To explain knowledge, we need to explain how it is produced, we need therefore to explain the functioning of the spirit, how it perceives, imagines, feels, thinks, wants and acts. How does it work?

The connection between sensors and effectors

Cognition is the production and use of internal representations that prepare or lead to action.

To use representations, one must be able to act, one must be an agent, that is to say an animated body: a living being or a robot. An agent is always a system that interacts with its environment through sensors and effectors (Turing 1936, Russell & Norvig 2010).

The sensors (the sensory organs) are connected by a nervous system (the brain, the spinal cord ...) to the effectors (the muscles, the excretory glands ...) in order to produce an intelligent behavior (Churchland & Sejnowski 1992, Gazzaniga & Ivry 2001).

Sensory perception is the production of internal representations from the signals provided by the sensors. It prepares for action by making the agent able to adapt to its present environment. But the senses are not the only sources of internal representations. All forms of perception and imagination are ways of producing internal representations that prepare or lead to action.

Brain modules and routine activities

A brain module is a network of neurons specialized in certain tasks of information processing. It has entrance ways, where it receives information and orders, and exit routes, where it emits information and orders itself. It can be very localized (a small nucleus of neurons, a cortical micro-column ...) or quite extensive (a vast network distributed over several brain regions). It has its own skills and a partially autonomous mode of operation.

Cerebral activity as a whole is the result of the coordinated activity of all modules. They exchange information and orders and thus produce all the internal representations which prepare for action and all the signals which trigger it and control it.

A brain module can be conceived as an autopilot. The most subordinate pilots are the most peripheral, the neural networks which control the muscles and the rest of the body. These subordinate modules are controlled by other modules, and so on. A brain module always has a fairly limited competence. It has access to only a small part of the information available in the brain, and the repertoire of tasks it can perform is also very limited. But the higher-level modules, that is to say those who command at the highest level other modules, are able in principle to mobilize all the resources of the body and its brain. Such a module is a kind of leader in the brain, an autopilot that pilots the other pilots.

A module can represent its own purposes, give information and orders to other modules, or receive them, and thus participate in the proper functioning of the organization. The spontaneous activity of the modules is sufficient to explain the routine behaviors that result from instincts or learning. The necessary resources are recruited automatically and perform their tasks as they are used to. There may be a leader who temporarily leads the march of the ensemble, or several, or none, because the modules can work separately and spontaneously coordinate their activities.

Decision and will

Actions are voluntary or involuntary. When they are voluntary, they result from a decision.

When we made a decision, we could have made another choice. The decision is preceded by an evaluation phase. The foreseeable consequences, external and internal, of the decision must be taken into account.

The assessment that precedes a decision is a kind of high-level perception, a perception of the whole situation that requires a decision. High-level perception is a synthesis that mobilizes most of our capacities to perceive.

Actions aim to transform reality, exterior, interior, or both.

We can use the will to transform ourselves internally. Most of our inner resources can be influenced and transformed by our decisions. This is how the will can end up transforming itself.

The evaluation which precedes a decision is a kind of collective deliberation, in which our internal resources are invited to participate. Once the decision is made, these same internal resources must respect it. The internal organization which allows the will to exist resembles a centralized administration without a central administrator. A common law is decided by all and is binding on all.

For the goals and rules we have voluntarily decided to mobilize our domestic resources, they must be kept in working memory. Some modules must be specialized in the recording of our decisions and the distribution of the orders that result. The stored decision is used to send orders to all the modules concerned by the execution of this decision, as long as the goal is not reached, or has not been waived. The modules that memorize our voluntary decisions are prime contractors. So we can call them executive modules. Other brain modules are usually subordinate to these executive modules.

Executive modules are not innovators. They simply record decisions made elsewhere and automatically distribute the orders that apply them. They are not homunculi, or little geniuses in the head, but only neuronal circuits capable of recording the decisions received on their input channels, and then giving the orders that apply them to their output channels. It's just about processing the information, not putting spirits in the machine.

Our voluntary projects are proposed, developed and evaluated by all of our internal resources, and once adopted, they impose themselves on these same internal resources, which must obey the orders given to them. But there is no chief, no central administrator. Executive modules only record decisions made by the community. They too therefore only obey common order.

As emotional systems evaluate the goals we decide upon, we can think of a model of the will that reduces it to a role of servant of emotions. Voluntary decision-making could simply involve submitting a project to emotional systems and then counting their evaluations. If the favorable opinions clearly outweigh the others then the decision is made. The will thus conceived would be heteronomous, it would only obey an external law, that of emotions.

