This series of essays has been rewritten from a Master's thesis defended at Temple University in 1970.

The Philosophical Foundations of Human Cognition (Section III)

©2004 Edward G. Rozycki

Part 4: The Things We Recognize


edited 7/26/15


(Section I) --------------



knowa and knowe

The Truth Condition

Enabling a Knowledge Claim


Epistemically Fundamental Verbs

Direct Reference

Epistemic Substitutes and Epistemic Sources

The Possibility of Illusion


Section Endnotes

(Section II) ------------


Contrasting 'recognize' and 'identify'

Recognition and Recall

The Primacy of Recognition


Conflicts of Recognition and Identification

Cognition is Recognition

(Section III) ------------


Confrontable Characteristics
Qualia, data ?

Recognition Equivalence
How things are the same

Goodman's Identity Theory
Indistinguishable vs substitutable

Systematizing the r/e-class
Paradigms and deviants

Goodman's Theory Continued

Knowing What a Thing Looks Like

Doesn't handle deviance

Dimensions of Individuation
r/e-classes are strange

Section Endnotes


(Section IV) ------------

PART 4: Continued

Multi-leveled Individuation

NIE's and Classes

Physical Objects as NIE's

The Act as NIE

Following a Rule

Recognizing Reconsidered

Goodman's Theory -- Conclusion

The Application of Concepts

To Conclude



Confrontable Characteristics (Qualia, data ?)

To recognize X, we must be confronted with X. To recognize X by some characteristic c, c must confront us, c must be a confrontable characteristic What characteristics are not confrontable? In particular, historical ones. In general, any characteristic which presumes the identification of entities other than the X we are confronted with and the Z (or Y) we are trying to identify it as. Characteristics such as: 'uniqueness', 'being an original', 'part of', 'American', 'first'. One cannot recognizer something to be unique or original or part of Y or first. One must first recognizer it to be something which is attributed to be unique or original, etc.

We often speak loosely saying things like "He was recognizably Armenian, Russian, ...etc." What is meant is that a confrontable characteristic of the person was recognized and presumed sufficient to identify him as Armenian, Russian or whatever. In fact, no one can be recognizably Russian, my brother, 20 years old or the last of the Mohicans. Were such characteristics recognizabler we would not have problems identifying as such counterfeits, impersonators, etc.

Recognition Equivalence (How things are the same)

If two objects are recognizably indistinguishable, say, a painting and a good copy of it, we will speak of them as being recognition-equivalent or say that they belong to the same recognition-equivalence class, abbrev. r/e-class. The distinction between a painting, a copy and a forgery is not formulable in terms of confrontable characteristics. All might be quite recognizablyr indistinguishable but an original is distinguished by its having been produced in a certain favored way, a copy is historically speaking uninteresting, and a forgery has the wrong history. None of these distinguishing characteristics can present themselves to us in a confrontation.

But confrontability may be merely relative to the thoroughness of examination given the object. What to the naked - or perhaps merely untutored - eye are two recognizably indistinguishable objects may - we would hope, have to - exhibit distinguishing, i.e. confrontably different, characteristics when subjected to a technically aided analysis. But - to reiterate - that an object is distinguishable from another at some level of analysis does not serve in and of itself to establish the authenticity of one of them. It is the presumed sufficiency of some comfrontable characteristic to identify the alleged original as the object with the "right" history.

Now X and U belong to the same r/e-class because they are confrontably indistinguishable. Suppose they are copies of the same book. What if we then acquire copy T, which but for a dog-eared page would be recognizablyr indistinguishable from X and U- - need we put it into a different r/e-class? We may, if it serves our purposes, e.g. we deal only in perfect copies of a particular book. But we can admit T into the same r/e-class as X and U with, so to speak, accompanying instructions to disregard the dogearring T is recognizablyr distinct from X and U but recognizablya indistinguishable from them.

