00100	CHAPTER 2--EXPLANATIONS AND MODELS
00200	
00300	
00400	  It is perhaps as difficult  to explain    scientific explanation as it
00500	is to explain anything else. The explanatory practices of different
00600	sciences differ widely but they all share the purpose of someone 
00700	attempting to answer someone else's why-how-what-etc. questions about
00800	a situation, event, episode, object or phenomenon. Thus explanation implies a 
00900	dialogue whose participants share some interests, beliefs, and values.
01000	A consensus must exist about admissable and appropriate questions and answers. The participants
01100	must agree on what is a sound and reasonable question and what is a
01200	relevant, intelligible, and (believed) correct answer.
01300	The explainer tries to satisfy a questioner's curiosity by making
01400	comprehensible why something is the way it is. The answer may be a
01500	definition, an example, a synonym, a story, a theory, a model-description, etc.
01600	The answer satisfies curiosity by settling belief. Nnaturally the task of
01700	satifying the curiosity of a five year old boy is different from that
01800	of satisfying a  forty year old psychiatrist.
01900	    Suppose a man dies and a questioner (Q) asks an expainer (E):
02000	       Q: Why did the man die?
02100	One answer might be:
02200	       E: Because he took cyanide.
02300	This explanation might be sufficient to satisfy Q's curiosity and he
02400	stops asking further questions. Or he might continue:
02500	       Q: Why did the cyanide kill him?
02600	and E replies:
02700	      E: Anyone who ingests cyanide dies.
02800	This  explanation appeals to a universal generalization under which
02900	is subsumed the particular fact of this man's death. Subsumptive explanations
03000	satisfy some questioners but not others who, for example, might want to
03100	know about the physiological mechanisms involved.
03200	       Q: How does cyanide work in killing people?
03300	       E: It stops respiration so one dies  from lack of oxygen.
03400	If Q has biochemical interests he might inquire further:
03500	       Q: What is cyanide's mechanism of drug action on the respiratory center?
03600	And so on, since there is no bottom to the questions which might be asked.
03700	Nor is there a top:
03800	       Q: Why did the man take cyanide?
03900	       E: Because he was depressed.
04000	       Q: What was he depressed about?
04100	       E: He lost his job.
04200	       Q: How did that happen?
04300	       E: The aircraft company let go most of their engineers because of the cut-back in defense contracts.
04400	Explanations are always incomplete because the top and bottom can be indefinitely
04500	extended and endless questions can be asked at each level.
04600	Just as the participants in explanatory dialogues
04700	decide what is taken to be problematic, so they also determine the termini of
04800	questions and answers. Each discipline has its characteristic stopping points.
04900	    In explanatory dialogues there exist larger and smaller constellations
05000	to refer to which are taken for granted as a nonproblematic background.
05100	Hence in considering  the function of paranoid thought `it goes without saying',
05200	that is, it transcends this particular field of function to say
05300	that a living teleonomic system as the larger constellation strives for
05400	maintenance and expansion of its life using smaller oriented, informed
05500	and constructive subprocesses. Also it goes without saying that at a lower
05600	level ion transport takes place through nerve-cell membranes. Every function
05700	of an organism can be viewed a governing a subfunction beneath and 
05800	depending on a transfunction above which calls it into play for a purpose.
05900	   Just as there are many alternative ways of describing, there are many
06000	alternative ways of explaining. An explanation is geared to some level
06100	of what the dialogue participants take to be the fundamental structures
06200	and processes under consideration. Since in psychiatry we cope with
06300	patients' problems using mainly symbolic-conceptual techniques,(it is true
06400	that the pill, the knife, and electricity are also still available.),
06500	we are interested in aspects of human conduct which can be
06600	explained, understood, and modified at a symbol-processing level. Hence I shall
06700	attempt to explain paranoid conversational interactions by describing 
06800	in some detail a simulation of paranoid interview behavior , having in
06900	mind an audience of mental health professionals and colleagues   in  fields
07000	of psychiatry, psychology, artificial intelligence, linguistics and philosophy.
07100	   Symbol processing explanations postulate an underlying intentionalistic
07200	structure of hypothetical mechanisms, functions or strategies, goal-directed symbol-processing
07300	procedures, having the power to produce and being responsible for
07400	the manifest phenomena. In this ethogenic (generating behavior, Harre[ ]) approach the term "mechanism"
07500	is not used in the  classical mechanical sense of the effects of forces on particles obeying laws of
07600	motion. Nor is it used in the sense of a mechanical contrivance such as a clock or an auto.
