| |
|---|---|
November 20, 1985
In doing philosophy, some philosophers have used the metaphor of rebuilding a ship while underway; the rest of the ship serves as the platform to permit operations on a selected plank. One can start almost anywhere, but one must first erect supports and cofferdams prior to removing the plank one wishes to work on. We may judge that some of the planks used for support may need replacement themselves, and the cofferdam may be none too sturdy. We take some risk in even undertaking the job. Some philosophers deny the premise that one can accomplish a shipyard overhaul while at sea. They could argue that the most that can be accomplished is minor damage control repairs, and that those are unsatisfactory, and at best, only temporary. We must learn to swim, and to fabricate a new vessel while swimming. Let us see.
Ask someone "How do you know that?" and see what kind of answers he or she responds with. "I don't know.", "I just know it.", "It just occurred to me.", "I inferred it (from ...).", "I read it (in ...).", "Someone told me.", "It happened to me.", etc. Some of these do not answer the question and some defer via communication to other sources, but the latter reports on the persons own experiences. In some ways we evaluate such reports as more reliable and 'closer to the truth'. Let us ignore deception, psychological and emotional problems, learning disabilities, and a host of other questions. (all planks which could stand refurbishing). "It happened to me." reports on one's memory of direct experience. We wish to avoid the problem of faulty (by whatever means) memory, and so choose direct experience as the answer to examine.
If we ask someone about their direct experience we get answers like "I see blue." or "I see a unicorn.". If we introspect we recognize that our direct experience is not the language we choose to describe it. How do we select the 'proper' language for expressing our direct experience? Philosophers have devised no pat answer for this hard philosophical question. I shant do so here either, but what do we know which can shed light on direct experience? If one accepts these direct experiences, along with the selection of language to describe it, as the basis of knowledge, one wears the mantle of empiricist. In examining carefully relations depending upon these direct experiences, and among them, one builds the edifice of modern science. But, what about these direct experiences? Introspection fails as a tool for looking at direct experience. One cannot look at a medium through the very medium in question; one does not see the structure one is seeing with.
We begin by assuming that we are new hands aboard a vessel. We assume that this vessel has some structure, and we have general notions about this structure (a tentative model). As we examine the vessel, we will revise our general notions and formulate specific notions about its structure. The specific notions comprise our tested model, while the remaining notions comprise our untested model.
In examining the vessel, we depend upon our direct experience, that is, what we experience through our senses. What we say about our sense data is not what we experience. If we take a strong sense for "communicate", we cannot communicate any of our direct experience to anyone else. The fact of the matter is that we encode our direct experience by abstracting into verbal levels according to our learned responses; we then transmit our codes. Our model must have implicit or explicit assumptions about structures relating our direct experiences, the codes used to express them, and the structures by which our direct experiences arise. In speaking about our direct experience, we must choose language with which to describe it. We usually do this implicitly by taking for granted the verbal associations we have learned to make; this is a plank of questionable strength.
It does not take an Indian Guru with decades of practice in meditation to experience the vast gulf between our direct experience and the verbal structures used to encode them. Most philosophers will grant that direct experience is non-verbal. When it comes to describing this non-verbal process of directly experiencing, we are faced with the exact same verbal encoding problem that we face in describing any of our sensory experiences. Introspection must be assumed to be an internal direct experience. The problem faced is that we must choose language structures to describe it. In fact, we select from among the language structures we already have at our disposal. Those language structures are not free from ontological commitments. Any word or statement we choose implicitly imports some structure into the description, and makes 'posits' about what we attempt to describe, and in this particular case, about the 'nature of' our direct experience.
If we posit that our direct experience is 'mental', we invoke the classical body-mind problem. By choosing 'mental' to encode our judgment of what direct experience 'is', we bring into play the entire superstructure involving a distinguishing between 'mental' and 'physical', and the question of their interaction. Yet such a choice represents a high level of abstraction about the process of experiencing. At the very first stage of attempting to describe our direct experience we import structure which makes a big difference in our ontology and metaphysics. But, while we have no choice but to import an ontology and metaphysics, consciousness of this abstraction allows us to choose from among alternatives.
We simply cannot directly experience another's direct experience, so we have no way to compare our own direct experience with that of another. We only compare the symbolic tokens we have learned to associate with the (idiosyncratic) direct experience. As a matter of fact, we bring our own experience to the decoding of language and compare our experience, the one we chose to associate with the word or statement being decoded, to our experience we intended to communicate about. We can only compare our experience we infer was intended by the speaker with our experience we intended.