We build up our r/e-class with an inner-core in which any member is recognizablyr indistinguishable from any other within the inner core and an outer-core in which any member is recognizablya indistinguishable from any other member in the r/e-class.

To illustrate: Consider an original watercolor, X. Any copy of X recognizably indistinguishable from X belongs with X to the inner core of X's r/e-class. But aren't U and V not the same? Why shouldn't they have their own r/e-classes?

Because - and this has not been stated before - we are trying to preserve a certain conventional distinction, i.e. that between X and other watercolors by the same artist.

In art, phenomena are individuated as works In our everyday experience phenomena are individuated as objects and events With respect to certain individuals, phenomena are subindividuated, so to speak, into acts. Our language gives us our traditional dimensions of individuation. We can, of course, overlook these or refine them as it suits our particular interests.

Both kinds of recognition play a part in our recognizing a confronted X to be a Y. r/e-class theory, as I will develop it here, is an attempt to elucidate our ability to "see" similarity in difference, to make allowance for deviance.

But how much deviance can be allowed?

Up to the point that a desired ability to individuate things breaks down. We continue this after reconsideration of Goodman's theory of work identity.

Nelson Goodman's Identity Theory (Indistinguishable vs. substitutable)

We saw earlier that a consequence of Goodman's theory was to make recognition epistemically irrelevant. What I will argue now is that Goodman's theory presumes the epistemic relevance of recognitionr.

That a performance p is a performance of the work W depends entirely upon its derivative score's being identical with the score of the work, according to Goodman. Recall that the derivate score is a description of p made in a notational system in which only the musically notatable characteristics of p are recorded as having presented themselves in the course of the performance to the transcriber of p.

The condition for the identity of scores is notational identity. It is both necessary and sufficient.

A score is a compound character in a notational system, i.e. it is composed of atomic characters which admit of no further analysis, e.g. whole notes, rests, sharps, etc. It is at the level of deciding whether a given instance is to be identified as an instance -- a replica -- of an atomic character in the notational system that Goodman falls back on our ordinary language distinctions.

For i to be a replica of j , i.e. for i and j both to be instances of the same atomic character, there must be either a high degree of similarity of appearance so that they are obviously instances of the same character, -- although

"there is in general no degree of similarity necessary or sufficient for replicahood." (Languages of Art, footnote, p.31"

or, we must decide that i is a replica of j from context and on the principle that there can be no unclassified or unclassifiable instances. Otherwise, our notational system will be theoretically inadequate as a notational system.

At the atomic level Goodman either recognizesr i as a replica of j or recognizesa i as a replica ofj, respectively. In virtue of these recognitions i is identical with j . But there are two kinds of identity here and Goodman confuses them. They are -- again respectively -- identity by recognizable indistinguishability and identity by mutual substitutability (i may be substituted for j and vice versa since they have been recognized as instances of the same character). But if i is recognizablyr indistinguishable from j and therefore a replica of j, i is thereby an instance with j of the same character and they are both mutually substitutable.

Thus recognizable indistinguishability entails mutual substitutability. But not the converse, i.e. if I and j are mutually substitutable they are not necessarily recognizably indistinguishable. Goodman does not observe this but uses a notion of identity whereby if i is identical to j anything predicable of i is predicable of j. Ignoring identity distinctions consequent to the distinction between 'recognizer' and 'recognizea' is a factor which necessitates Goodman's rejection of deviance. We will continue this discussion after the next section.

Systematizing the r/e-class (Paradigms and deviants)

In this section I will try to relate the notion of r/e-class to the more familiar notion of paradigm.

For any r/e-class, members of the inner core are paradigms for that class. Members of the outer core are deviants. Paradigms are recognizablyr equivalent. Deviants are recognizeda equivalent. r/e-classes are defined extensionally, i.e. by actual objects, since recognizabler indistinguishability is a criterion of membership. Although actual objects define (but do not well-define -- see note 23 below) the r/e-class. there may exist no objects that are either paradigms or deviants. That is to say that there may exist no objects that belong to the, say, outer core of an r/e-class (as when we have something of which there are no copies.)

