Handbook of epistemology/Perception and imagination
Perception is the imagination of the present
Imagination is the production of internal representations.
Sensory perception is a form of imagination, stimulated and guided by the senses. The internal representations are produced from the signals provided by the sensory organs.
The representations of the present environment are not necessarily of sensory origin. 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.
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.
Perception and imagination are often thought in opposition. What is perceived is present, what is imagined is not. But this opposition forbids talking about the imagination of the present and defining perception as a form of imagination.
In the strict sense of this book, perception is only the imagination of the present. 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 heard perception and imagination are synonymous.
Associations and silent inferences
Most internal representations influence actions not directly by controlling the muscles but indirectly, by participating in the production of other representations. Internal representations, except perhaps those directly used to control the effectors, always have the function of producing, modifying or suppressing other internal representations, or at least of participating in the dynamics of production of internal representations.
Because the brain is a massively parallel machine, a representation can excite or inhibit many other representations. This is why representations are usually produced in associations. Mutually excited representations are awakened simultaneously.
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 inference as soon as it connects consequences and conditions.
An inference is a special kind of association. The link of excitation of the consequence by the condition must be sufficiently strong so that the consequence is always awaken when the condition is.
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.
Sensory perception is to sensation what reason is to intuition. Intuitions come spontaneously but we must reason from them to develop the rational knowledge to which they sometimes lead. In the same way sensations are produced spontaneously by the senses but it is necessary to proceed by inference, to produce new internal representations, in order to develop a well-informed perception of the perceived being. Sensory perception is like a reasoning on sensations.
Silent associations and 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.
Imagine to simulate other souls
The imagination of the past, of the future and of purely imaginary worlds is a simulation of perception and action. Part of the resources of perception is mobilized to represent an environment that is not present, only imagined. This simulation of perception activates at the same time the systems of emotion, motivation and action. We can imagine what we would perceive, what we would feel and what we would do, if we were elsewhere, at another time, or in place of someone else (Goldman 2006, Rizzolatti & Sinigaglia 2006).
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.
Perception is conceptual
Concepts are properties or relations. A property, or a quality, is attributed to a single object. A relation is between several objects. When a relation is between two objects, it can be considered as a property of the couple. A relation between three objects is a property of the triplet, and so on for relations between more objects.
Perception automatically attributes concepts to perceived objects. Visual perception attributes visual qualities (color, brightness, texture, shape ...) to the objects seen. The same applies to other forms of sensory perception.
Iconic representations, such as visual images, and conceptual representations, which can be formulated with words, are sometimes distinguished. But this distinction is not fundamental. A visual image attributes visual qualities to all its points, so it is already conceptual. Conversely, a verbal description such as blue-white-red can be considered as an image of the French flag, because the words are aligned like the parts they represent.
To perceive is already to conceive. Perceiving, imagining and conceiving are essentially the same thing. It is always about producing internal representations that prepare for action.
When a being is perceived, it is always perceived with qualities or relations. A being without concepts, a kind of thing in itself, to which no concept is attributed, can not be perceived. Beings never come completely naked. They are always dressed with the concepts that perception has given them. (Kant 1787).
For an object to be perceived, there must be at least a detector which signals the presence of the object. The signal emitted by the detector is an internal representation of the perceived object. It also determines a concept attributed to the object: the quality of being detectable by this detector, or even more precisely, the ability to trigger the detection signal emitted by the detector. Detection automatically assigns the detected object the quality of being an object that can be so detected. The same detection signal can serve as a representation of the detected object and at the same time a representation of the concept attributed to this object, because the object is identified by its concept.
Beings are very often represented by the concepts attributed to them. For example, "the tree in the yard" is an expression that uses the concept of being a tree in the yard to represent a tree.
A concept is determined by the set of detection systems that signal the presence of an object by attributing this concept to it. This definition is not just for sensory perception and empirical concepts. It can be generalized because any information processing unit can be considered as a detection system. An information processing unit produces output signals from signals received at the input. An output signal can be considered as a detection signal of the input signals that produced it.
