Consciousness of Abstraction

December 26, 2003

I'll begin by describing a number of levels or types of abstracting.  The first four of these more-or-less correspond to  Levels or perspectives on the use of language.  Because general semantic levels involve nervous systems, I will then describe the neurological process of abstraction.  And because language is our focus, I will then particularize the neural abstraction process to the context of words, so we can get to verbal levels. Once these are presented, I will then be able to illustrate the complexity and the additional requirements needed to illustrate abstracting between verbal levels.

Lexical abstraction.

1. A large set of sentences.
2. Selecting some of these sentences to represent the set or a subset.

Logical abstraction

1. A large set of sentences of propositional form.
2. A set of valid rules of inference
3. Constructing more sentences using (1) and (2)
4. Selecting some of the resulting sentences to represent the set or a subset.

Semantic abstraction.

1. A large set of words and sentences
2. A collection of objects
3. A mapping function from objects and sets of objects to words
4. A set of rules for determining "well formed sentences".
5. A set of valid rules of inferences applied to well formed sentences.
6. Constructing more sentences using (4) and (5).
7. A process for determining if any sets of objects "satisfy" a well formed sentence.
8. Selecting some of the resulting sentences to represent the set or a subset of well formed sentences.

Note: Lexical, logical, and even semantic abstraction can be performed by (properly programmed) computers.  In 1963 a theorem proving program rediscovered a more simple and elegant proof of the Euclidean geometry theorem that an isosceles triangle (a triangle with two equal sides) has two equal angles. Research showed that the relatively unknown proof had originally been discovered around 300 AD by Pappus, but it had apparently been largely forgotten, especially in educational contexts.

General semantic abstraction

1. A large set of words and sentences
2. A collection of objects
3. A mapping function from objects and sets of objects to words
4. A set of rules for determining sentences.
5. A set of rules of inferences applied to well formed sentences.
6. Constructing more sentences using (4) and (5).
7. A process for determining if any sets of objects "satisfies" a sentence.
8. A set of nervous systems to observe and respond to (1) through (8) and other nervous systems.
9. Selecting some words to represent the set or a subset of words, objects, rules, process, nervous systems, and responses of nervous systems.  (Note the absence of both the "valid" and "well formed" criteria at this level.)

Neural abstraction underlying general semantics abstraction.

Korzybski's highly abstract model describes this using only three levels - event, object, and verbal.

Our current model of the process is significantly more complex.

1. An energy source in the environment.
2. A medium to conduct energy (with losses [=abstraction])
3. A sensory interface that can detect (some of) (some kinds of) energy. (electromagnetic, vibratory, chemical, pressure, etc.)
4. (Multiple) sense cells that undergo an electro-chemical cascade reaction in response to various of the energy presences (with minimum threshold levels of energy for activation -each responding to different kinds of energy)
5. Synapses that allow one nerve cell to trigger (or inhibit) another.
6. Neurons that respond to sensory cells and trigger (or inhibit) other cells (nerve, glands, muscles)
7. Collections of neurons that selectively respond to different combinations of sensory cell responses.
8. "Organs" within the brain that communicate via nervous process with each other, with sensory input paths, with motor control functions, with glandular functions, etc.
9. Specialized muscle groups that move limbs for releasing pressure energy ( making marks on paper, tapping keyboards, pushing people, etc.) and vibratory energy (grunts, growls, speech, etc.), as well as organs for releasing chemical energy, (pheromones, hormones, etc.), both internally and externally to the person.

In this process each of us has learned to "identify" certain marks and sounds as "letters" and "words".

Neural abstraction to verbal levels process.

Assume light (electromagnetic energy) reflected from certain "marks" are the input to the above process. It then expands as follows:

