00100	 EXPLANATIONS AND MODELS
00200	 The Nature of Explanation
00300		It is perhaps as difficult to explain explanation  itself  as
00400	it is to explain anything else. (Nothing, except everything, explains
00500	anything). The explanatory practices  of  different  sciences  differ
00600	widely but they all share the purpose of someone attempting to answer
00700	someone  else's  (or  his own) why-how-what-etc.    questions about a
00800	situation, event, episode, object  or  phenomenon.  Thus  explanation
00900	implies  a dialogue whose participants share some interests, beliefs,
01000	and values.   A consensus must exist about what  are  admissible  and
01100	appropriate  questions and answers.    The participants must agree on
01200	what is a sound and reasonable  question  and  what  is  a  relevant,
01300	intelligible, and (believed) correct answer. The explainer  tries  to
01400	satisfy   a  questioner's  curiosity  by  making  comprehensible  why
01500	something  is  the  way  it  is.  Depending  on  what  mystifies  the
01600	questioner,the  answer  may be a definition, an example, a synonym, a
01700	story, a theory, a model-description, etc.  The  answer  attempts  to
01800	satisfy  curiosity  by  settling  belief, at least temporarily, since
01900	scientific  beliefs  are  corrigible  and  revisable.  A   scientific
02000	explanation  aims  at  convergence  of  belief in the relevant expert
02100	community.
02200		Suppose a man dies and a questioner (Q) asks an explainer (E): 
02300		Q: Why  did  the  man  die?  
02400	One answer might be:
02500		E: Because he swallowed  cyanide.
02600	This explanation might be sufficient to satisfy Q's curiosity and  he
02700	and he stops asking further questions. Or he might continue:
02800	        Q. Why did the cyanide kill him?
02900	and E replies:
03000	        E: Anyone who swallows  cyanide dies.
03100	This explanation appeals to a universal generalization under which is
03200	subsumed  the  particular  fact  of  this  man's  death.  Subsumptive
03300	explanations  satisfy  some  questioners  but  not  others  who,  for
03400	example,  might  want  to  know  about  the  physiological mechanisms
03500	involved.                                                            
03600	        Q: How does cyanide work in causing death?
03700	        E: It stops respiration so the person dies from lack of oxygen.
03800		If Q has biochemical interests he might inquire further: 
03900		Q:What is cyanide's mechanism of drug action on the
04000		respiratory center?
04100		The last two questions refer to causes. When human action  is
04200	to  be  explained,  confusion  easily  arises  between  appealing  to
04300	physical, mechanical causes and appealing to  symbolic-level  reasons
04400	which   constitute  learned,  acquired  strategies  seemingly  of  an
04500	ontological order different from causes. (See Toulmin, 1971).
04600		It is established clinical knowledge that the phenomena of
04700	the paranoid mode can be found associated with a variety of  physical
04800	disorders.    For example, paranoid thinking can be found in patients
04900	with  head   injuries,   hyperthyroidism,   hypothyroidism,   uremia,
05000	pernicious   anemia,   cerebral  arteriosclerosis,  congestive  heart
05100	failure, malaria and epilepsy.      Also drug  intoxications  due  to
05200	alcohol,  amphetamines,  marihuana  and LSD can be accompanied by the
05300	paranoid mode. In these cases the paranoid mode is not a  first-order
05400	disorder  but  a  way  of  processing information in reaction to some
05500	other underlying disorder. To account for the association of paranoid
05600	thought  with  these  physical  states  of  illness,  a psychological
05700	theorist might be tempted to hypothesize that a  purposive  cognitive
05800	system would attempt to explain ill health by attributing it to other
05900	malevolent human agents. But before making such an explanatory  move,
06000	we must consider the at-times elusive distinction between reasons and
06100	causes in explanations of human behavior.
06200		One  view  of  the  association  of  the  paranoid  mode with
06300	physical disorders might be that the physical illness  simply  causes
06400	the  paranoia,  through  some  unknown mechanism, at a physical level
06500	beyond the influence of deliberate self-direction and self-control.
06600	That  is,  the  resultant  paranoid  mode  represents  something that
06700	happens to a person as victim, not  something  that  he  does  as  an
06800	active  agent.   Mechanical causes thus provide one type of reason in
06900	explaining behavior. Another view is that the paranoid  mode  can  be
07000	explained  in terms of symbolically-represented reasons consisting of
07100	rules and patterns of rules which specify an agent's  intentions  and
07200	beliefs.   In  a given situation does a person as an agent recognize,
07300	monitor and control what he is doing or trying to do?    Or  does  it
07400	just happen to him automatically without conscious deliberation?