Conversely, we can imagine something being destroyed so that only copies or replicas of it remain. Still, we would want to retain the distinction between copies of the destroyed object and copies of other objects. That there exists neither paradigm nor deviant means that nothing is recognizable -- in either sense -- as belonging to that class.

We may have conglomerate r/e-classes -- the most usual kind -- with recognizablyr distinct paradigms. This is a conflation by performative recognition of two or more basic r/e-classes done because our language is less discriminatory than our senses.

Consider the r/e-class of teacups. It is conglomerate. Teacups come in different patterns and we could imagine circumstances in which we would differentiate an r/e-class for each pattern, e.g. we deal in tea-services. In other circumstances we could not however reject a teacup of a particular style as being a teacup if all we had specified when asking for it was a teacup.

The words style, type, pattern, kind, species, etc. provide us with means to designate less and less conglomerate r/e-classes down to the level of actual recognizabler indistinguishability.

What do any two objects correctly called , say, a Y have in common? They are both recognizeda as a Y by speakers of a language with that category. In terms of confrontable characteristics they may be entirely different, but no man is less a man for being disguised as a bear. What is it then that relates X as a deviant to Y if they have nothing recognizablyr in common? A deviance justification, i.e. a statement of explanation as to why recognizabler distinguishability can be disregarded.

Learning what an object is, is as much learning why it is what it is, as what it is.

It is generally conceded that for most classes of objects there are no necessary conditions for membership, because for a class Z and alleged necessary condition of membership n, specifying a Z without n is not contradictory, quite imaginable and may actually be commonly encountered. I would add to this that there can be no sufficient characteristics for class membership either since for any set of alleged sufficient characteristics for membership in class Z, we might encounter a Y, not a Z, which possesses all the characteristics of the set. 23

Of course these might all be quite deviant so that one might say -- presuming normal conditions to obtain -- sufficient (perhaps even necessary ) conditions of class membership can be specified, i.e. conditions stated in terms of confrontable characteristics of objects.

(The necessary and sufficient conditions for a given object X to be a Y are

1) There exists a language having the category Y;

2) X be recognizabler as a paradigm Y and/or

3) X be justifiably recognizablea as a deviant Y).

We might try to set up sufficient sets of recognitionr (confrontable) characteristics for paradigms only. But conditions change and the deviant may become the paradigm. Our language has the wherewithal to maintain its usefulness despite this.

The notion that the paradigm should serve as a pedagogical model is understandable when we consider that paradigm recognition is recognitionr and requires only normal capacities of perception and memory. Deviance recognition is recognitiona and demands ability to justify in addition to that of recognizingr a particular as the same from occasion to occasion. Paradigm recognition demands minimal intellectual ability.

One learns first to recognizer the paradigms, then to recognizea and consequently recognizer some deviants for each paradigm. One then learns to abstract the "way that they can deviate" so as to be able to apply the original categories to new cases. It is not surprising that two common activities of small children should be asking what things are, and asking why.

Goodman's Theory of Identification - continued

Goodman holds that a theoretical requirement for any system to be a notational system is that for anything alleged to be an instance of some character in the system, that it be identifiable as an instance of a particular and unique character in the system. Let us call this the principle of individuation i.e. within a genus, an individual must be identifiable as to its species. More commonly, once one picks a dimension of individuation appropriate to a particular, the particular must be classifiable within one of the categories available in that dimension.