Direct experience is decidedly non-verbal. Every statement about it is an abstraction from direct experience into pre-existing linguistic structures. Such verbal level abstractions represent selecting from the superstructure which ladder to climb, and which signal flag to fly on the yardarm at the top of that ladder. The most popular ladder is common currency among philosophers; 'mental' activity is divorced from 'physical' activity, and their interaction is, to date, inexplicable.
Because we have two different words, with a history of distinct usage, does not mean that, within the limited context of the body-mind problem, they actually do refer to two different things. The morning star and the evening star were not always known to have the same referent. On our vessel, the ladders may diverge and go into different parts of the super-structure, but they start at the same place on deck. Perhaps they simply go up different sides of the same part of the superstructure, but we think they are different masts because they look different from astern and from afore. In any case, traditional usage of terms has found to be 'wrong' in the light of modern knowledge in more than one case. Perhaps the use of "body" and "mind" is such an erroneous case.
Abstracting from direct experience, which is at non-verbal levels, to descriptions of those direct experiences, which is at verbal levels, requires choosing from among pre-existing verbal categories with their attendant implicit premises, ontological commitments, and metaphysics. When you say "direct experience is mental", you have already performed such an abstraction. The abstraction is into the pre-existing linguistic category distinction between mental and physical. Let us instead assume that mental activity is physical (an unpopular view), that my direct experiences are the movements of the pointer over the meter dial; that verbal level abstractions are the judgments regarding the significance of the position of the pointer in terms of the numbers pointed at. As a brief aside, it should be pointed out that we may reasonably infer that if mentation simply 'is' certain kinds of physical structure and processing, then it should be (theoretically) possible to instantiate mentation in non-organic forms, such as computers, a notion which most philosophers would reject. I do not; I view it as an ultimately testable conclusion.
We can posit that our direct experience 'is' a manifestation of neurological brain-function, and a direct experience is had for each particular configuration of memory data and nervous processing. In view of the non-verbal nature of direct experience and the abstraction process involved in formulating descriptions of it, the assumption that direct experience is physical has no more and no less credibility than to assume that direct experience is independent of physical structure. If anything, it should be more credible, since it does away with the body-mind problem as an immediate consequence. 'Mental' states are simply (unknown) 'physical' structures.
Whatever the case, we abstract from direct experience to a very high level of abstraction and choose some word or statement to make which purports to describe our direct experience. Consider an electrical multimeter. Opening the meter and applying its probes to its inner parts effectively adds circuits to the inside of the meter. As a result, the meter's response to its probes is altered. (Its output is now computed in terms of different circuitry.) What we read on the face of the meter cannot be assumed to be the same reading as we would get if we used a different copy of the meter. Such is the problem with introspection. What someone says they experience is not only not what they experience, but the abstraction process has been 'disturbed' and produces results which is even less reliable than abstraction about what is experienced. Unfortunately, one cannot observe the 'inner workings' of another's experiencing as would be possible in the case of the multi-meters. Therefore, there is no 'undisturbed' meter reading with which to compare the reading obtained by 'introspection'.
There may be some objection to the use of metaphor and analogy in describing these situations; these metaphors are much more well based than the covert metaphors used in describing mentation. We speak of things or ideas as being 'in' or 'before' the mind. Concealed in this usage is a container metaphor or a camera metaphor. When you 'grasp' something, a physical object metaphor is used. According to the mystics, "understanding" comes from a time when the initiate to a secret teaching 'stood under' the platform of the teacher and 'received' the teaching. All the paraphrases of "know" involve reports of information, metaphors, or reference to direct experience. As another aside, recent work on brain structure has suggested a greater dependence on physical organization than heretofore suspected. It has been observed that the representation of sensory reception and motor control areas in the brain form a, more or less, continuous mapping which is spatially organized with the same physical orientation as the normal adult posture. The strange new finding is that in bats, where the normal adult posture is with the wings back and up, the physical structure is also mirrored in their brains with the representations back and up (the reverse of all other mammals in which the limbs are mirrored forward and down). It may be that the significance of this aside is that there is no more to understanding elaborate and abstract spatial relationships than are captured in circuits which mirror those relationships physically. But that is a speculation beyond the purpose of this paper.