07700	Instead it is used here , and throughout the monograph,in the more general sense of modus operandi as
07800	in the mechanism for electing a president or the mechanism of evolutionary
07900	change. Thus I shall avoid the terms "mechanical" and "mechanistic" in order
08000	to avoid metaphors and images of Newtonian physics and contrivances. As will become clear,
08100	this ethogenic viewpoint uses the terms "mechanisms", "functions", "procedures"
08200	and "strategies" as roughly synonoymous.
08300	
08400	2.2 Symbolic models
08500		An algorithm composed of symbolic computational
08600	procedures converts input symbolic structures into output symbolic
08700	structures according to certain principles. The modus operandi
08800	of a symbolic model is simply the workings of an algorithm when run on
08900	a computer. At this level of explanation, to answer `why?' means to provide             
09000	an algorithm which makes explicit how symbolic structures go together,
09100	how they are organized to work to generate patterns of manifest phenomena.
09200	
09300	   To simulate the input-output behavior of a system using symbolic
09400	computational procedures, we construct a model which produces I/O
09500	behavior resembling that of the subject system being simulated. The
09600	resemblance is achieved through the workings of an inner postulated
09700	structure in the form of an algorithm, an organization of intentionalistic
09800	symbol processing procedures which are responsible for the characteristic
09900	observable behavior at the input-output level. Since we do not know the
10000	structure of the `real' simulative mechanisms used by the mind-brain,
10100	our postulated structure stands as an imagined  theoretical analogue,
10200	a possible and plausible organization of mechanisms analogous to the
10300	unknown mechanisms and serving as an attempt to explain the workings
10400	of the system under study. A simulation model is thus deeper than a
10500	pure black-box explanation because it postulates functionally equivalent
10600	mechanisms inside the box to account for observable patterns of I/O
10700	behavior. A simulation model constitutes an interpretive explanation
10800	in that it makes intelligible the connections between external input
10900	internal states and output by postulating intervening symbol-processing procedures operating
11000	between symbolic input and symbolic output. An intelligible description
11100	of the model should make clear why and how it reacts as it does under
11200	various circumstances.
11300	    To cite a universal generalization to explain an individuals behavior
11400	is unsatisfactory to a questioner who is interested in what powers and
11500	liabilities are latent behind manifest phenomena. To say `x is nasty
11600	because x is paranoid and all paranoids are nasty' may be relevant,
11700	intelligible and correct but it does not cite a structure which can account
11800	for `nasty' behavior as a consequence of input and internal states of
11900	a system. A model explanation specifies particular antecedants and mechanisms
12000	through which antecedants generate the phenomena. This ethogenic approach to
12100	explanation assumes perceptible phenomena display the regularities and
12200	irregularities they do because of the nature of a (currently) imperceptible
12300	and inaccessible underlying structure.
12400	   When attempts are made to explain human behavior, principles in
12500	addition to those accounting for the natural order are invoked. `Nature
12600	entertains no opinions about us' said Nietsche but human natures do and
12700	therein lies a  source of complexity for the understanding of human nature.
12800	Until the first quarter of the 20th century, natural sciences  have been guided by the Newtonian ideal
12900	of perfect process knowledge about inanimate objects whose behavior can
13000	be subsumed under lawlike generalizations. When a deviation from a law was
13100	noticed,it was the law which was modified, since by definition physical objects do not have the power to break laws.
13200	When the planet Mercury was observed to deviate from the orbit predicted
13300	by Newtonian theory, no one accused the planet of being an intentional agent
13400	breaking the law; something was wrong with the theory.  Subsumptive explanation is the acceptable norm in physics
13500	but it is seldom satisfactory in accounting for the behavior
13600	of living intentionalistic systems. In considering the behavior of falling bodies
13700	no one nowadays follows the Aristotelian pattern of attributing an intention
13800	to fall to the object in question. But in the case of living systems, especially
13900	ourselves, our ideal explanatory practice remains Aristotelian in utilizing
14000	a concept of intention.(Aristotle was not wrong about everything).