1. Light reflected from marks strikes a person's retina.  Light is transduced into nerve firings
2. The retina abstracts from the billions of cells responding into millions of nervous processes, sending a continuous stream of nerve impulses into the optic centers of the brain. Nerve firings are transduced to neurotransmitters and neurotransmitters induce some and inhibit other nerve firings in a multiply repeated sequence of processes.
3. The structure of the connections process various combinations of retinal patters into distinct areas of the brain. 
4. Connections to other parts of the brain stimulate different regions to interact with the incoming patterns. Patterns are "recognized" and associations are brought forth from memory. 
5. Other parts of the brain are stimulated to interact with the resulting patterns. Various centers respond to the patterns, responses, and associations - centers that have been identified as "emotional", "cognitive", "verbal", "spatial", etc. (and many others whose functions we have no understanding of).
6. Here we have a significant gap in our model, because no allowable instrumentation has yet been devised to observe in detail any particular part of a brain functioning when an individual is presented with a visual stimulus that we might call a particular word.  We presume that there are, perhaps highly individualized to each individual, combinations of brain structures, neural firing patterns, synapse distributions, and specific chemical changes that we could, in principle, sometimes be correlate to particular visual or auditory patterns that we would call a "word". These presumed structural-functional processes could be labeled that person's "semantic reaction".
7. A small portion of any person's "semantic reactions" could, theoretically, with the advent of appropriate non-invasive instrumentation (Star-Trek like) be independently correlated to specific marking and auditory patterns (which we might call specific words). These "identifications" are learned, and they are necessary for time-binding.
8. Hypothesized highly individualized semantic reactions are abstracted from ongoing process and stimulate other parts of the brain to interact with various organs and motor control areas.  Of the many sensory experiences in combination with evoked associated memories (including prior responses), and subsequent neural processing (including devising new responses), a limited abstraction is selected to activate motor neurons to effect physical gestures and or the emission of audible vibrations. 
9. Some of these motor actions coordinated with physical objects produce additional marks.

And the above process repeats over and over throughout the life of the individual

10. Assume that, in principle, the correlation presumed by (7) above has been done for the population of a specific language and culture for all the individual words accepted by the members of that culture, and that each time a person learns a new word, his or her brain is mapped by the hypothesized process to his semantic reaction to that word.  Moreover, assume that this culture has in place a process to keep the individual mappings between individual semantic reactions to words up-to-date.

With this information at hand, we are in a position to define "verbal level" in relation to neurological level.  Whenever a person is experiencing the particular semantic reaction associated with a word, we can choose to focus our attention on the semantic reaction (neurological level) or the associated word (verbal level).  With this connection established, we can now refer to the "marks on paper" as "words", because we have a process of abstraction going from the energy emitted by the marks through the sense organs, into the brains, to the semantic reactions of persons, which because of our constantly up-to-date map, we can then say what the words are.  Note that the 9 step process from seeing words to writing words has a branch, like an antenna sticking up on a little remote control toy, at the 7th level of abstraction that performs an identification to bridge the gap from neurological to "verbal" levels.

Of course, using Korzybski's three level model is much simpler, but it is disastrously so, because it completely leaves out the fact that any verbal semantic reactions we have are not simple abstractions from the object level. Our semantic reactions include our prior experiences - our memories.

Now consider what is involved in abstracting from one "verbal" level to another.  For lexical, logical, or even semantic abstraction the process is very simple, as a review of the above processes shows, however a "general semantic" abstraction from one "verbal" level to another involves much more.

The starting point is the marks on paper that have been associated with and identified by the cultural mapping of words to semantic reactions.

1. First, the person involved must go through the first 7 steps in the process to develop his or her semantic reactions to the set of words that we will call the baseline verbal level.
8. Second, the person must undergo "subsequent neural processing (including devising new responses)"
9. Third, the subsequent neural processing resulting from combining the responses to individual words (mapped as semantic reactions) stimulate areas of the brain, some new, some re-stimulated, generating additional semantic reactions.
10. Next, from these composite semantic reactions, the person must select (abstract) which of his or her many ongoing neural processes (aka semantic reactions) that have been identified with words need to be emitted.
11. These selected reactions in turn stimulate the motor responses that result in the writing of new marks (words).

Evaluating whether the words emitted at step 11 "constitute" a "higher level of abstraction", in any conventional sense of the term, requires that we have additional structure not yet described.  We need, in addition to words, sentences, rules of inference, and rules of composition, objects, and reference rules, a dictionary, and general knowledge.

A dictionary is a set of sentences that identify a word with a set of other words, usually in the form of either a synonym or other sentences, but may also include "negative" defining by using antonyms and explanations of what the word does not "mean". A dictionary specifies a "definition" of a word in terms of other words, or processes using the names of objects.  Sometimes a dictionary also includes non-verbal illustrations of objects for which the words stand in semantic relation to.

General knowledge includes, for our purposes, relations between words and objects, ways of describing in words relations among objects, relations among words, and relations among words and objects, as well as ways of responding to objects or words, and the ability to specify, in words, appropriate actions, (and other things too).