The will is autonomous because it gives itself its own law. It is autonomous in its evaluations when it makes its decisions based on rules or evaluation criteria that it has itself decided.

The goals on which we decide ourselves may be suggested by the systems of perception and emotion independently of any voluntary control. In such cases the will only has to agree to projects that it has not elaborated. All that is asked of it is to give its signature. But we can also decide to develop projects on which we will decide later. The will is autonomous in its execution when it decides to elaborate the projects which it will submit to its evaluation. An system autonomous in its evaluation and execution is able to pose and solve problems.

The model of a centralized administration without a central administrator explains why the ego is like a strange loop (Hofstadter 2007). I can decide the criteria for evaluating my decisions, because the will is autonomous in its evaluations. I can also decide which objects will be the subject of my next decisions, because the will is autonomous in its execution.

When their behavior is routine, agents do not need to look for solutions for long. They find them spontaneously because their brain modules know how to produce them, by instinct or habit. The agents are content to solve the problems they already know how to solve. But faced with a new situation, the usual reactions are not always adapted. The agent may have the domestic resources to react appropriately, but it does not know how to mobilize them, because it would have to invent a new mode of coordination between its brain modules. None of them have the means to recruit others, even though it would be sufficient for them to work together to achieve the desired ends. The agent would need an inner composer-conductor, able to find truly new solutions (Shallice & Cooper 2011). The model of a centralized administration without a central administrator shows, without postulating the existence of a spirit in the machine, how the brain can function as if it were equipped with such a composer-conductor.

Attention and consciousness

We pay attention to our decisions when we make them. We also pay attention to all the information that could be used to adopt or reject a decision. We also pay attention to the information that is used to control the implementation of a decision.

Our executive modules, responsible for the application of our voluntary decisions, are necessarily limited in number. Their memory resources are also limited. This is why the inner order of goals and rules that direct our behavior is of limited complexity. We can not do too much at once. Each proposal must be examined in turn.

Attention is the selection of representations to make decisions and control their execution.

A signal is conscious if and only if it is a signal that one can pay attention to.

An empirical test of conscience is a decision test. For example, we give as a rule to a subject that she must press a button when a certain signal appears. The subject shows that she is aware of the signal by passing the test.

Imagination and voluntary control of attention allow the will to operate in closed circuit, because it can decide itself the information from which it makes new decisions. We can thus be focused on what we imagine and feel as cut off from the world. But such isolation is never complete. An unexpected event is enough to get us out of our meditation and capture our attention. Unexpected event information, evaluated for a voluntary decision, was selected by an unintended process (Lachaux 2011). The will is autonomous and can decide for itself what holds its attention, but not to the point of completely avoiding external influences.

The model of centralized administration without a central administrator explains how the brain makes us able to have an autonomous will, to pay attention, to form beliefs and to voluntarily control perception, imagination and thought. This is a theory that explains the functioning of our brains when we are conscious, but this is not enough to explain the appearance of consciousness from brain activity (Chalmers 1996). Attention is the selection of representations to make decisions and control their execution, but selection alone does not explain why representations so selected become particularly conscious. A robot can also select representations to make decisions without implying any consciousness.

Consciousness arises from brain life, but we do not know how to explain it. Nerve impulses are produced by electric currents in neurons and through their membranes. These are very ordinary ion currents. There is nothing to suggest that they must be messengers of the spirit.

The imagination of the present

Perception and imagination are often thought in opposition. What is perceived is present, what is imagined is not. If for example I am in a familiar place, I can imagine the layout of the place even in the dark. I know that various objects are present and where they are but I do not perceive them directly.

A simplistic and partially false modeling of perception assumes that it is unidirectional. The information is first produced by the sensory detectors and then synthesized in successive steps to the high-level representations, which determine the main perceived objects and the main concepts attributed to them. It is assumed that complex representations emerge from elementary perceptions, as in a pointillist painting. Such a dynamic of production of representations is called upward or bottom-up, because sensory signals are considered as low-level representations, while concepts attributed to complex objects are high-level. This modeling ignores the anticipation effects. It makes it possible to explain the internal representations of objects actually detected, but not the representations of objects or qualities whose presence is only assumed.

In the strict sense, perception is only sensory. Perceived representations are awakened or at least confirmed by information that comes from the senses. But the representation of the present goes beyond strictly sensory perception. We could not even take a step if we limit ourselves to the data directly perceived by the senses, because we have to anticipate that the ground will resist before feeling it directly.

