaching mates and avoiding danger, as well as prey catching. We can also think of these as being operated by circuits. When there is just one stimulus the appropriate circuit can respond. When there is more than one stimulus simple rules can determine which circuit is able to trigger behaviour. For example, in toads, prey catching and threat avoidance is suppressed during the mating season.

Over the millennia evolution has favoured animals which developed more complex brains. In addition to circuits to acquire chemicals to keep internal processes going and to avoid toxins, mammals have circuits to avoid damage, to maintain homeostasis, to compete with rivals, to pass their genetic code on to a new generation, to support other copies of their genetic code and to cope with the planets cycles like day and night.

Much of our life is shaped by the preconscious thoughts of these eight circuits: Our days begin and end with food and hygiene. We sleep in dwellings that offer safety and an ideal temperature. Our work and leisure is filled with competition. For many of us raising healthy children and grandchildren is what gives purpose to our lives.

The tasks carried out by these circuits shape so much of the fabric of our lives that we rarely question how we think about them. Yet when we do focus on these specific tasks it becomes possible, and at times easy, to observe a preconscious circuit in action in ourselves or others. One simple example; sitting on a park bench I had a sense of the need to look left it was a danger alert. On looking I saw a runner heading past just behind me. In a park in sunlight the risk of danger from the runner was very low. For my ancestors tens of thousands of years ago that prompt to look, from a preconscious circuit using information from the edge of my retina, could mean the difference between life and death if the movement was a lion.

The circuit does not need to know, in the common sense of the word, that there might be danger. Evolution simply favoured animals that associate that type of stimulus in a way that is appropriate for a potential threat. This is one of a great many associations between information detected and responses by the body which have accumulated over millions of years of evolution.

Different circuits store different information about the same stimulus. In the example of the apple, the danger circuit needs to associate the loud crunch sound of biting into the apple with it not being a threat, while the toxin circuit associates the absence of the crunch with evidence that the apple is bad. So information needs to be shared across the circuits.

For social animals like us an apple can have many associations in many circuits. It can be food for us, our offspring, or a rival. Taking it could be a challenge to a dominant animal or it could let us repay a social obligation. These single pieces of information in isolation are of some value, but where survival is at stake what’s needed is a more complete picture—a way of using all the available information to make a decision.

Combining the information from all circuits allows a model of the world surrounding the animal to be built. In visual animals it is likely that the foundation for the model is part of the damage avoidance circuit used for simulating the rotation of three dimensional objects close to the animal. While damage avoidance is mapping the position of the branches close to the face the food circuit is drawing attention to the apples hanging on them.

Details about the animal itself are a critical part of the model. Knowing the speed of a lion is only of use if you know how fast you can run to time an escape.

Things happen suddenly when animals interact with their environment. Responding with a reflex or pre-programmed pattern of muscle actions as soon as an appropriate stimulus is detected can improve the species chances of survival. While the task specific circuits can work by passing control to one circuit at a time—all thought focused on escape from immediate danger for example—this does not work well in complex situations. Rivalry where control switches from one to another and back again is of limited use.

When no pre-programmed or reflex response is available or appropriate what is needed is a whole animal perspective from which to make decisions, and a simplified set of data to work on.

This is what consciousness provides, a running model of us and our surroundings. When we are awake this model is running. The models function is to provide just enough information to a circuit that can make decisions at the level of the animal. A decision making conscious self is better equipped to meet novel challenges.

The decision making processor working on behalf of the whole animal is closely identified with the representation of the animal within the model. The self that is aware is the self within the model. Only part of your brain is concerned with you as a whole. Most of your brain is getting on with looking after the needs of the cells in your body.

Take the analogy of a large business, there are people ordering supplies, hiring staff, making and selling products, and departments in charge of each of these functions. The Chief Executive Officer only sees the overall figures. They do not have access to the fine grained data on which products were sold in specific stores at specific times. The summarising of data for a CEO makes sense because providing all the detail would swamp them and make timely effective decisions impossible.