Forced by the principle of individuation to identify, Goodman identifies replicas either on the basis of recognitionr, where some i and j are recognizablyr indistinguishable, or on the basis of recognitiona where they are merely mutually substitutable. He then overlooks the distinction and thus comes to the conclusion that deviance is sufficient to withhold identity. He writes,

"If we allow the least deviation, all assurance of work-preservation and score preservation is lost; for by a series of one-note errors of omission, addition and modification, we can go all the way from Beethoven's Fifth Symphony to Three Blind Mice. (Languages of Art, p.187)

What is Goodman's logic here? Two works are identical iff scores derivate of their performances are identical. Suppose we allow deviant scores to be identical to the original scores by some criterion other than notational identity. Consider work W and a deviant d(W); and work V and a deviant d(V). Now by hypothesis W and d(W) are identical by some other criterion than Goodman's; likewise V and d(V). Suppose the scores d(W) and d(V) are notationally identical, then (and this is Goodman's presumption) d(W) and d(V) are identical. But identity is a transitive relation, therefore W is identical to V.

Goodman fails to distinguish between the identity of performative recognitiona that exists between the work and its deviants, and the identity of recognizabler indistinguishability that may exist between the deviants of the two works. However, this problem of work equivalence is Goodman's, not ours and is consequent to his choosing a criterion of identification for musical works which derives entirely and solely from the musically notatable characteristics of the performances. He must identify d(W) and d(V) as work-equivalent. We would ordinarily just say, "So what if they sound alike? They're different works. The reason they sound alike is..."

We take into account other things besides the musically notatable characteristics of the performance. We take into account mistakes, interruptions, popping strings, splitting reeds, etc. If we were forced to or interested enough to work-categorize every miserable performance that occurs we most likely could. However, between Beethoven's Fifth Symphony and Three Blind Mice lies a vast barrier of unrecognizable musical non-entities -- so far as we are normally concerned.

What considerations do come into our normal work-identification will be discussed below in the section entitled Multi-leveled Individuation.

Knowing What a Thing Looks Like

In light of our above discussion we can see that knowing what a thing looks like is a complex notion, especially when we consider that it is knowing what a thing looks like in varying circumstances. Normal conditions are paradigm conditions, i.e. those considered appropriate for the presentation of certain objects or events. The conditions are those which promote clear and distinct individuation to the degree desired in the presentation. Thus, paintings are not hung in dark closets nor is chamber music played in ballparks. Nor are children taught colors under green lights; nor writing,wearing baseball mitts.

Normal conditions are those "normally" encountered and the concepts of our language give us much leeway in discriminating a tree from a bush, or hot from cold. Our perceptions are almost always finer than the categories provided in our everyday language require for correct categorization - in normal circumstances. To know what a thing looks or feels or smells or tastes like is to know within a range of conditions appropriate to each mode of sensing what a thing looks or feels or smells or tastes like.

But to know -- in general -- what an X looks like is not merely to know what a group of particular X's look like in varying, circumstances. It is to know whether or not the looks of a thing can qualify or disqualify it for inclusion in the class of X's.

Abstraction (Doesn't handle deviance)

It is sometimes suggested that coming to learn how to use class concepts is learning to abstract from particulars. This is misleading. If by "abstraction" is meant a process where by individual characteristics among objects are ignored and common, supposedly class defining characteristics are attended to, then abstraction is not the means - the sole means - by which we come to learn class concepts and apply them to the world. Abstraction cannot explain how we come to handle deviance - anything may present itself as deviant depending on the circumstances in which we confront it. Abstraction may well work in normal circumstances with paradigm cases, but let us not, in examining abstraction as a step in the acquistion of intellect overlook the role of justificatory behavior in concept learning.

Dimensions of Individuation (r/e-classes are strange)

The r/e-class is only well defined in the context of a dimension of individuation. Given a particular object, 0, we may choose to individuate in terms of everyday English. It is, say, a hammer. As a hammer it must be differentiated from horses, screwdrivers, telephones, etc. A hammer with a broken handle or a misshapen head must remain a hammer and not become any of the things we contrast with hammers. If it becomes recognizablyr indistinguishable from an object in some contrast category, we recognizea it as a hammer if it suits our purposes, as for example an archeologist might categorize something as a hammer which looks like and can be used as a crowbar; he justifies the categorization by saying, "It was smashed."