In a general way our representations of the present come from both sensory and memorized informations. For example, when we grasp a familiar object, the gesture is prepared to fit the weight of the object. If we do not anticipate the weight, the gesture is not adapted. This shows that we have an internal representation of the weight before we hold the object in the hand. The weight is therefore represented before the muscular tension sensors provide this information. We can say that the weight was imagined, but we can also say that it was perceived indirectly from the visual image, thanks to a memorized knowledge on the ordinary weight of such an object.

What is perceived is not only determined by the senses but also by expectations and desires, by previous perceptions, memories, prejudices, culture and knowledge. The waiting effects can be so strong that sometimes we think we have seen what we could not see, because it did not exist. Our perceptions therefore have inner sources, they are not only elaborated from the senses. The dynamic of representations is not only upward, but also downward, top-down. The sensing systems that receive the sensory information also receive higher level information. We must model a kind of permanent dialogue between the various stages of perception. Information can travel in all directions, from bottom to top, from top to bottom, and horizontally (Hofstadter & FARG 1995). Any representation can have an influence on the output of others, whatever their level of complexity.

As the representation of a present situation is always part of a system of presuppositions, the present is always as much imagined as really perceived. Sensory perception can even be considered as a form of imagination, stimulated and guided by the senses. It is an imagination of the present in accordance with the data of the senses.

A detection system is above all a warning system. The detection signal warns of the presence of the being detected. The warning function is more fundamental than the detection function, because the system can signal the presence of a being which has not been detected. It is enough that this presence is supposed, or inferred from the detection of other beings.

An inference consists in passing from a condition to a consequence. The consequence is a representation produced, inferred, from the representations which determine the condition. If these representations are verbal, an inference is a step in a reasoning, but representations are not necessarily verbal. Perception proceeds by silent inference as soon as it connects consequences and conditions.

Inferences can be chained because consequences can themselves be conditions that have consequences, and so on. A chain of silent inferences resemble very much a reasoning. The sequence of representations of the conditions and their consequences is similar to that of their verbal descriptions chained in a reasoning.

Silent inferences make that there is no clear boundary between sensory perception and the imagination of the present. When a representation has been produced by inference, as a consequence of a condition already perceived, we can say that it is imagined, but we can also say that it is perceived indirectly from the perception of the condition.

A schema, or a conceptual framework, is a system of preconceptions, that is to say, what is held true before having verified it. A schema determines the beings we expect to perceive with the concepts that we think we should attribute to them and the inferences that we believe we can apply to them.

Sensations are the sources of the bottom-up processes of perception, schemas are the sources of downward processes. They are part of the normal functioning of perception. They are necessary to adapt quickly to the environment, because to act we often do not have time to check everything.

Knowing the right schemas makes all the difference between the expert and the neophyte. An expert often needs only a glance to correctly analyze a situation and draw the necessary conclusions, because she already knows the schemas that make it possible to understand it and she only has to check their adaptation. A neophyte is overwhelmed by the flow of new information, does not know what to look at, does not distinguish the essential from the negligible and rarely asks the right questions, because she does not know the schemas that would allow her to organize her perception of the situation.

In the strict sense, perception is only the imagination of the present when it is awakened or confirmed by the senses. But we can also define perception in a more general sense and speak of the perception of the past (the remembrance, and more generally any form of imagination of the past), of the future (anticipation), of the imaginary (to dream of beings that do not exist) and even of abstract beings (abstract, mathematical knowledge for example). Thus understood perception and imagination are synonymous.

The imagination of the absent

To act we must perceive and imagine the present, because we have to adapt to reality, but we must also imagine the absent, the goals that we set for ourselves and that we have not yet achieved, the means to implement and the foreseeable consequences of our decisions.

Perception systems can function as alarms even if the beings whose presence they signal have not been detected. They can report a hypothetical presence. They thus make it possible to completely overcome the senses and simulate the perception of a scene that is not present. The imagination of the past, of the future and of purely imaginary worlds is perception without detection, therefore a simulation of perception. The resources of perception are mobilized to represent an environment that is not present, only imagined.

Simulating perception consists in simulating the activation of our detection systems. One can simulate sensory perception and partially reconstitute sensory images or impressions, but imagination is not necessarily associated with sensory images. To imagine a dangerous being it is not necessary to make a visual image of it, or to imagine its voice, or any other form of simulated sensory perception, it is sufficient to simulate the activation of a danger detector. You can imagine yourself in the vicinity of a dangerous being even if you do not perceive anything of it, except that it is dangerous.

The mind is first known through the experience of oneself. By remembering all that one has lived, one recognizes oneself as a mind. But we extend our knowledge, by the imagination, by putting ourselves in the place of other minds, of all the minds that we can imagine.