14100	   Consider a man participating in a high-diving contest. In falling towards
14200	the water he falls at the rate of 32 feet per second per second. Viewing
14300	the man simply as a falling body, we explain his rate of fall by appealing to a physical
14400	law. Viewing the man as a human intentionalistic agent, we explain his dive as the result
14500	of an intention to dive in a cetain way in order to win the diving contest.
14600	His action (in contrast to mere movement) involves an intended following
14700	of certain conventional rules for what is judged by humans to constitute, say,
14800	a swan dive. Suppose part way down he chooses to change his position in
14900	mid-air and enter the water thumbing his nose at the judges. He cannot break
15000	the law of falling bodies but he can break the rules of diving and make a 
15100	gesture which expresses disrespect and which he believes will be interpreted
15200	as such by the onlookers. Our diver breaks a rule for diving but follows
15300	another rule which prescribes gestural action for insulting behavior.
15400	To explain the actions of diving and nose-thumbing, we
15500	would appeal, not to laws of natural order, but to an additional order, to
15600	principles of human order, superimposed on laws of natural order and which
15700	take into account (1)standards of appropriate action in certain situations
15800	and (2) the agents inner considerations of intention, belief and value 
15900	which he finds compelling from his point of view.
16000	   In this type of explanation the explanandum, that which is being explained
16100	is the agent's informed actions, not simply his movements. When a human
16200	agent performs an action in a situation, we can ask:(1) is the action
16300	appropriate to that situation and if not, why did the agent believe his
16400	action to be called for.
16500	   As will be shown, symbol-processing explanations rely on concepts 
16600	of action, intention, belief, affect, preference, etc. These terms are
16700	close to the terms of ordinary language as is characteristic of  early
16800	stages of explanations. It is also important to note that such terms are commonly utilized 
16900	in describing computer algorithms in which final causes guide efficient causes. In
17000	an algorithm these ordinary terms can be explicitly defined and
17100	represented.
17200	   Psychiatry deals with the practical concerns of inappropriate action,
17300	belief, etc. on the part of a patient. His behavior may be inappropriate
17400	to the onlooker since it represents a lapse from the expected, a
17500	contravention of the human order. It may even appear this way to the 
17600	patient in monitoring and directing himself.But sometimes, as in severe cases of the paranoid mode
17700	the patient's behavior does not appear anomalous to himself. He maintains
17800	that anyone who understands his point of view, who conceptualizes
17900	situations as he does from the inside, would consider his outer behavior
18000	appropriate and justified. What he does not understand or accept is
18100	that his inner conceptualization is mistaken and represents a misinterpretation
18200	of the events of his experience.
18300	    The model to be presented in the sequel constitutes an attempt to
18400	explain some regularities and particular occurrences of conversational
18500	paranoid phenomena observable in the clinical situation of a psychiatric
18600	interview. The explanation is at the symbol-processing level of
18700	linguistically communicating agents and is cast in the form of a dialogue
18800	algorithm. Like all explanations it is only partially accurate, incomplete
18900	and does not claim to represent the only conceivable structure of mechanisms.
19000	
19100	2.3 The nature of algorithms
19200	
19300	   Theories can be presented in various forms such as natural language
19400	assertions, mathematical equations and computer programs. To date most
19500	theoretical explanations in psychiatry and psychology have consisted
19600	of natural language essays with all their well-known vagueness and
19700	ambiguities.Many of these formulations have been untestable, not because
19800	relevant observations were lacking but because it was unclear what
19900	the essay was really saying. Clarity is needed.
20000	     An alternative way of formulating psychological theories is now
20100	available in the form of ethogenic algorithms, computer programs, which have
20200	the virtue of being clear and explicit in their articulation and which
20300	can be run on a computer to test internal consistency and external correspondence with the data of observation.
20400	Since we do not know the `real' mind-brain algorithms,
20500	we construct a theoretical model which represents a partial
20600	paramorphic analogue. (See Harre, 1972). The analogy is at the symbol-
20700	processing level, not at the hardware level. A functional, computational
20800	or procedural equivalence is being postulated. The question then becomes
20900	one of determining the degree of the equivalence. Weak functional equivalence
21000	consists of indistinguishability at the outermost input-output level.
21100	Strong equivalence means correspondence at each inner I/O level, that is
21200	there exists a match not only between what is being done but how it is
21300	being done at a given level of operations.(These points will be discussed
21400	in greater detail in Chapter 3).