Less dependence on direct metaphor can be achieved with some of Korzybski's1 definitions. Terms which are used in such a manner as to have the same function at all levels of abstraction are called "multi-ordinal terms". Examples of such terms are "all", "never", "true", "know", etc. Terms which may be used in such a way as to apply to themselves are called "self-referential terms". Terms which may be used at many levels of abstraction are called "multi-level terms". Most multi-ordinal terms are both multi-level and self-referential. "Know" and "truth" are multiordinal terms, that is, they function at all levels of abstraction in more or less the same way, though their 'meaning' is different at different levels of abstraction.
At low levels of abstraction the terms 'verification', 'sense data', 'synthetic', etc., apply. At high levels of abstraction 'prove', definition, 'analytic', 'necessary', etc., apply. To 'know' what is meant by these terms requires knowing the level of abstraction intended by the user of the term. (Note that there seems to be circularity in this formulation; it is both unavoidable and ambiguous regarding the intended level of abstraction. I hope it becomes less ambiguous as we proceed.) What is intended by the use of multi-ordinal terms, such as 'true', 'know', etc., depends upon the level of abstraction intended by the user of the term.
The same statement can 'mean' many different things, depending upon the intent of the speaker. Language and communication fundamentally spans many levels of abstraction; these levels are ordered by the directions 'more abstract' contrasted with 'less abstract', but have no explicit designations. There is an ordinal scale, but no nominal scale for describing levels of abstractions (except for levels designated by Korzybski's structural differential).
The context of a use of the term 'know' requires an agent; and to be an agent is to have a purpose or intent. "Purpose" is not (normally) a multi-ordinal, or multi-level term. Applying the multi-ordinal term 'know' in the limited level context of a purpose provides a means to select the level of abstraction and thus instantiate a restricted use of the term. The restricted use depends upon the level of abstraction required by the purpose of the agent. Lower levels of abstraction require reports of direct experience; higher levels of abstraction require second hand reports, arguments or proofs, etc. When a philosopher says something "is true", we do not have any way of limiting the level of abstraction for inferring what is intended by the speaker, but we have confidence that, if pressed, the speaker could provide a description of the specific instantiation of "know" in his case.
Epistemology looks at the specific instantiation; whereas, metaphysics looks for relations which are independent of any such instantiation, in short, metaphysics deals with the abstract levels in which all information about the specific instantiations for 'true' or 'know' is abstracted out.
"Is true" (metaphysics) corresponds to "I know" (epistemology). It is always clear when someone says "I know..." that there is an agent involved; it is not clear when someone says "... is true." that an agent was ever involved. In empiricism, all "... is true" statements are limited by the preferential bias of the purposes of empiricists to lower levels of abstraction. In other words, when an empiricist says "... is true", he intends for you to understand "We know ... ", and that you assume that he also intends you to understand that every (most) "We know ..." statement can be ultimately translated into statements about reports of direct experience (some turn out to be untested assumptions).
What do we know about our mechanism of direct experience -- perception? What can we infer about it? Are there philosophical consequences? The epitome of empiricism, modern science, is constructed on perceptual experience. The philosophy of its structure involves premises which have been tested against untested assumptions. The method of testing statements in a theory, or premises, depends upon the level of abstraction of the statement. More abstract statements are tested by primarily logical means; less abstract statements are tested by primarily observational means. Still, there are the untested statements.
In metaphysics, the untested assumptions are made explicit, and the logical relations among abstract premises serve as the only criteria for determining consistency. The assumptions are accepted or rejected, and consequences logically dependent upon them are examined. In science, an appeal to perception is also made for acceptance or rejection of less abstract premises. According to Carl R. Popper, the best we can do is to operate from a model which is not yet disconfirmed; our model is not truth, it is not 'it', it is not any 'final' answer; it is merely the best model we have yet found.
In regard to Quine's doctrine about the Indeterminacy of radical translation2, it seems clear that the level of abstraction intended by the native who assents or dissents may differ from the level of abstraction intended by the field experimenter, not only in the single incident, but in the overall scope of the experiment. Different translation manuals would be obtained for different levels of abstraction as limited by the varying purposes of the native. Our models would differ primarily in their postulate or inference regarding the native's purpose in participating in the communication experiment, and in all consequences of these differing assumptions, yet would correctly account for all observed speech utterances.
A very large superstructure of assumptions is taken into the field. How large can be apprehended if we substitute "dolphin" for "native". Now, that's radical translation for you.
Also, indeterminacy of radical translation directly parallels Popper's philosophy of science's dictum that we only have models. There can be equally effective, though different and incompatible, models which account for the observations.