To know someone as a soul, as a being who imagines, who feels and who wants, by putting oneself in his or her place, enables us to anticipate the immediate consequences and the long-term effects of our present acts on his or her behavior.

By imagination we can combine representations in new configurations that we never perceived. The parts have been perceived, but their assemblage is invented, it is purely imaginary, it represents a fictitious being, a kind of chimera. By assembling fragments of sensory images, like a patchwork, we can create an image of a being that does not exist. In general, the assembly of concepts makes it possible to create representations of beings which have never existed and which may never exist. Combinatorics multiplies the possibilities endlessly.

With silent inferences, we can predict the chain of consequences of our decisions. We can thus explore by imagination the paths that we could follow. We thus discover at the same time the goals that we could reach and the means to reach them.

The importance of representations of the present and the future for the preparation of action is evident, that of representations of the past is a little less. Remembering prepares us for action indirectly, if only by helping us to perceive the present and the future, by inference from the knowledge of the past. But the imagination of fictions, how can it prepare for action? It seems it is taking us away from it. To act well one must have feet on the ground, one has to adapt to what really exists. What is the use of imagining beings who will never exist?

The work of the novelist is similar to that of the mathematician. She posits conditions, an initial situation and constraints, and then exposes their consequences, which are often inescapable, in the same way that a mathematician proves theorems from axioms and hypotheses. When we imagine fictions, we can fully utilize our abilities to infer. It is not only a question of inventing assemblages of representations, but above all of imagining all that results from it, all that our inner dynamic of production of representations by inference can provide from these inventions. The imagination of fictions reveals the power of inference.

Introspection

The perception of one's own body can be seen as a kind of self-perception. For example, the information provided by muscle tension sensors makes it possible to construct an internal model of the body, the position of the limbs and the efforts to which they are subjected. But the knowledge of oneself is more than the perception of one's body, because the spirit is constantly a witness of himself.

If I see that the sky is blue, I am not only informed about the state of the sky, I am also informed about myself, namely that I see the sky, I know myself as a being who perceives the sky.

Introspection is the knowledge of oneself as a spirit, that is, as a being who perceives, imagines, feels thinks and wants.

Does introspection require sensory organs? Is there a sensory interface between the perceived self and the perceiving self? When I know that I see the sky, is it an introspective eye that shows me that I see the sky?

A sensory organ is always an interface between an inside, the nervous system, and an outside, the environment beyond the skin or the interior milieu under it. The external signals are received by the sensory interface and translated into internal signals, usable by the nervous system.

Introspection does not require a sensory organ because there are no external signals to translate into internal ones, no separation between a perceiving self and a perceived self. Everything happens inside. All information about the agent, as it perceives, imagines, feels, thinks or wants, is already present inside the agent. To develop its faculties of introspection it has only to exploit these internal sources of information. A sensory organ of introspection is not necessary because the information sought is already present inside.

To know oneself, one has to perceive oneself, so to represent oneself. But where does one find that self which must be perceived? And how does it represent itself?

La Gioconda is not only a representation of Mona Lisa, it is also a representation of Leonardo da Vinci, because it carries a lot of information about him. My representations do not only provide information about the beings represented, they can also say a lot about my way of representing them, and therefore about me. They are also representations of me. My representations of the world do not only represent the world, they also represent me.

Emotions

The concept of emotion is difficult to define and its use is often very imprecise. Should we distinguish moods and emotions, moods because they are durable, emotions because they are brief? Is tranquility an emotion or independence from emotions? Is jealousy an emotion or a more complex state that mixes emotions and will?

We can define emotions from some basic emotions (sadness, fear, anger, disgust, shame, joy, appeasement, pride, surprise ...) and include all the variations and combinations, or from some general characters:

• An emotion is triggered by the detection of specific conditions, fear by the detection of danger, sadness by the detection of misfortune, anger by the detection of the unacceptable ...
• This detection is followed very quickly by reflex reactions and physiological changes that enable the body to adapt to the novelty of its situation.
• Emotions determine motives, ie desires or aversions. They tell us the goals that deserve to be pursued, and what we must flee or avoid (Damasio 1994). They are therefore very important to the will, because they serve us to evaluate our projects, and for learning, because they point out what deserves to be memorized.

Because it is triggered by specific conditions and because it causes specific reactions, a particular emotion, such as fear, can be characterized by the activity of a brain module, or a system of modules that coordinate their activities. Input channels carry the signals that awaken, modify or suppress the emotion. Output channels carry the signals that provoke typical emotional reactions (LeDoux 1996). As an emotion can mobilize a large part of the body's resources, such an emotional system can be considered as a kind of leader in the brain. An emotion, especially if it is strong, can exert a sort of empire over all the bodily activity, interior and exterior.