Only a simplified summary of data which have passed a threshold for positive identification are included in our conscious model worlds.

We know it is a model because it has many features which we can show only represent reality such as sound, colour and taste. Synesthesia is a condition where the senses are blended causing some people to see a sound in colour or to taste a word. The fact that conscious sensations can be caused by the ‘wrong’ input shows that they are only representations.

Daniel Dennett has pointed out that there is nothing intrinsically sweet about sugars, so why do we consciously experience it? The extra layer of data, sweetness, is added so that it can be compared with data from other circuits.

Because the data processed by the task specific circuits is so different in kind—food, threat, mate, rest—it has to be shared in a common language of simple choice based options. Each element fed to the model by the circuits has a value on a scale; either positive-neutral-negative, or too little-optimum-too much. Sweetness is simply one form of simplified data for the decision making self. Consciousness only has access to these limited data types.

We know consciousness is used for decision making because it is directed to model different parts of the animals immediate environment as the current situation unfolds. Attention is directed by different circuits to different salient features, and the conscious us decides. In the business analogy the director of sales may point out that there has been a shift in demand from one product to another, while the production director reports on changes in material costs. The CEO decides what, if anything, it all means for the business.

That consciousness is for rapid decision making is supported by the fact that it excludes information on which the model is based. I can only see that this is a lion, I cannot consciously check the process by which patterns of light on my retina were matched with the pattern for lion. I can refuse to believe, or blink and look again but, if my danger circuit has detected a lion it wants me to act. I cannot not see it. No matter what I consciously think chemical and electrical signals will be preparing my body to move.

The rule that conscious sensations are generated from data about the body and its environment to get the self to act is proved by the exception. I have concluded that in some instances the content of the model is censored to exclude data that might prevent immediate action. This can happen in trivial situations of the ‘I can’t believe he just said that’ kind. There is a delay before we become aware of a statement that cannot immediately elicit a socially acceptable response. Information can also be censored in critical situations.

The following account is by a soldier recalling his experience of finding the bodies of his comrades following the D-Day battle in 1944.

‘I said come on. Get up. Get up. Get up. You know I just couldn’t comprehend so quick that they were dead.’

I suggest this may describe an instance where the model self was presented with a censored version of reality. Perhaps his affiliation circuit’s detection of the loss of the people he could rely on was prevented from forwarding the news to consciousness. His danger circuit acting to block the signal after calculating that news of the loss might prevent him from taking measures to protect himself.

The model includes feedback on previous decisions. Whether it comes as an inner voice telling us we did well, or a feeling of shame, the feedback is from circuits attending to specific tasks. Feedback is in the same simple common language used for sharing data.

Feedback is only relevant where decisions are taken. In many instances decisions are taken by one circuit prompting action where there are no conflicting prompts from other circuits. Feedback on the outcome of the decision will come from how it impacts on other circuits priorities. Where an action has positive or negative impact on one circuit it will report and record that for evaluating similar situations in the future.

If our conscious experience includes feedback, and it can come in the form of feelings after an event as well as our internal dialog, then we can take it as proof that our consciousness is sometimes empowered to make decisions.

The self within the model is identified with the processor for decision making where there is no clear overriding course of action from a task specific circuit. In social animals this self, on whose behalf the decision is to be made, can be the extended self with affiliations and obligations. The reason the self within the model is tasked with making decisions, rather than having rules decide which circuit should take charge, is that this allows more subtle nuanced decisions to be made.

Though many decisions are taken by preconscious circuits, without conscious thought, we ascribe them all to the self. This illusion of agency is further proof that the purpose of consciousness is to take decisions. If a circuit acting alone was part of our self image then, faced with a difficult decision, we might wait for one circuit to take charge. This could nullify the enormous expense of energy made in translating and sharing data and building a model capable of taking more subtle decisions based on the overall needs of the animal. Better to waste a little energy telling the model it takes all decisions so that it will act when it is needed.