It seems to me that Quine has taken Popper's philosophy of science3 and applied its general structure and form to the problem of building a model for an unknown language. All the dictums about the world view espoused by Popper's philosophy of science apply. Quine's doctrine of the indeterminacy of radical translation is also a translation of Popper's doctrine that we never know 'reality', that the best we can do is have a set of equally effective but conflicting models of it.
So Quine's doctrine on the indeterminacy of radical translation reflects both Popper's philosophy of science, and the multitude of levels of abstraction in communication.
Popper, however, does point out that there are many implicit and untested assumptions upon which our model depends, not the least of which is a collective assuming as to how 'what is perceived' is to be 'cut up into blocks'.3 At the base of our theories are untested assuming regarding the fundamental building blocks of perception. We assume that the 'objects' we perceive as gestalten are, and less obviously, that everyone perceives the same 'objects'. Psychologists know better, and buttress their position as supported by the classical perceptual illusions which are seen to flip-flop between two different reported 'objects'. In such cases, it is argued that the unstable perception results from an 'ambiguous' collection of lesser perceptual element which are themselves stable, and that the gestalten formed take place in our 'mind' (nervous system).
Examination of this assumed structure may shed some light on some philosophical problems. One assumption involves a lack of regularity. At the level of the illusion, the 'objects' of perception are said to be "in the mind", while at the lower levels, the 'objects' of perception, namely the lesser perceptual element, are said to be in the things perceived. To allow such a direct contradiction between different levels of perception clearly identifies 'perception' as not free from equivocation.
Arguments which depend upon the two different ways of apprehending perception commit the fallacy of equivocation. We cannot have it both ways. Either the objects of perception are in the 'mind', or they are in the thing.
If the object is in the thing, then we must allow each thing to have more than one object. Another consequence of this is that objects are perceived whereas things are not perceived. If the objects are not in the 'mind', how then are things different from objects? Naive realism says they are the same. But, this gives rise to the problem that the same thing is two different things. Allowing objects to be in things leads to unacceptable problems. So, objects are not things, and it seems that there is no way to know what things are.
Suppose, then, that objects are in the 'mind' (nervous system). How can it be that our lesser perceptual elements are less subject to ambiguous interpretation than are the more abstract gestalten? It depends upon the number of lesser perceptual elements and into what combinations they may be placed.
Suppose a visual field is divided into two areas, one black and one white, against a grey background. This structure can be apprehended as a two colored figure on an uncolored background, as a black object on a white background, as a white object on a black background, or as an uncolored surface marred by the presence of contrasting colors. There are two distinctions; one separates the figure from the background, and one separates the figure into two parts. As just enumerated, it can be seen that with only two distinctions there are 4 possible combinations. With three distinctions there are 8; with 4, 16; etc. As the number of 'simple' perceptual elements increases, the number of ways of drawing a simple distinction among those elements rises as 2 raised to the power of the number of elements. And, there are more ways to structure the perceptual elements than just drawing a simple distinction!
A further question about this view would be how small the perceptual elements can get. If we were arguing from a conceptual point of view, then Aristotle's infinite divisibility argument would go here. Fortunately, we have the empirical bias of science and the appeal to what has been observed to guide us. These suggest that we examine the way the human nervous system is used in responding to visual stimuli and determine what are the smallest perceptual elements of a given stimulus. Unfortunately, this direction does not give great promise; any perceptual element can be subjected to 'enhancement' such as using a microscope, or some other device to enlarge the object, and increase the apparent number of 'smallest' perceptual elements. Measurement and reliability theory specifies that one should take the least division in an instrument as the uncertainty of any measurement. One specifies that the least perceptual element is the size of the smallest division on the measuring tool being used.
If the ruler is marked only in whole centimeters, and we place it up against an object and the edge of the object falls between the 6 and the 7 centimeter mark, we could equally choose to say that the object is 6 +/- 1 centimeters long, or that it is 7 +/-1 centimeters long. It's length falls into the 6 to 6+1 range, and into the 7 to 7-1 range.
Well, you say, get a better ruler? Sooner or later, we get a ruler so finely divided that our unaided eye cannot distinguish into which division the end of the object falls. Well, you say, get a magnifying glass? Ok, but now that we can see into which division it falls, the divisions appear large enough that we go back for a better ruler.
Sooner or later, we get a ruler so finely divided that our aided eye cannot distinguish into which division the end of the object falls. Now, you say again, get a better magnifying glass. Ok, but now that we can again see into which division it falls, the divisions appear large enough that we go back for a better ruler. [Read this paragraph again.]