21500	   An algorithm represents an organization of symbol-processing mechanisms or functions
21600	which represent an `effective procedure'. It is essential here to grasp this concept.
21700	An effective procedure consists of two ingredients:
21800	       (1) A programming language in which procedural rules of behavior
21900	          can be rigorously and unambiguously specified.
22000	     (2) A machine processor which can rapidly and reliably carry out
22100	          the processes specified by the procedural rules.
22200	The specifications of (1), written in a formally defined programming
22300	language, is termed an algorithm or program while (2) involves a computer
22400	as the machine processor, a set of deterministic physical mechanisms
22500	which can perform the operations specified in the algorithm. The
22600	algorithm is called `effective' because it actually works, performing
22700	as intended when run on the machine processor.
22800	     It is worth remphasizing that a simulation model postulates
22900	procedures analogous to the real and unknown procedures. The analogy being 
23000	drawn here is between specified processes and their generating systems.
23100	Thus
23200	
23300	      mental process    computational process
23400	      --------------:: ----------------------
23500	    brain hardware      computer hardware and
23600	    and programs           programs
23700	The analogy is not simply between computer hardware and brain wetware.
23800	We are not comparing the structure of neurons with the structure of
23900	transisitors; we are comparing the organization of symbol-processing
24000	procedures in an algorithm with symbol-processing procedures of the
24100	mind-brain. The central nervous system contains a representation of
24200	the experience of its holder. A model builder has a conceptual representation
24300	of that representation which he demonstrates in the form of an algorithm.
24400	Thus an algorithm is a demonstration of a  representation of a representation.
24500	    When an algorithm runs on a computer the postulated explanatory
24600	structure becomes actualized, not described. (To describe the model
24700	is to present , among other things, its embodied theory). A simulation model such as the
24800	one presented here can be interacted with by a person at the linguistic
24900	level as a communicating agent in the world. Its symbolic communicative behavior
25000	can be experienced in a concrete form by a human observer-actor.
25100	Thus it can be known by acquaintance, by first-hand knowledge, as well
25200	as by the second-hand knowledge of description.
25300	   Since the algoritm is written in a programming language, it is hermetic
25400	and opaque except to a few people, who in general do not enjoy reading
25500	other people's code. Hence the intelligibility requirement for explanations
25600	must be met in other ways. In an attempt to open the model to scrutiny
25700	I shall describe the model in detail using diagrams and interview
25800	examples profusely.
26000	
26100	2.4 ANALOGY
26200	    I have stated that a simulation model of a symbolic system reproduces
26300	the behavior of that system at some input-output level. The reproduction
26400	is achieved through the operations of an algorithm which represents
26500	an organization of hypothetical symbol-processing strategies or procedures
26600	which have the ability to generate the I/O behavior of the processes
26700	under investigation.The algorithm must be an effective procedure, that is
26800	one which  really works in the manner intended by the model-builders. In the model
26900	herein described our paranoid algorithm generates linguistic I/O behavior
27000	typical of patients whose thought processes are dominated by the paranoid mode.
27100	Given that the manifest outermost I/O behavior of the model is
27200	indistinguishable from the manifest outward I/O behavior of paranoid
27300	patients, does this imply that the hypothetical underlying processes used
27400	by the model are analogous to or the same as the underlying processes
27500	used by persons in the paranoid mode. This deep and thorny question
27600	should be approached with caution and only when we are first armed with some clear notions
27700	about analogy, similarity, faithful reproduction, indistinguishability and functional  equivalence.
27800	    In comparing two things (objects, systems or processes ) one can cite properties they
27900	have in common, properties they do not share and properties  regarding which
28000	it is difficult to tell. No two things are exactly alike in every detail.
28100	If they were identical in respect to all their properties then they would be copies. If
28200	they were identical in every respect including their spatio-temporal
28300	location we would say we have only one thing instead of two. One can
28400	assert with some justification that a given thing  is not similar to
28500	anything else in the world or it is similar to evrything else in the world 
28600	depending upon how we cited properties.
28700	    Similarity relations are used in processes of classification in which
28800	objects are grouped into classes , the classes then representing object-
28900	concepts. The members of a class of object-concepts resemble one another
29000	in sharing certain properties. The resemblance between members of the class
29100	is not exact or total. Members of a class are considered more or less alike
29200	and there exist degrees of resemblance. A classification may involve only single
29300	properties while a taxonomy seeks to classify things according to their
29400	structure or organization. Thus a simulation model contributes to taxonomy
29500	in that since model X is structurally analogous to its subject Y, Y is to be
29600	viewed as belonging to the same class as X.