07500		This question raises a third view, namely  that  unrecognized
07600	reasons,  aspects of the symbolic representation which are sealed off
07700	from reflective deliberation, can function like mechanical causes  in
07800	that  they  are  inaccessible  to  voluntary  control. If they can be
07900	brought to consciousness, such  reasons  can  sometimes  be  modified
08000	voluntarily  by  the  agent,  who, using ordinary language as its own
08100	metalanguage, can reflexively talk to  and  instruct  himself.   This
08200	second-order  monitoring  and control through language contrasts with
08300	an agent's inability to modify mechanical causes or symbolic  reasons
08400	which    lie    beyond    the   influence   of   self-criticism   and
08500	self-emancipation  carried  out   through   linguistically   mediated
08600	argumentation.     Timeworn  conundrums  about concepts of free-will,
08700	determinism, responsibility, consciousness and the powers  of  mental
08800	action  here  plague  us  unless  we can take advantage of a computer
08900	analogy in which a clear and  useful  distinction  is  drawn  between
09000	levels  of mechanical hardware and symbolically-represented programs.
09100	This important distinction will be elaborated shortly.
09200	
09300		Each of these three views provides a serviceable  perspective
09400	depending on how a disorder is to be explained and corrected.    When
09500	paranoid processes occur during amphetamine intoxication, they can be
09600	viewed  as  biochemically  caused and beyond the patient's ability to
09700	control volitionally through internal self-correcting dialogues  with
09800	himself.  When a paranoid moment occurs in a normal person, it can be
09900	viewed as involving a symbolic misinterpretation.   If  the  paranoid
10000	misinterpretation  is  recognized as unjustified, a normal person has
10100	the emancipatory power  to  revise  or  reject  it  through  internal
10200	debate.  Between  these  extremes of drug-induced paranoid states and
10300	the self-correctable paranoid moments of the normal person, lie cases
10400	of  paranoid  personalities, paranoid reactions and the paranoid mode
10500	associated   with   the   major    psychoses    (schizophrenic    and
10600	manic-depressive).
10700		One opinion has it that the major psychoses are a consequence
10800	of unknown  physical  causes  and  are  beyond  deliberate  voluntary
10900	control.   But  what  are we to conclude about paranoid personalities
11000	and paranoid reactions where no hardware disorder  is  detectable  or
11100	suspected?  Are  such  persons  to  be  considered  patients  to whom
11200	something is mechanically happening at the physical level or are they
11300	agents whose behavior is  a  consequence  of  what  they  do  at  the
11400	symbolic level?   Or are they both agent and patient depending on how
11500	one views the self-modifiability of their  symbolic  processing?   In
11600	these  perplexing  cases  we  shall take the position that in normal,
11700	neurotic and characterological  paranoid  modes,  the  psychopathlogy
11800	represents  something  that happens to a man as a consequence of what
11900	he has experientially undergone, of something he now does, and
12000	something he now undergoes.    Thus he is both agent and victim whose
12100	symbolic processes have powers to do and liabilities to undergo.  His
12200	liabilities  are  reflexive  in that he is victim to, and can succumb
12300	to, his own symbolic structures.
12400	
12500		From this standpoint I  would  postulate  a  duality  at  the
12600	symbolic  level  between  reasons  and causes. That is, a consciously
12700	unrecognized reason can operate like  a  mechanical  cause  in  being
12800	inaccessible  to voluntary modification by symbolic reprogramming. It
12900	is, of course, not reasons themselves which operate as causes but the
13000	execution  of  the  reason-rules  which  serves  as  a determinant of
13100	behavior.  Human symbolic behavior is non-determinate to  the  extent
13200	that  it  is  autonomously self-determinate. Thus the power to select
13300	among alternatives, to make some decisions freely and to change one's
13400	mind  is  non-illusory.  When a reason is recognized to function as a
13500	cause  and  is  accessible  to  self-monitoring  (the  monitoring  of
13600	monitoring),  emancipation  from  it  can  occur  through  change  or
13700	rejection of belief. In this sense an at least two-levelled system is
13800	self-changeable and self-emancipatory, within limits.
13900		Explanations  both  in  terms  of  causes  and reasons can be
14000	indefinitely extended and endless questions  can  be  asked  at  each
14100	level of analysis.  Just as the participants in explanatory dialogues
14200	decide what is taken to be problematic, so they  also  determine  the
14300	termini  for a series of questions and answers.   Each discipline has
14400	its characteristic stopping points and boundaries.