Everything that the self is consciously aware of is simplified data included in the model to either prompt or enable a current decision or give feedback on previous decisions.

Seen from this perspective while you have no conscious awareness of exactly which cells are damaged or which nerves are firing, when you stub your toe, a circuit in your brain for avoiding physical damage is prompting a decision making processor to take action to limit the damage, while at the same time giving feedback on recent action to prompt a higher priority for spatial navigation in the future.

Only simplified instructions are given by the model. You may move your toe to ease the pain but you have no conscious control over precisely which muscles contract to cause the movement.

So far this account has been one dimensional, based only on the goals of data processing in the brain. While it gives an insight into the nature of conscious experience it does not yet describe a mechanism that we might be able to replicate. To do this we also need to look at the types of data processed.

Animal brains use data to detect information about their surroundings to control their movement and to store useful associations such as that between the red light from an apple and acquiring sugar. This involves processing three separate types of data:

Causing muscles to act together for movement requires the storing of data for contracting and expanding groups of cells in sequence. I describe the management of this as a motor control system. Our motor control system can store and manage complex sequences of movement, running jumping and catching a ball for example without conscious control. We have to consciously learn some movements, such as the sequence of notes to play the piano. Once the motor control sequence is stored we do not need to focus consciously to play a tune.

The information gained by detecting light or pressure on cells, with that gained by detecting chemicals, are the results of what I call the sensory system. When you consciously see a tree it is the firing of a top down icon with the canonical shape of a tree. Depending on the distance away the icon used could be leaf, branch, tree or forest. I suggest that our language for trees reflects the icons we use to see them. If you see the same tree outside your door every morning your brain will create and store a specific icon for that tree. Using icons for familiar objects can speed up their recognition. Cut a branch from that tree and, later when you walk out the door, your sensory system will have to modify the familiar icon. Consciously you will notice the branch gone.

Making a useful link associating two types of sensory information—a pattern of red light and the presence of sugars in the case of an apple—requires a different types of data. The information would be something like: Detect red light. Move towards red light. Stop and bite when red is reached. Digest sugars. This is managed by what I call the narrative system. The narrative system has to be able to sort and store the flow of events to find patterns which are useful for predicting the future.

A coherent narrative is an essential part of a self-model. To survive a rabbit needs to know its own size and strength, what it likes to eat and where it can run for protection. The rabbit needs to remember where it fits in all of its relationships. Going to the wrong individual for support in a dangerous situation could be as fatal as mistaking yourself for a fox. Records of all of our daily actions are shaped by how they fit into our narrative self-model. The self-model is difficult to change because its coherence is vital for survival.

In humans rich verbal language makes our narrative system a powerful tool as our narrative records can be quickly compared using the shorthand of words. It also enables us to share patterns of association. This means we can learn from experiences we have not witnessed or been part of.

The narrative system is stronger than the sensory in humans, with narrative information literally changing what we see. We can demonstrate this by using narrative prompts to change the viewer’s interpretation of a bistable image for example.

Most of the time what we are consciously aware of is a sensory check on a narrative record. The sensory check is thorough, testing everything repeatedly. There will be many times when the sensory system does detect a change. Where there is no consequence to that change the sensory information is updated but awareness of the change is not promoted to consciousness. Where the change alters the current narrative we will become consciously aware; startled by a person in the empty room or gladdened by the face of a friend in a crowd.

All three systems have both on line and off line modes. For example my motor control system can work off line to prepare to catch a ball while online I type.

The systems can work independently. Our inner narrative can be reminding us to get food for dinner while our hands and eyes are busy working.

Consciousness emerges as subjective ‘what it’s like to be us’ experiences because data is being simultaneously processed by motor control, sensory and narrative systems using both chemical and electrical signalling.