Emotion and cognition are sometimes thought in opposition, but it is a mistake. Emotions produce and use internal representations that prepare for action, so they are part of cognition. They are valuable informants on the external and internal realities.

Beliefs

Perception and imagination continually attribute concepts to perceived and imagined beings. All of these concept assignments are beliefs, as long as they are remembered.

When we become aware of a belief, we can choose to doubt it. A conscious belief is evaluated before the decision to approve it. It is then taken over by the executive system, which stores it in working memory and uses it to control perception and imagination.

A conscious belief can have an effect, through the executive system, on the whole functioning of the brain. We thus rediscover an element of the cognitive theory of consciousness of Baars (Baars 1988, Changeux 2002, Dehaene 2014). As long as a representation does not hold the conscious attention, it remains attached to its place of production and can not affect the entire system. Its effect is necessarily limited and localized. But if one becomes aware of it, it can be used to influence other parts of the brain, it is like written on a blackboard which can be read by other brain modules.

After remaining active for a moment in working memory, beliefs are usually recorded and consolidated in long-term memory, where they remain as dormant, or latent, beliefs. They are awakened and become active again if we remember them.

As a conscious belief is evaluated before being approved, the whole evaluation process is part of the perception of the assigned concept. The evaluation capacities which precede the decision are concept detection capacities. As all concept assignments can be subject to conscious approval, the decision system works like a universal detector, capable of detecting any concept, as soon as it has learned to do so.

The invention of perception

Our capacities of perception are not fixed. We can always learn new ways of perceiving.

Sensitivity can be refined by experience. Our perception systems are sensitive to context and to many parameters that we can vary to make ourselves more sensitive to what we perceive.

We can invent concepts by combining the concepts we already have. If one expects to perceive a combination of concepts again, one can acquire the ability to detect it as a new unique concept. The same can be done for the detection of combinations sufficiently similar to other combinations already perceived and considered as typical examples.

Learning by experience to make new silent inferences increases our ability to perceive, because we thus learn to perceive the consequences from their conditions.

Our capacities of perception depend on our anticipations, and therefore on the schemas, the systems of presuppositions, which we have adopted. By modifying our schemas we can modify our ways of perceiving and interpreting reality. By making us free to adopt or reject beliefs and schemas, the will makes us free to discover or invent new ways of perceiving.

Concept detection can be an almost instantaneous process, if the detection signal is produced as soon as information about the detected being is provided, or progressive, if the detection system takes the time to accumulate information before giving its result. The assessment process that precedes a decision is generally progressive. There is time to take into account a lot of information before deciding.

As we are free to decide how we evaluate our decisions, we are free to invent ways of perceiving. We can invent all the concepts we want by giving ourselves the means to detect them.

To put oneself in the other's shoes

We know another spirit by imagining that he perceives, that he imagines, feels, thinks, wants and acts. We imagine that he perceives by imagining what he perceives. We imagine that he imagines by imagining what he imagines. Sympathy is to feel what he feels. In general, we know him as a spirit by putting ourselves in his place (Goldman 2006, Rizzolatti & Sinigaglia 2006). We can imagine that we want what he wants and that we do what he does. A spirit is a universal simulator because he can simulate all other spirits, at least if they have the same faculties - for a human being it is easier to put oneself in the place of a human being than in the place of a bat.

A spirit knows the spirit both by knowing himself and by putting himself in the place of other spirits, of all the spirits he can imagine.

Everything that can be perceived, imagined, felt, thought or decided by some, can be perceived, imagined, felt, thought or decided by all the others. To know oneself as a spirit is at the same time to know what everyone else can do with their spirit. Conversely, everything that others do with their spirit shows us what we can do ourselves. "Nothing human is alien to me." (Terence, Heautontimoroumenos, v. 77)

I know another spirit by putting myself in his place in imagination. Likewise, he knows me by putting himself in my place. When I wonder what he thinks of me, I try to imagine what he imagines when he puts himself in my place, I imagine what he imagines of me. I can also imagine what he imagines when he puts himself in the place of a third.

The same content can be imagined in various ways: a past that we have lived, a projected future, a simple hypothesis, a content imagined by others, and even a content that others believe we imagine. When we remember, we put ourselves in the place of the other that we were. When we look to the future, we put ourselves in the place of who we could be. From this point of view we know ourselves in the same way as we know others, putting ourselves in the place of ourselves in imagination. When I imagine what I could be, I am in a position similar to that of another who imagines what I could be. When I imagine how someone else imagines me, I imagine what I could be if I were like he imagines me.