14500		Underlying such explanatory dialogues are larger and  smaller
14600	constellations   of   concepts   which   are  taken  for  granted  as
14700	nonproblematic background.    Hence in considering the strategies  of
14800	the paranoid mode "it goes without saying" that any living teleonomic
14900	system ,as the larger constellation ,  strives  for  maintenance  and
15000	expansion  of life. Also it should go without saying that, at a lower
15100	level, ion transport takes place through nerve-cell membranes.  Every
15200	function  of  an  organism  can  be viewed as governing a subfunction
15300	beneath and depending on a transfunction above which  calls  it  into
15400	play for a purpose.
15500		Just as there are many alternative ways of describing, there
15600	are many alternative ways of explaining.  An explanation is geared to
15700	some  level  of  what  the  dialogue  participants  take  to  be  the
15800	fundamental  structures  and processes under consideration.  Since in
15900	psychiatry   we   cope   with   patients'   problems   using   mainly
16000	symbolic-conceptual  techniques,(it is true that the pill, the knife,
16100	and electricity are also available), we are interested in aspects  of
16200	human  conduct  which can be explained, understood, and modified at a
16300	symbol-processing  level.  Psychiatrists  need  theoretical  symbolic
16400	systems from which their clinical experience can be logically derived
16500	to interpret the case histories of their patients. Otherwise they are
16600	faced  with  mountains  of  indigestible  data  and  dross.  To quote
16700	Einstein: "Science is an attempt to make the chaotic diversity of our
16800	sense  experience correspond to a logically uniform system of thought
16900	by correlating single experiences with the theoretic structure."
17000	
17100	 The Symbol Processing Viewpoint
17200	
17300		Segments and sequences of human behavior can be studied  from
17400	many  perspectives.   In  this  monograph, I shall view sequences of
17500	paranoid symbolic behavior from an information-processing  standpoint
17600	in  which  persons  are  viewed  as symbol users. For a more complete
17700	explication and justification of this perspective , see Newell (1973)
17800	and Newell and Simon (1972).
17900		In brief, from this vantage point we  define  information  as
18000	knowledge   in   a  symbolic  code.  Symbols  are  considered  to  be
18100	representations  of  experience  classified   as   objects,   events,
18200	situations and relations. A symbolic process is a symbol-manipulating
18300	activity posited to account for observable symbolic behavior such  as
18400	linguistic  interaction. Under the term "symbol-processing" I include
18500	the seeking, manipulating and generating of symbols.
18600		Symbol-processing   explanations   postulate   an  underlying
18700	structure  of  hypothetical  processes,  functions,  strategies,   or
18800	directed  symbol-manipulating procedures, having the power to produce
18900	and being responsible for observable patterns of  phenomena.  Such  a
19000	structure  offers an ethogenic (ethos = conduct or character, genic =
19100	generating)  explanation  for  sequences  or  segments  of   symbolic
19200	behavior.  (See Harre and Secord,1972).  From an ethogenic viewpoint,
19300	we can posit processes, functions, procedures and strategies as being
19400	responsible  for  and  having  the  power  to  generate  the symbolic
19500	patterns  and  sequences  characteristic  of   the   paranoid   mode.
19600	"Strategies"  is  perhaps the best general term since it implies ways
19700	of obtaining an objective - ways which have suppleness and pliability
19800	since choice of application depends on circumstances.        However,
19900	I shall use all these terms interchangeably.
20000	
20100	 Symbolic Models
20200		Theories and  models  share  many  functions  and  are  often
20300	considered  equivalent.   One  important distinction lies in the fact
20400	that a theory states a subject has a certain structure but  does  not
20500	exhibit  that  structure in itself. (See Kaplan,1964). In the case of
20600	computer simulation models there exists a further useful distinction.
20700	Computer  simulation  models  which  have  the ability to converse in
20800	natural language using teletypes, actualize or realize  a  theory  in
20900	the  form of a dialogue algorithm. In contrast to a verbal, pictorial
21000	or  mathematical  representation,  such  a  model,  as  a  result  of
21100	interaction,  changes  its  states  over  time and ends up in a state
21200	different from its initial state.