There have been many reports from people who have survived sudden accidents such as falling from a mountain that they experienced time slowing down during the event. The reports liken the experience to a film in slow motion, where it seems to take three seconds for one second to pass.

One explanation is that perception involves at least two systems in operation, systems which normally work to the same time-base. If the time-base of one system is speeded up in a critical situation then the other system will appear slowed down. A high speed narrative system examining a normal speed sensory system will experience it as being in slow motion. This is the equivalent of film shot at 24 frames per second being viewed at 8 frames per second.

Time slowing down in accidents shows that our consciousness involves the interaction of different systems. It may also demonstrate how ‘what it is like to be us’ consciousness emerges as one system can examine the others.

If we try to mechanically model how consciousness emerges when our brains task-specific circuits interact we must model all three systems; sensory, narrative and motor control. We can model a sensory system using icons, from top down information processing, to selectively model some of the animal and its surroundings. We can model a narrative system by creating and using records of associations of events over time and across circuits, processed to chunk the data into stories with a beginning middle and end. We can model a motor control system that records patterns of actions to enable nimble movement.

Our model systems should be able to work both on and off line.

We can imagine a machine with three such systems that would report its experiences as being like our consciousness. I think of a ship controlled by such a machine. I call it the Clever Viking. It has circuits managing its needs which feed in data for decisions to be made.

Housed in a round box, the Clever Viking has central shafts on which are mounted arms with magnets on the end. This is the sensory system. As items are detected beyond the ship magnetic icons are placed on the wall of the housing at a height matching the appropriate arm. When food is detected the magnet is placed level with the food arm so that the arm swings towards it. The movement of the arm cause the shaft to turn and move a gear wheel, linked to gears that record which way the ship is pointing. The ship is prompted to move in the direction of the food.

When the crew is fed the food arm can be shortened so that the magnet has less pull. On the other hand if the food arm is fully extended it would be so close to the wall that it might start to respond to icons on other levels as if they were food. The Clever Viking could report that it is so hungry it can only think about food, and might even report that it perceived other things as food, like a starving dog in a cartoon imagining objects turning into steaks.

The central shafts are connected to a great series of gears so that the movement of the shafts can start the movement of the ship. Once a movement has begun the gears can control it and trip the next in a series of actions. This is the motor control system.

Finally the placing of the magnets on the wall of the housing, the position of all of the shafts and the movement of the starting gears are recorded by punching holes in paper tape. This is the narrative system. The system can match the current pattern of holes in the tape with previous patterns. Readout from the tape can prompt turning of the sensory arms or movement of the gears.

All three systems have online and offline modes. They move in synchronised steps like clockwork. They examine current situations, use records of past events to make predictions about the future, and decide which way to move the ship.

When the Clever Viking starts up in the morning the first reading of the narrative tape and sensor arms are used to check immediate surroundings. At the same time, each of the task specific circuits will scan for open task threads and may prompt an action. If the surroundings match familiar sensory icons, for the harbour say, a standard routine is triggered. Based on meeting everyday needs it may be testing the oars and sails or taking food from the store. If the surroundings are unfamiliar a deeper search of the narrative will take place to orient the ship for the day. As the day progresses a standard set of tasks, based on the main identity narrative, will be followed. For a lion this would be hunting for a human going to work or school.

The systems move together like clockwork. Each of the circuits can promote their goal by extending their sensory arm, by moving gears, or by loading a paper tape. A change in one system signals all three because they are interconnected, and this changes priorities. A change in the arm positions can cause the Clever Viking to change task. As the icons on the wall are changed, these too can trigger a change in task. A new icon will change the position of the shafts and so the narrative and motor control systems become aware of the change. A lion may be quick to decide to change task if they see easy prey, while a stronger narrative in humans will usually keep us on track.