When a concept is defined by a series of conditions, which together are necessary and sufficient to determine it, a system of detection of the presence of the concept is defined at the same time, because the concept defined is detected by detecting the conditions which define it.
When a concept is defined by similarities with one or more examples, detecting the concept consists of detecting similarities and differences.
Even a single being can be identified by a concept, as soon as we are able to perceive it or imagine it as a single being, because to perceive or imagine it requires a system of detection, and because a such system defines a concept. For example I can have the concept of a person who is familiar to me because I can perceive it and distinguish it from all other people.
Concepts are often conceived as products of language. Concepts are signified by the expressions used to name them, and they are not known until they have a name. According to the accepted meaning in this book, concepts precede language. As soon as a perception system is able to detect objects, it automatically assigns concepts to them. Concepts are widely used by animals, whether or not they use language (Gould & Gould 1994). For example, all animals capable of fear show by their behavior that they are capable of detecting danger. The concept of danger is therefore one of their internal representations.
Are concepts concrete beings?
« Form finds itself identical at the same time in several places. It is as if you spread a veil over several human beings and say, « The veil remains one in its totality, when it is stretched on several things. » (Plato, Parmenides, 131b)
Beings are concrete when they exist in space and in time. Objects perceived directly with our senses, or indirectly through other material beings (systems of observation and measurement) are all concrete. Are there other modes of existence? Are there beings that are not concrete?
Are the concepts we attribute to concrete beings objects? Should we consider them as beings? And if so, as concrete beings? Do they exist in space and time?
Beings, or objects, are all that can be perceived, represented, or thought. As concepts can be represented and thought, they are beings like others. But can one perceive their presence in space and in time?
It can be argued that concepts are present whenever the objects of which they are true are present. The existence of the concept of horse is simply that of all horses. The concepts are manifested and revealed by the existence of the beings of which they are true and they exist at the same time. Thus conceived a concept exists concretely as soon as it is true of a concrete being.
More commonly concepts are conceived as abstract beings. Only beings that can be identified with a material substratum, a body, are considered concrete beings. But a concept is not a body, it may be true of many concrete beings, all of which have a body, but it is different from each of them. A body can not claim to be on its own a concept that is true of it, even if it is a concept that is true only of it, because the concept may be present in the head while the body is absent, only imagined. The concept may remain, Socrates, for example, while the body it represents has long since disappeared.
It can also be argued that concepts are present every time they are conceived, that is, whenever they are detected, or when this detection is simulated. The concrete existence of the concept can be identified with that of the signal that detects it. Thus conceived concepts exist concretely in space and in time, but not in the way of bodies, since they exist in a transient and dispersed way, only when a body detects them or imagines them.
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. But this modeling is insufficient because it does not explain the effects of preparation for perception.
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.
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 rules of inference 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.
To understand how schemas are used to control perception and imagination, we need a model that explains the will, the attention, and the formation of beliefs. It will be presented in a forthcoming chapter.
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. Similarly, by assembling concepts we can create conceptual representations of beings that have never existed and may never exist. By abstraction we separate the concepts from the realities they represent. By imagination we assemble them in new configurations and thus create fictions. Abstraction and imagination are creative. By separation and composition we can create all the conceptual possibilities we want. We discover the richness of the concepts and the creative freedom they give us.
We can say representations of fictitious beings that they are always false or always true. They are always false because to be true a representation must represent what really exists. They are always true because they make fictitious beings exist as objects of imagination or thought, and by asserting their truth it is only said that objects are thus represented or thought. The representations of fictitious beings are automatically true because they define the beings of which they are true.
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 do not exist?
The goals we set for ourselves begin as fictions. They will only really exist if we reach them. In this way fictions prepare us for action as anticipations of all that we could do. We discover our abilities through the imagination. But we also imagine beings that will never exist, fictions that will never be goals of action, purely imaginary beings and that will remain so.