29700	   In an analogy a comparison is drawn between two things. `Newton did not
29800	show the cause of the apple falling but he showed a similitude bewteen the
29900	apple and the stars.'(D`Arcy Thompson). Huygens suggested an analogy between
30000	sound waves and light waves in order to understand something less well-understood
30100	(light)in terms of something better understood(sound).To account for species
30200	variation, Darwin postulated a mechanism of natural selection. He constructed
30300	an analogy from two sources, one from artificial selection as practiced
30400	by domestic breeders of animals and one from Malthus' theory of a competetion
30500	for existence in a population increasing geometrically while its resources
30600	increase arithmetically. Bohr's model of the atom offered an analogy between
30700	solar system and atom. These few well-known historical examples make vivid
30800	the role of analogies in theory construction. Such analogies are partial
30900	paramorphs (Harre,1971) in that two systems are compared for parallelisms
31000	and they are compared only in respect to certain properties which   
31110	constitute the positive and neutral analogy. The negative analogy is ignored.
31111	Bohr's model of the atom as a miniature planetary system was
31200	not intended to suggest that electrons possessed color or that planets
31300	jumped out of their orbits.
31310	2.4 FUNCTIONAL EQUIVALENCE
31400	   When human thought is the subject of a simulation model, we draw from
31500	two sources, symbolic computation and psychology, an analogy between
31600	systems known to be able to process symbols, persons and computers. The
31700	properties compared in the analogy  are obviously not physical or substantive
31800	such as blood and wires, but functional and procedural. We want to assume
31900	that the not well- understood procedures of thought in a person are
32000	similar to the somewhat better understood procedures of symbol-processing
32100	which take place in a computer. The analogy is one of functional           
32200	or procedural equivalence. If model and human are indistinguishable at a manifest
32300	I/O level, then they can be considered weakly equivalent. If they are
32400	indistinguishable at deeper and deeper I/O levels, then strong equivalence
32500	becomes  achieved. (See Fodor,1968). How stringent and how deep are the
32600	demands for equivalence to be? Must there be point-to-point correspondences
32700	at every level? What is to count as a point and what are the levels?
32800	Procedures can be specified and ostensively pointed to in an algorithm
32900	but how can we point to the inaccessible symbolic processes in a person's head?
33000	Does a demonstration of functional equivalence constitute an explanation of observable
33100	behavior?
33200	   In constructing an algorithm one puts together an organization
33300	of collaborating functions. (As mentioned, we use the terms `function',   
33400	`procedure' and `strategy' interchangeably.) A function takes some symbolic
33500	structure as input and yields some other symbolic structure as output.
33600	Two computationally equivalent functions, having the same input and yielding
33700	the same output, can differ `inside' the function at the instruction level.
33800	   Consider an elementary programming problem which students in symbolic
33900	computation are commonly asked to solve. Given a list L of symbols,
34000	L=(A B C D), as input, construct a function or procedure which will
34100	convert this list to the list RL in which the order of the symbols is
34200	reversed, i.e. RL=(D C B A). Here are some examples of functions which
34300	will carry out the operation of reversal. (They are written in the high-level
34400	programming language MLISP).                                               
34500	        REVERSE1 (L);
34600	          BEGIN 
34700	            NEW RL;
34800	            RETURN FOR NEW I IN L DO
34900	              RL ← I CONS RL;
35000	          END;
35100	
35200	       REVERSE2 (L);
35300	         BEGIN
35400	           NEW RL, LEN;
35500	           LEN ← LENGTH (L);
35600	           FOR NEW N ← 1 TO LEN DO
35700	             RL[N] ← L [LEN - N + 1];
35800	           RETURN RL;
35900	         END;
36000	       REVERSE3 (L);
36100	         REVERSE3A (L,NIL);
36200	
36300	       REVERSE3A (L,RL);
36400	         IF NULL L THEN RL
36500	         ELSE REVERSE3A (CDR L, CAR L CONS RL);
36600	   Each of these computational functions takes a list of symbols, L, as
36700	input and produces a new list, RL, in which the order of the symbols on the
36800	input list is reversed. It is at this I/O level that the functions can
36900	be said to be equivalent. Looking inside the functions one can see
37000	similarities as well as differences at the level of the individual
37100	instructions. For instance, REVERSE1 steps down the input list L, takes
37200	each symbol found and inserts it at the front of the new list RL. On the
37300	other hand, REVERSE2 counts the length of the input list L using another
37400	function called LENGTH which determines the length of a list. REVERSE2
37500	then uses index expressions on both sides of an assignment operator, ← ,
37600	(a) to obtain a position in the list RL, (b) to obtain a symbol in the list 