21300		Einstein once remarked, in contrasting the act of description
21400	with  what  is  described,  that it is not the function of science to
21500	give the taste of the soup.  Today  this  view  would  be  considered
21600	unnecessarily  restrictive.  For  example, a major test for synthetic
21700	insulin is whether it reproduces the effects, or at least some of the
21800	effects  (such as lowering blood sugar), shown by natural insulin. To
21900	test whether a simulation is successful, its effects must be compared
22000	with  the  effects produced by the naturally-occuring subject-process
22100	being modelled.     An interactive simulation model which attempts to
22200	reproduce  sequences of experienceable reality, offers an interviewer
22300	a first-hand experience with  a  concrete  case.  In  constructing  a
22400	computer  simulation, a theory is modelled to discover a sufficiently
22500	rich  structure  of  hypotheses  and  assumptions  to  generate   the
22600	observable  subject-behavior  under  study.      A dialogue algorithm
22700	allows an observer to interact with a concrete specimen of a class in
22800	detail. In the case of our model, the level of detail is the level of
22900	the symbolic behavior of  conversational  language.   This  level  is
23000	satisfying  to  a clinician since he can compare the model's behavior
23100	with its natural human counterparts using familiar skills of clinical
23200	dialogue. Communicating with the paranoid model by means of teletype,
23300	an interviewer can directly experience for himself a  sample  of  the
23400	type  of  impaired social relationship which develops with someone in
23500	a paranoid mode.
23600		An algorithm composed of  symbolic  computational  procedures
23700	converts  input  symbolic  structures into output symbolic structures
23800	according to certain principles.    The  modus  operandi  of  such  a
23900	symbolic  model  is simply the workings of an algorithm when run on a
24000	computer. At this level of explanation, to answer  a  "why"  question
24100	means  to  provide  an  algorithm  which  makes explicit how symbolic
24200	structures collaborate, interplay and interlock - in short, how  they
24300	are organized to generate patterns of manifest phenomena.
24400	
24500		To simulate the sequential input-output behavior of a  system
24600	using  symbolic  computational  procedures,  one writes  an algorithm
24700	which, when run on a computer, produces symbolic behavior  resembling
24800	that  of  the  subject  system  being  simulated.    (Colby,1973) The
24900	resemblance is achieved through the  workings  of  an  inner  posited
25000	structure   in   the   form  of  an  algorithm,  an  organization  of
25100	symbol-manipulating procedures which  are  ethogenically  responsible
25200	for the characteristic observable behavior at the input-output level.
25300	Since we do not know the structure of the "real" simulative processes
25400	used  by  the mind-brain, our posited structure stands as an imagined
25500	theoretical  analogue,  a  possible  and  plausible  organization  of
25600	processes  analogous  to  the  unknown  processes  and  serving as an
25700	attempt to explain the  workings  of  the  system  under  study.    A
25800	simulation  model  is  thus  deeper  than  a  structureless black-box
25900	explanation because it postulates functionally  equivalent  processes
26000	inside  the  box  to  account  for  outwardly  observable patterns of
26100	behavior.    A   simulation   model   constitutes   an   interpretive
26200	explanation  in  that  it  makes intelligible the connections between
26300	external input, internal states and output  by  positing  intervening
26400	symbol-processing  procedures  operating  between  symbolic input and
26500	symbolic output. To be  illuminating,  a  description  of  the  model
26600	should  make  clear  why  and  how it reacts as it does under various
26700	circumstances.
26800		Citing a universal generalization to explain an  individual's
26900	behavior  is unsatisfactory to a questioner who is interested in what
27000	powers and liabilities are latent behind manifest phenomena.  To  say
27100	"x is nasty because x is paranoid and all paranoids are nasty" may be
27200	relevant, intelligible and correct. But another type  of  explanation
27300	is  possible:  a model-explanation referring to a structure which can
27400	account for "nasty" behavior as a consequence of input  and  internal
27500	states  of  a  system.     A  model  explanation specifies particular
27600	antecedents and processes  through  which  antecedents  generate  the
27700	phenomena.   An ethogenic approach to explanation assumes perceptible
27800	phenomena display the regularities and nonrandom irregularities  they
27900	do  because  of  the  nature  of  an  imperceptible  and inaccessible
28000	underlying structure.    The  posited  theoretical  structure  is  an
28100	idealization,  an  imagined analogue to the unobservable structure in
28200	persons.
28300		When attempts are made to explain human behavior,  principles
28400	in  addition  to  those accounting for the natural order are invoked.