Then again we may have a specific task to perform and 0 might be individuated in terms of hammer-types. Thus we would distinguish among and contrast a Mason's hammer with a ballpeen, or a claw, or a sledge.

But 0 might be a Hittite artifact. We as antique dealers care little about its being a - to judge by its looks - crowbar, for Hittite artifacts of any kind bring high prices. 0, as a hammer, contrasted with crowbar; its deviants were broken or misshapen hammers, 0 as a type of hammer, contrasted with other types of hammer; its deviants were broken or misshapen hammers of the same type. 0 as Hittite artifact contrasts with Sumerian artifacts or Attic artifacts; its deviants are -- perhaps - -Hittite artifacts showing Sumerian influence.

But 0 is an object, an individual in its own right. It is in one dimension of individuation this hammer; in another, this artifact. What this object contrasts with is that object. What may vary or deviate is its appearance in one set of conditions or another.

Consider N, a book. As such it contrasts with newspapers. Its deviants are large thin paperbacks which look like magazines, or folios, or newspapers.

But N is a novel. Here it contrasts with a biography, or textbook. Its deviants are historical novels using non-fictional personages, or overly didactic novels, etc.

But N is a particular novel. It contrasts with different novels. Its deviants are plagiarisms, play adaptations, movie scripts, Reader's Digest digestions, etc.

N, as an object, contrasts with other objects, any other object. As object it does not contrast with a newspaper as newspaper, a textbook as textbook, or other novels as novels. It deviates in appearance from situation to sitiation depending on handling and conditions of presentation.

Let's consider some human behavior. John moves in, a certain manner. We record his movement on motion picture film, Let us indicate the particular movement defined by the beginning and end of the film as Z. Z is at one level of individuation a body in motion. It is, say, body M executing a motion N. It contrasts with other bodies executing N, or with itself executing some other motion. Its deviant might be M executing N in conditions -- total immersion in water -- which damp the motion.

Z is in another dimension of individuation a particular sequence of muscle contractions, call it a-b-c-d. It contrasts with b-c-a-d. Its deviants might be an a-b-c-d sequence with a simultaneous, rather than sequential, b-c contraction.

Z is John moving his arm. It contrasts with John's moving his leg. A deviant might be a movement resulting from a nudge John receives the impulse of which John utilizes to initiate his own arm movement.

Z is John reaching for a book. It contrasts with John's throwing a book. A deviant might be a sort of semi-reflex action on John's part when he notices, out of the corner of his eye, a book begin to fall from a shelf.

Z is John getting himself something to read. It contrasts with getting himself something to throw, eat, etc. A deviant might be John's reaching into the refrigerator because he absent-mindedly mistakes it for a bookcase.

So much for examples illustration dimensions of individuation and their possible deviants. In the next section we will discuss multi-leveled individuation.


23 This amounts to a theory of the logic of classes. Traditionally, a class is defined by a set of properties in terms of which a decision is made to include or exclude individuals insomuch as they do or not possess the defining properties of the class. The logic of r/e-classes presented in this thesis is such that individual objects or events can be use to define classes insomuch as they are by some tradition or to some purpose, e.g. linguistic, moral, aesthetic, specified to be paradigm or deviant and interrelated by explanations which account for the incongruity between that which is recognizablyr equivalent and recognizablya equivalent.

Such classes are not well defined in the abstract but only in terms of some dimension of individuation, i.e. whether we are considering the same object to be a ball as opposed to a bat or a sphere as opposed to a cube or a toy as opposed to a weapon or an inedible as opposed to a foodstuff. I maintain that this is the actual logic of our language and clarifies Wittgenstein's notion of "families of concepts."

For a visual representation of the difference between traditional and recognition-equivalence analysis see, E. G. Rozycki, (2004) The Fractalization of Social Enterprise, "An ancient view of understanding" as well as "Another conception of understanding."