Pretty soon we catch up to another guy doing the same thing. His name is Aristotle and he's in search of infinite divisibility. (Also, modern physicists keep building bigger and better particle accelerators to probe the finest structures of the atom.) Why does this circularity occur? Because we have forgotten that the eye itself is a measuring device, and we are not examining its structure, we are using this unknown measuring device to perform other measures.
Examination of the structure of the eye shows a certain non-uniform granularity. There are cells (rods) which simply respond to the presence of light above a minimum threshold. A rod cell cannot report on any divisions in the stimulus, it merely reports the presence or absence of stimulation. In metaphysical terms, it can only report that there is or there is not something there. It is combinations of these reports that are made up into gestalten in the brain-mind.
An exact study of perception by examination of the detailed construction and operation of the human eye is simply not possible; however some philosophical import can be inferred from a simplified model of such a structure. Marvin Minsky did just such a study and my view here is influenced by his efforts.4
The simplest perceptual device just reports yes or no to its stimulation. We may interpret that stimulation as 'meaning' something "is there" or "is not there". (Metaphysical realism perspective.)
Combining some of these into patterns can yield different 'objects'. Consider these two patterns, which are clearly different, but which, in the present form, do not make any particular sense to us.
...#.....##......#......#......#.....###.. ..###...#...#.....#.....#.....#.....#####.
However, rearranging them makes much more sense. The information content is exactly the same in the two patterns. It is the arrangement in terms of our (learned) structure which makes the difference. We can, after a time, learn to make sense of the first two patterns directly.
...#... ..###.. ..##... .#...#. ...#... ....#.. ...#... ...#... ...#... ..#.... ..###.. .#####. What of this structure? .#.#.....##..#.#...#.#.##..#.#.#...#.#.#.. or: .#.#... ..##..# .#...#. #.##..# .#.#... #.#.#..
It's random, and doesn't make any particular sense to us in either form.
A tentative philosophical conclusion of such a perception structure is that only the least elements of perception can be said to 'be there' or not. Higher level combinations of structure are simply patterns of combinations of stimulation which are interpreted by the structure of the nervous system.
For example, a structure (circuit) to recognize the first pattern can be constructed as follows, using two AND circuits and one NOR circuit. I will call this structure the "1-seer".
Input AND gate: - Figure 1.
Inputs are ANDed together [ | || | | | ||| ] [------------------------------------------] [ | ] Output is + if all inputs are #. Input NOR (NOT OR) gate: - Figure 2. Inputs are ORed together and result complemented [||| ||||| |||||| |||||| |||||| ||||| ||] [------------------------------------------] [ | ] Output is - if any inputs is #. (or conversely, + if all inputs are .) Next, feed the outputs of the two input circuits into an ANDer.
Decider: - Figure 3. [|.|] [---] | Output is + if both inputs are +. Use the 1-seer to compute the results in the following three cases. Case 1. Input ...#.....##......#......#......#.....###.. [ | || | | | ||| ] [------------------------------------------] [ | ] AND Output > + \\ -------------------- Input \\ ...#.....##......#......#......#.....###.. | [||| ||||| |||||| |||||| |||||| ||||| ||] | [------------------------------------------] | [ | ] / NOR Output > + -------------------- | / Decider Input > + + [|.|] [---] | Final Output > + (There IS the figure "1".)
Case 2.
Input ..###...#...#.....#.....#.....#.....#####. [ | || | | | ||| ] [------------------------------------------] [ | ] AND Output > - \\ -------------------- Input \\ ..###...#...#.....#.....#.....#.....#####. | [||| ||||| |||||| |||||| |||||| ||||| ||] | [------------------------------------------] | [ | ] / NOR Output > - -------------------- | / Decider Input > - - [|.|] [---] | Final Output > - (there IS NOT the figure "1".) Case 3. ...#.....##......#......#......#.....###.. [ | || | | | ||| ] [------------------------------------------] [ | ] AND Output > - \\ -------------------- \\ ...#.....##......#......#......#.....###.. | [||| ||||| |||||| |||||| |||||| ||||| ||] | [------------------------------------------] | [ | ] / NOR Output > - -------------------- | / Decider Input > - - [|.|] [---] | Final Output > - (there IS NOT the figure "1".)