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. In this way the imagination of fictions is an exploration of oneself. We discover our inferential abilities.
All knowledge of a being can help to know those who are like it, because part of what is true of it is also true of the others. As fictions are always more or less similar to real beings, they can serve to know them. By imagining fictions, we can exercise and develop our faculties of representation and inference on typical beings, fictitious but sufficiently similar to certain realities to be used for the knowledge of reality.
To prepare oneself for action, knowing one's environment is not enough, one must also know oneself, if only to know one's position and abilities.
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 soul is constantly a witness of itself.
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.
Reflection is the knowledge of oneself as a soul, that is, as a being who perceives, imagines, feels and wants.
Does reflection 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.
Reflection 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 is happening inside. All information about the agent, as it perceives, imagines, feels, or wants, is already present inside the agent. To develop its faculties of reflection it has only to exploit these internal sources of information. A sensory organ of reflection 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. We know ourselves by knowing our relationship to the being we represent. When we remember, we know that it is a representation of the past and we know ourselves as a being who remembers.
Reflection makes it possible to develop a silent knowledge about knowledge, because one knows knowledge when one knows oneself as a being who knows.
Reflection is fundamental to develop intelligence. For example, an agent can often imagine how to act before acting. As soon as it correctly anticipates the results of the actions it might undertake, it becomes capable of attaining them. By knowing itself as a being who imagines, so by reflecting on its abilities, it discovers how to develop them. Reflecting on our abilities makes us capable.
The nature of matter and the truth of perception
When we perceive an object with our senses we think we know it well. For example, if we see that the wall is yellow, we naturally believe it is really yellow. But is it not a mistake? All we know is that our eyes give us a sensation of yellow. Yellow seems to be on the wall but is truly in our eyes. It could even be that the wall does not exist, that we only have the illusion of a yellow wall. Should we conclude that we will never know the outside world, that we can only know our sensations and ourselves, that perception is always introspective?
The nature of matter is to interact with matter. The properties of a piece of matter (elementary particle, atom, molecule, solid, liquid or gaseous material ...) are always determined by its ways of interacting with other pieces of matter. Matter always does that, interact with matter, and nothing else. There is nothing more to know about matter than its interactions. When we know how material beings interact with each other, we know everything that is to be known about them.
We are sensitive to a being when it acts on our senses. Our sense organs are specialized to undergo the action of external objects. They are not sufficient to know all material beings and their interactions, but they still bring a lot of very useful informations. Observation and measurement instruments, and all detection systems that we can build, are like sensory prostheses. They extend the scope of perception. They enable us to know material beings which the senses are not directly sensitive to. They reveal other forms of action and sensitivity.
Matter can always be detected because its nature is to interact. As soon as it acts on another piece of matter, the latter is a detector. Our senses, complemented by all conceivable detection systems, enable us in principle to know all material beings and all their properties. Nothing can remain hidden. Everything can be perceived, because the nature of matter is to be perceptible (Dugnolle 2017).
The wall is really yellow simply because it is able to excite the sensation of yellow on our eyes, or on any other detector sensitive to yellow light. More generally, all the qualities and all the relations that determine the existence of a material being are detectable by other material beings. We do not have to fear that perceptions maliciously deprive us of what it seems to give us, true representations of perceived beings.
But this argument in favor of the truth of perceptions seems to prove far too much, since it suggests that all perceptions should be true. If the quality detected is still the quality of being so detectable, it follows that any detection is true, since what is detected is necessarily detectable. How can false perceptions then exist?
The possibility of falsity comes from the existence of a standard of truth. If a measuring instrument has not been correctly calibrated, it provides a false result. The result is false only with reference to the measurement standard. The same goes for perception. They can only be false if there is a standard that determines what should be perceived. In the absence of norms, they are always true, because they always reveal the effect of the object on our senses. Even a false perception reveals a truth about the object, because it is true that it can be so perceived.