What is speech?

In the beginning was the Word,
and the Word was with God,
and the Word was God.
He was with God in the beginning.
Through him all things were made;
without him nothing was made that has been made.
In him was life,
and that life was the light of all mankind.
The light shines in the darkness,
and the darkness has not overcome it.

(John 1, 1-5)

We can reason about the world as if it were a big book. To reason about beings is always to reason about what we say about them. All truths are determined from elementary and fundamental truths, all atomic truths about all beings. All these truths together are like the big book of the world.

If one hears the concept of speech in its broadest sense, everything speaks. Being is a speech. To be, for all bodies, is to act on other beings. All material beings do it, simply by being material. To be visible, or audible, or to be perceived in any other way, is always to show oneself, and that is always to speak. Living beings perceive other beings and live by reacting to what they have perceived. Their reactions show how they understand the speech expressed by the beings they perceive.

In a more restricted sense, animals speak only when they emit signals so that they are perceived and influence those who perceive them. They show themselves to the eyes which look at them, or they make their presence felt in another way. It is not necessary to assume that the animal gives its signals voluntarily. A wasp indicates by its yellow and black robe to a possible predator that it is a dangerous prey. Evolution by natural selection is sufficient to explain such animal communication. Wasps have an advantage in showing that they are dangerous and predators have an advantage in perceiving it. There is no need for will or conscience. Blind instincts are sufficient to establish this communication.

In an even more restricted sense, animals speak only if they emit signals because they want to influence the imagination and the will of those who perceive them. This assumes that the animals which emit and receive these signals are capable of imagining and wanting, and especially that they have the knowledge of the mind, that is, they know themselves and others as beings who imagine and want.

Speech is the voluntary emission of signals to influence the imagination and the will of those who receive them.

Meaning through imagination

Why do words make sense? What makes sound sequences useful for communicating? What gives words and statements their meaning?

When one understands a description, one imagines what is described. Words and verbal expressions arouse the imagination as soon as we understand their meaning. One imagines what is described when one simulates its perception, when one activates, in simulation mode, the detection systems which would be awakened if we perceived what is described. When concepts detected by our perception systems are associated with verbal expressions which name them, we can both describe what we perceive, by naming perceived concepts, and imagine what is described by simulating the detection of named concepts (Saussure 1916).

Silent knowledge is the knowledge that precedes speech and results from perception, imagination, emotion and will. It can be translated into words as soon as the detection systems that it uses are named by verbal expressions. Descriptions are then a translation into words of the knowledge of perception and imagination. The laws which relate the description of conditions to the description of consequences are a translation of silent inferences. A reasoning which combine these laws is a translation of a sequence of silent inferences. In this way, silent knowledge can be translated into words, and thus communicated.

Silent knowledge is fundamental to the development of reason, because talking knowledge begins as a translation of silent knowledge. It can then fly on its own because it can speak about speech, but it needs silent knowledge to take off, because words must awaken the imagination to make sense.

A verbal expression has a meaning when it names a concept or an individual. The named concept, that is to say, the property or the relation, is the meaning of an expression that names a concept. The named individual is the reference of an expression that names an individual. The reference of an expression that names a concept is sometimes identified with the extension of the named property or relation.

A concept is empirical when it is an observable property or relation. Silent knowledge is always empirical.

We understand the meaning of an expression that names a concept or an individual when we know how to detect them.

The same verbal expression can have several meanings. The same name can be used to name several concepts or several individuals. It can be interpreted in many ways.

The understanding of speech can be conceived in a purely passive mode, as if the words were the notes of a score and the understanding, the music performed by an inner mechanical piano. But imagination is not only awakened by words in this passive way. The understanding of speech is also and especially active. We understand because we want to understand, and what we understand, that is, what we imagine, often depends on our expectations.

To understand a discourse, one must identify the concepts named by words and expressions. We have to find or invent the ways of perceiving the named concepts. When a child learns to speak, at the same time he learns new words or expressions and ways of perceiving the named concepts. When a speech invents new expressions, it invites us at the same time to invent new ways of perceiving.

Speech gives us the means to invent and communicate ways of perceiving. All the concepts invented by one can be communicated to the others.

A description can be communicated for itself. In this simple case, the speaker understands what she says if she perceives or imagines what she describes, and the listener understands what is said as soon as she perceives or imagines what is described. Knowing how to describe what we perceive or imagine and knowing how to perceive or imagine what is described are the foundations of language understanding language. All uses of speech rely on descriptions.