Each of the systems can also make a quick forward prediction. For example the sensory system can ‘see running with scissors taking someone’s eye out’. Each clockwork step allows a comparison of current priorities for the whole ship. Where there is no clear single option for the next action the Viking can decide. What makes the Viking clever is how flexible it is in finding patterns of association from past experience to use as solutions to new problems

Once tasks are stared they are worked on till they are complete or the task thread is otherwise closed.

Imagine if one day such a Clever Viking, when the ships crew is reporting that they are cold and hungry, is given cake soaked in warn liquid. As the cake is taken in the food circuit signals that its composition is exactly what it has been searching for. First the narrative task of a search for food is ended, as is the search for warmth. The gears to move the ship are disengaged and the vessel relaxes its motion. At the same time the system for distributing the food to the crew is cranked up.

Next all of the sensory arms are extended. The food circuit is prompting the narrative to record details which will enable successful food acquisition in the future. The food arm will record a strong signal from the icon of the cake. The paper tape will record all the current details. The events surrounding the cake would be recorded in the food circuit, as would the warmth in the circuit for maintaining temperature. The affiliation circuit would record a debt to the one who presented the cake, and mark that this individual can be trusted for support.

Years later the taste of cake soaked in tea, clearly identified when the food circuit arm finds the same strong icon, causes a search of patterns of association stored by other circuits.

Finding a match, the icons from the previous time are loaded in the offline sensory and narrative systems. Not just the icons for the cake but all the events and patterns associated with it. The machine might report that the taste of cake would

‘... immediately causes the memory of the old grey house upon the street, where her room was, to rise up like the scenery of a theatre ...’

The circuits within our bodies and brains responsible for life’s tasks create an internal model of the external world a Cartesian puppet theatre. Centre stage in that model world is the model us, a Pinocchio-like puppet who can think. The circuits pull the strings to guide us. Because they often pull in different directions the circuits shape our model world, like the scenery of a theatre, to tempt us one way or warn us away from another.

When we understand how the cells that make up our bodies have evolved solutions to meet their needs like food and warmth, and to avoid threats like attack or poison, phenomenal consciousness is explainable. We have states of experience to enable our conscious self to make decisions. I am aware because the cells that make up my body need a conscious me to get what they need.

There are many cells involved in detecting and processing wavelengths. The data gained gives animals a survival advantage; detecting a predator before they are close enough to attack for example. We do not have zombie-like processing of wavelength; we are aware of the redness of an apple or the cord C played on a guitar. Our bodies devote even more cells and energy to translating the wavelength data into a shared language of feeling because there is even more value in having a model self to make decisions.

All our conscious thoughts and sensations are simplified data provided to the model self by task specific circuits. They are; decisive best guesses as to what is outside the body, reports on the body’s current condition, prompts for action from one of the body’s controlling circuits or feedback from those circuits on action taken.

The pain when you stub your toe is a signal from a circuit to limit damage and to remind the conscious you to take care in the future.

The pain of bereavement is a signal from a circuit that has recorded debts, like that for the care that handed us warm cake when we were cold and hungry, that can now never be repaid. It comes when we find that the image icons we use every day now have to have a piece ripped out, and that well worn patterns of behaviour have to change.

The joy of a new baby is part of your brain celebrating that its work is finished, for now, while another part is signalling that its real work has just begun.

Everything that we are aware of—all of the features of the model world we experience—are signals from circuits. They are as like reality as blips on a radar screen are like the sky. The mechanical nature of the signal can not make where the signal comes from any less real than a blip on a radar screen can make a plane full of passengers.

All our actions are driven by the circuits responsible for life’s tasks. We have circuits for food, poison, mating, competition, danger, affiliation, homeostasis and circadian rhythms. They are there to meet the same basic needs as all organisms. Behind all of that is whatever drives matter and energy to become atoms, atoms to become cells and cells to make life. Our conscious experience, and the job it has to do, is perhaps the most interesting part of that whole story.

We may be puppets in a puppet theatre but, like Pinocchio, the more we act free of the puppeteers’ strings the more human we become.