Both dictionaries and knowledge are highly individualized for each person. Moreover the absence of the requirement for "valid reasoning" or "well formed sentences" together with the interaction in step 5 of the neural levels to verbal levels process with emotional and other portions of the brain can result in semantic reactions not logically consistent or even linguistically well formed.

Evaluating whether the words emitted at step 11 constitute a "higher level of abstraction" (in the conventional sense) requires that the person performing the evaluation have at his disposal the dictionary for all the words, and the knowledge associated with the words and sentences forming the basis level. In doing so, process step (9) may require a significant amount of processing. 

Because of the highly individualized nature of dictionaries and knowledge, as well as the similarly individualized response patters, vis-a-vis cognition, emotion, etc., there is no way to guarantee that any two individuals will arrive at the same evaluations regarding whether or not, between two sets of statements, one is at a higher level of abstraction than the other.

A high degree of consistence in the acculturation process regarding definitions of words, and the associated knowledge is required for agreement to be possible, and this is where time-binding comes in.  The greater the "enforcement" of standard dictionary definitions and the acquisition and sharing of knowledge, including valid reasoning, well formed sentences, and the appropriate proportion of emotional, cognitive, and other responses, the more effective time-binding is.

Evaluating the relative level of abstraction between sets of verbal statements with any degree of reliability requires consistency of use of terms, agreement on definitions, the use of valid reasoning, well formed sentences, and the primacy of conditioning of brain function to give primacy of valid reasoning over emotional responding. There is nothing in the neurological to verbal levels that provides any mean for processing between verbal levels of abstraction.  The evaluating of the relative levels of abstraction between different verbal statements depends entirely on the culturally shared definitions of the words involved, the knowledge associated with these words, and the emotional, and other brain reactions, to the words in their context and the person's situation, as well as the propensity of the person to use valid or invalid reasoning methods.  It is only through time-binding that agreement is achieved regarding any particular set of statements as to their relative levels of abstraction.

With that in mind, we have some shared knowledge of certain definitions that are fairly consistently evaluated regarding each other.  These comprise the following (not necessarily complete) general levels.:

1. Observation statements.
2. Description statements
3. Theories to explain observations
4. Inferences as to facts not explicitly observed (usually based on 1, 2, and 3)
   (including making assumptions and "jumping to conclusions" - not different in kind, just in validity of reasoning.)
5. Speculations as to causes - (usually based on the above, but sometimes without valid logic)
6. Attribution of motive - (similar to cause, but where people are involved)
7. Judgments of value - (good, bad, desirable, undesirable, "worthless", etc.)
8. Emotional reactions - (love, hate, like, anger, depression, etc.
9. Etc., - recycling (4) through (8) above repetitively - with and without more of (1) through (3)

To go from each of the above levels to the next involves the 11 levels described earlier.  To go from observation to description involves the eleven steps previously described.  To go from descriptions to theories involves 10 more steps. (If you think the math is funny, you forgot to take into consideration that the end of the first process is the start of the next.  The step 1 of the next abstraction begins with the step 11 of the previous process.) That means that the full complexity of going from observation to emotional reactions expressed in words involves 81 levels of abstracting, and there are ten (10) more levels for each additional higher level of abstracting.

Consciousness of abstracting traditionally involves keeping in mind, from moment to moment, all the steps that involve abstracting or "leaving out" information, when we are making assumptions, being aware of our inferences, recognizing our judgments, and realizing when we are experiencing emotional reactions.  We are expected to do this so that we can insure the appropriateness of all these steps, and to be aware that they are based on incomplete and possibly incorrect information.

1. We always miss possible information, because we cannot see or know everything.
2. Our verbal descriptions may be inadequate or possibly incorrect, because we cannot say "all" about anything.
3. Our theories may be inadequately tested, and they just might fail in this instance. (Time_n+1 is not time_n.)
4. Our inferences or assumptions could be wrong.  (Assumptions are not "bad"; we just have to be prepared.)
5. Our guesses about cause could be wrong. (We might have used the wrong theory.)
6. Our guesses about other people's motives may be wrong. (We can't know the "whole" story.)
7. Our judgment about the worth of something may be out-of-step with our culture. (One man's poison . . .)
8. Our emotional reactions may be counter-productive.
9. Our reactions to our reactions might take us "way out in left field".

But let not all these possibilities paralyze us. Go forth and act, but be prepared.