Understanding words means knowing how to use them

Wittgenstein has invited us not to separate the study of the meaning of language from that of its use:

« To imagine a language means to imagine a form of life.» (§19) « Here the term "language-game" is meant to bring into prominence the fact that the speaking of language is part of an activity, or of a form of life. » (§23) « For a large class of cases-though not for all-in which we employ the word "meaning" it can be defined thus: the meaning of a word is its use in the language. » (§43, Philosophical investigations, 1953)

For example, he proposes to reason on a simplified example of linguistic communication:

« The language is meant to serve for communication between a builder A and an assistant B. A is building with building-stones; there are blocks, pillars, slabs and beams. B has to pass the stones, and that in the order in which A needs them. For this purpose they use a language consisting of the words 'block', 'pillar', 'slab', 'beam'. A calls them out; --B brings the stone which he has learnt to bring at such-and-such a call. » (§2)

Any way of using speech as a speaker or listener is a way of understanding it. The comprehension of a language is nothing other than its use. The speaker understands what he says when he knows what he is doing by saying it. The listener understands what is said when he knows what to do with it. To understand words is to know how to use them. To explain how we understand words we must explain how they prepare us for action.

Mutual understanding

A speaker acts on those who listen to her. To know what she does when she says what she says, she must know what the audience does or can do with it. A speaker must be able to put herself in the place of the listeners and to understand what they understand, otherwise she does not really understand herself. Conversely, to know what to do with what is said to them, listeners must understand the speaker's intentions, why she says what she says. They must therefore be able to put themselves in the speaker's place and understand what she does, otherwise they do not really understand what is said to them. The understanding of words is one of the forms of mutual understanding, in which everyone knows others and oneself, and knows that one is known by others in the same way as one knows them.

A speaker imagines herself as a speaker. She knows her intentions. She knows the role she gave herself. When there is no misunderstanding or deceit, the listener imagines the speaker in the same way that the speaker imagines herself. She assigns the speaker the same intentions as the speaker assigns to herself. In this way, it is easy for the speaker to imagine how the listener imagines the speaker, because that is already how she imagines herself. The same content imagined by the speaker is attributed at the same time to the speaker who imagines herself and to the listener who imagines the speaker. The same is true of the content imagined by the listener which is assigned at the same time to the listener who imagines herself and to the speaker who imagines the listener. When there is no misunderstanding, the speaker and the listener imagine the same scene and the same roles, as if they were the viewers of the same film. The only difference is that they give themselves different roles.

Theoretical truth

As soon as they have an empirical meaning, the expressions of the theory can be used to formulate descriptions, to state observations. The statements are true as soon as the observations they translate are (Locke 1690).

Observation is not the only criterion of truth, because theories impose the truth of their principles (Leibniz 1705, Kant 1787). Theories, or theoretical frameworks are the talking equivalent of silent conceptual frameworks. A theory is determined with a system of names, intended to name concepts, and a system of axioms and definitions, which make it possible to reason with the named concepts.

A theory receives an empirical interpretation when the named concepts are identified with observable properties or relations that together form an empirical conceptual framework. The same theory can be interpreted by several different empirical conceptual frameworks, but it places constraints on acceptable empirical meanings. The interpretation must not contradict the principles of the theory. For example, the truth of the principle of transitivity, 'if x is greater than y and y is greater than z then x is greater than z' is accepted by definition of the relation 'is greater than'. This relation can be interpreted in many empirical ways, but the principle of transitivity can never be contradicted by our perceptions. If it leads to an erroneous anticipation, it will be said that the observed relation has been misnamed, that it is not an empirical meaning that can be given to the expression 'is greater than'.

The paradox of Condorcet (1785), in political science, illustrates the priority of an a priori principle:

One can assume that electoral results give strength to the various candidates and consider measuring this force. Assume that in an election where each voter must rank three candidates A, B and C in order of preference, the three ABC, BCA and CAB orders have each been chosen by one-third of the electorate. It seems that the force of A is greater than that of B, since two-thirds of the electorate prefers A to B. Similarly B is stronger than C, and C is stronger than A. The principle of transitivity is therefore contradicted by experience. But it is not refuted, it has only been poorly applied, because such an electoral system does not make it possible to measure the strength of the candidates. It will not be said that the theoretical framework (the measure of the magnitude of the forces) is false, but only that it is not adapted to the perceived reality.