37700	L and (c) to assign the symbol to that position in the reversed list RL.
37800	Notice that REVERSE1 and REVERSE2 are similar in that they use FOR loops
37900	while REVERSE3, which calls another function REVERSE3A, does not. REVERSE3A
38000	is different from all the others in that it contains an IF expression.
38100	   Hence similariries and differences can be cited between functions as
38200	long as we are clear about levels and degrees of detail. The above-described
38300	functions are computationally equivalent at the input-output level since
38400	they take the same symbolic structures as input and produce the same
38500	symbolic output.
38600	2.5 Functional equivalence
38700		  If we propose that an algorithm we have constructed is functionally
38800	equivalent to what goes on in humans when they process symbolic structures,
38900	how can we justify this position ? Indistinguishability tests at, say,
39000	the linguistic level provide evidence only for weak equivalence. We
39100	would like to be able to get inside the underlying processes in humans
39200	the way we can with an algorithm by inspecting its instructional code.
39300	The difficulty lies in identifying, making tangible and counting processes
39310	in human heads. Many experiments  must be designed and carried out.
39400	We must have great patience with the neural sciences and psychology.
39500	   In the meantime, besides weak equivalence and plausibility arguments, 
39600	one can appeal to extra-evidential support from other
39700	relevant domains. One can offer analogies between what is known to go on at  
39800	a molecular level in living organisms and what goes on in an algorithm.
39900	Foe example, a DNA molecule in the nucleus of a cell consists of an
40000	ordered sequence (list) of nucleotide bases (symbols) coded in triplets
40100	termed codons (words). Each element of the codon specifies which amino
40200	acid during protein synthesis is to be linked into the chain of polypeptides
40300	making up the protein. The codons function like instructions in a 
40400	programming language. One codon is known to operate as a terminal symbol
40500	analogous to symbols in an algorithm which terminate the end of a list.
40600	If a stop codon appears in the middle of a sequence rather than at its
40700	normal terminal position, as in a point mutation, further protein
40800	synthesis is prevented. The polypeptide chain resulting is abnormal
40900	and may have lethal or trivial consequences for the organism depending
41000	on what other collaborating processes require to be handed over to them. Similarly
41100	in a algorithm. To use our previous programming example, the list L
41200	consisting of the symbols (A B C D) actually contains the terminal
41300	symbol NIL which is left unwritten because it is taken as a convention.
41400	If in reversing the list (A B C D NIL) the symbol NIL appeared in the
41500	middle of the list,i.e. (A B NIL C D), then the reversed list RL would
41600	contain only (B A) instead of the expected (D C B A) because
41700	the terminal symbol had been encountered. Such a result may be lethal
41800	or trivial to the algorithm depending on what other functions require
41900	as input from the reversing function. Each function in a algorithm
42000	is embedded in an organization of collaborating functions just as
42100	is the case in living organisms.
42200	   We know that at the molecular level of living organisms there exist
42300	rules for processes such as serial progression along a nucleotide
42400	sequence which are analogous to stepping down a list in an algorithm.
42500	Further analogies can be made between point mutations in which DNA
42600	codons can be inserted, deleted, substituted or reordered and  symbolic
42700	computation in which the same operations are commonly carried out.
42800	Such analogies are interesting as extraevidential support but obviously
42900	closer linkages are  needed between the macro-level of thought processes
43000	and the micro-level of molecular information-processing .
43100	   To obtain evidence for the acceptability of the model empirical tests
43200	are utilized in evaluation procedures. Such tests should also tell us
43300	which is the best among alternative models. Once we have the `best available'
43400	model can we be sure it is correct or verisimilar?  We can never know with certainty. Theories
43500	and models have a short half-life as approximations and become superseded by better ones.