In case 1, the 1-seer 'saw' a "1" and reported a + (yes). In case 2, the 1-seer did not see a "1" and reported a - (no). In case 3, the 1-seer did not see a "1" and reported a - (no). Case 3 is unusual in that we know, (metaphysically), that the "1" 'is there', but is simply not correctly aligned with the 'eye' of the 1-seer.
This illustrates that applying the metaphysical judgment 'is there' or 'is not there' is not appropriate for perceiving systems. The perceiving system can report "I see ...", or "I don't see ....", but cannot say that what is seen, (the object), is what is there, (the thing).
Moreover, as the problem of measurement shows, what 'is there' is illusionary in that the measuring device can only record yes or no; it cannot report on structures too fine for the perceiving structure to discern. Whatever fine structure 'is there' is lumped together into a single average value which either is enough to trigger a yes, or is not. If we say something 'is there', we naively assume that the finest level of structure providing stimulation corresponds directly to the finest level of structure perceiving it, and moreover, that there is a point for point correlation between points of stimulation and points of reception. Both of these assumptions seem to be without justification.
For an illustration, consider the fine structure:
#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.
And the perceiving structure whose smallest element is as large
as 5 of the fine structure elements. 3 of 5 #'s are enough to
register a +. Further, suppose there is fine structure space
between perceiving elements; the diagram is:
Figure 4. [-----] [-----] [-----] [-----] [-----] [-----] [ | | | | | | ] [ +/- +/- +/- +/- +/- +/- ] [ ----------------------------------------- ] I see [//////] (interpretation).
Compare the following two cases, when the perceiving structure reports opposite results to what we know is metaphysically the same fine structure.
Case 4: #.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#. [-----] [-----] [-----] [-----] [-----] [-----] Three .'s [ | | | | | | ] outweigh [ - - - - - - ] two #'s. [ ----------------------------------------- ] I see [------] (dark). Case 5: #.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#.#. [-----] [-----] [-----] [-----] [-----] [-----] Three #'s [ | | | | | | ] outweigh [ + + + + + + ] two .'s. [ ----------------------------------------- ] I see: [++++++] (bright).
If we observe the stimulation patterns on our receptors, we can construct a model to account for that pattern. The model will be limited by the 'size' of the smallest sensor. Notice that in the following two case, neither of these structures can see what the 1-seer can see because the "1" is composed of structure too fine for the 'eye's' of cases 6 & 7.
Case 6:
..###...#...#.....#.....#.....#.....#####.
[-----] [-----] [-----] [-----] [-----] [-----]
[ | | | | | | ]
[ - - - - - + ]
[ ----------------------------------------- ]
I see: [-----+] (1/6 bright)
Case 7:
..###...#...#.....#.....#.....#.....#####.
[---] [---] [---] [---] [---] [---] [---] [---]
[ | | | | | | | | ]
[ - + - - - - - + ]
[ ----------------------------------------- ]
I see: [-+-----+] (1/4 bright)
Also, by devising a finer sensor, that is, one with a greater resolving power, our model may need to be changed to account for finer levels of distinction among patterns of stimulation.
It is not, in principle, possible to have a perception system which responds to what 'is'. A perception system simply responds. Our direct experience is the output of such a perception system, albeit more structured and more complex. Our direct experience is the response of our perception system, not what stimulated it. Examination of the structure shown in figure 4 reveals that many fine structures will simply not be seen, and many other structures will be seen as 'the same'. Each perception segment cannot tell the differences among inputs consisting of any configuration of 3 or more "."s or any configuration of 3 or more "#"s; it only 'sees' a "-" or a "+" respectively (and that is not the "." or "#" of which the stimulation is comprised). Fine structure differences will be differences that 'don't make a difference' to us.
In information theory these differences are referred to as equivocation in input. Consequently, what we experience is the response of our perception system and its fundamental equivocation. All our 'knowledge' attributed to direct experience is based upon this equivocation. How can we have any confidence at all in our direct experiences? The answer provided by modern science is that natural selection weeded out those perception systems which equivocated among life and death differences; the only progeny were from those perception systems which did not equivocate between differences which were significant to survival. It should be noted that the ability of nervous systems to respond at the levels of structure illustrated herein takes place in the most rudimentary nervous processes, at the lowest levels in the phylogenetic scale. Even bacteria can detect the difference in a chemical gradient so as to swim toward food and away from toxins. We can have relative confidence that our direct experience of environmental conditions directly relating to our survival will be most the reliable, as long as we remember the dictum "Direct experience entails abstraction and abstraction entails equivocation".