In general, an expression can have many empirical meanings, and new ones can always be invented. But when principles are true by definition they impose constraints on interpretation, limits on the empirical meanings we can give to our expressions. Empirical interpretations are not independent of abstract interpretations. Truth by definition is generally a priority. If an expression is used in a way which contradicts a principle, it will be said that the interpretation is not correct, or that it is not one of the interpretations permitted by the principles. In this way, we can be sure that our principles are always true, because an interpretation that would make them false is a priori excluded.

The principles of a theory are accepted by definition of their terms. Their truth is supposed to be known as soon as the meaning of words is understood (Pascal 1657).

Statements provable from a system of axioms and definitions are the theorems of the theory defined by such principles. A theory gives an abstract meaning to the names with which it forms its statements. Sometimes axioms are said to be disguised definitions, because they serve to give meaning to their terms, and thus to define them.

A theory can be used in a way that resembles perception, like a system of detectors. To know if a statement is true or false, it is enough to prove it or to prove its negation. In this way the names of concepts are associated with theoretical detectors that determine whether the concepts are true of the named individuals. Theoretical detectors detect what they must detect by finding logical proofs, based on accepted principles. If they do not find any, their detection has failed. « The eyes of the soul, whereby it sees and observes things, are none other than proofs. » (Spinoza, Ethics, Book V, prop. 23, Scolie) Just as the eyes of the body enable us to see visible beings, logical proofs based on principles enable us to know abstract beings. The perception of abstract beings consists in reasoning from the principles which define them.

Words and verbal expressions can be interpreted in many ways and thus receive many empirical or abstract meanings. To determine an empirical interpretation, it is enough to associate to the named concepts perceptual systems which enable us to detect them. In order to determine an abstract interpretation, it is enough to associate to the named concepts systems of principles which enable us to reason with them.

According to the interpretation given to it, the same statement can be at the same time an abstract truth and an empirical truth. Such an ambiguity can be very useful, because by developing an abstract knowledge, one at the same time obtains an empirical knowledge. We do not even need to change the wording. Obviously for such magic to occur, the theory and its interpretation must be adapted to the observed reality in order to make it intelligible. The encounter between the abstract truth and the empirical truth of our statements is the goal of all empirical sciences, because we want to know by reasoning what we know by the senses.

Logical principles always make us pass from the true to the true (Aristotle, Prior analytics). When affirmations are true, their logical consequences can not be false. More precisely, whatever interpretation we give to affirmations, if these affirmations are true, according to the supposed interpretation, then the logical consequences are also true, according to the same interpretation. The relation of logical consequence does not depend on the interpretation of what we affirm, it depends only on the meaning of the logical operators.

When we prove a conclusion by logical reasoning, the premises determine sufficient conditions of truth. Whatever interpretation is chosen, if the premises are true, then the conclusion is true. Reasoning serves not only to prove, but also to explain the conditions of truth. To understand a theorem, one must know its proof, because it gives truth conditions which specify how one must interpret it.

When we learn a language, we learn at the same time new expressions, new ways of perceiving, which give empirical meanings to these expressions, and new principles with which we can reason. We develop thus our empirical knowledge and our abstract knowledge at the same time. The two are intertwined in an inextricable way because in general one and the same expression combines both empirical and abstract meanings.

To know if a statement is true, one must first specify its meaning. The same speech may be sometimes true, sometimes false, it depends on its interpretation. Most of our controversies come whether from misunderstandings, because we give different meanings to the same expression, or lack of precision, because we leave in the vague the truth conditions of what we say. We do not explain the principles which decide abstract truth or we do not specify the systems of perception which decide empirical truth.

The diversity of interpretations can make communication of knowledge very difficult. The speaker must respect a principle of clarity: to provide clarification so that his speech can be interpreted correctly. The listener must respect a principle of charity: always to interpret a discourse in the way that is most favorable to it, as far as possible. It is always possible to dispel misunderstandings and to reach consensus, because we can all do the same reasoning and perceive the same world.

Thought

« Do you accept my description of the process of thinking?

- How do you describe it?

- As a discourse that the mind carries on with itself about any subject it is considering. You must take this explanation as coming from an ignoramus ; but I have a notion that, when the mind is thinking, it is simply talking to itself, asking questions and answering them, and saying Yes or No. When it reaches a decision — which may come slowly or in a sudden rush — when doubt is over and the two voices affirm the same thing, then we call that its judgment. So I should describe thinking as discourse, and judgment as a statement pronounced, not aloud to someone else, but silently to oneself.

- I agree. »

(Plato, Theaetetus, 189e-190a)

Thought is the imagination of speech.

To think is to explore by imagination everything that can be done with speech.

Just as one can train others and be trained by them with speech, so can one train oneself with thought.

We build ourselves by word and thought, but we can also say that it is the word that builds us.