28500	"Nature entertains no opinions about us", said Nietzsche.  But  human
28600	natures  do,    and  therein  lies  a  source  of  complexity for the	
28700	understanding of human conduct. Until the first quarter of  the  20th
28800	century,  natural  sciences  were  guided  by  the Newtonian ideal of
28900	perfect process knowledge  about  inanimate  objects  whose  behavior
29000	could  be  subsumed  under lawlike generalizations.  When a deviation
29100	from a law was  noticed,  it  was  the  law  which  was  subsequently
29200	modified, since by definition physical objects did not have the power
29300	to break laws. When the planet Mercury was observed to  deviate  from
29400	the orbit predicted by Newtonian theory, no one accused the planet of
29500	being an intentional agent disobeying a law. Instead it was suspected
29600	that something was incorrect about the theory.
29700		Subsumptive explanation is the acceptable norm in many fields
29800	but  it is seldom satisfactory in accounting for particular sequences
29900	of behavior in living purposive systems.  When physical  bodies  fall
30000	in the  macroscopic world, few find it scientifically useful to posit
30100	that bodies have an intention to fall .  But in the  case  of  living
30200	systems,  especially  ourselves,  our  ideal  explanatory practice is
30300	teleonomically Aristotelian in utilizing a concept of intention. (For
30400	a thorough discussion of purpose and intentionality see Boden,1972).
30500		Consider  a  man participating in a high-diving contest.   In
30600	falling towards the water he accelerates at the rate of 32  feet  per
30700	second. Viewing the man simply as a falling body, we explain his rate
30800	of fall by appealing to a physical law.  Viewing the man as  a  human
30900	intentionalistic  agent,  we  explain  his  dive  as the result of an
31000	intention to dive in a  certain  way  in  order  to  win  the  diving
31100	contest.  His  conduct  (in  contrast  to  mere movement) involves an
31200	intended following of certain conventional rules for what  is  judged
31300	by  humans  to constitute, say, a swan dive. Suppose part-way down he
31400	chooses to change  his  position  in  mid-air  and  enter  the  water
31500	thumbing his nose at the judges. He cannot disobey the law of falling
31600	bodies but he can disobey or ignore the rules of diving. He can  also
31700	make  a gesture which expresses disrespect and which he believes will
31800	be interpreted as such by the onlookers.   Our diver  breaks  a  rule
31900	for  diving but follows another rule which prescribes gestural action
32000	for insulting behavior.    To  explain  the  actions  of  diving  and
32100	nose-thumbing,  we would appeal, not to laws of natural order, but to
32200	an additional order, to principles of  human  order.  This  order  is
32300	superimposed  on  laws  of  natural  order  and  takes  into  account
32400	(1)standards of appropriate action in certain situations and (2)  the
32500	agent's  inner considerations of intention, belief and value which he
32600	finds compelling from his point of view. In this type of  explanation
32700	the  explanandum,  that  which  is  being  explained,  is the agent's
32800	informed actions, not  simply  his  movements.  When  a  human  agent
32900	performs  an  action  in  a  situation,  we  can  ask:  is the action
33000	appropriate to that situation and if not, why did the  agent  believe
33100	his action to be called for?
33200		Symbol-processing explanations of human conduct rely on
33300	concepts of intention, belief, action, affect, etc. These  terms  are
33400	close to the terms of ordinary language as is characteristic of early
33500	stages of explanations. It is also important to note that such  terms
33600	are  commonly utilized in describing computer algorithms which follow
33700	rules in striving to achieve goals. In an  algorithm  these  ordinary
33800	language terms can be explicitly defined and represented.
33900		Psychiatry deals with the practical concerns of inappropriate
34000	action, belief, etc. on the part of a patient. His  behavior  may  be
34100	inappropriate  to  onlookers  since  it  represents  a lapse from the
34200	expected, a contravention of the human order. It may even appear this
34300	way  to  the  patient  in  monitoring  and  directing  himself.   But
34400	sometimes, as in severe cases of the  paranoid  mode,  the  patient's
34500	behavior  does  not  appear  anomalous to himself.  He maintains that
34600	anyone  who  understands  his  point  of  view,  who   conceptualizes
34700	situations  as  he  does  from the inside, would consider his outward
34800	behavior appropriate and justified. What he does  not  understand  or
34900	accept is that his inner conceptualization is mistaken and represents
35000	a misinterpretation of the events of his experience.
35100		The  model  to  be  presented  in  the  sequel constitutes an
35200	attempt to explain some regularities and  particular  occurrences  of
35300	symbolic   (conversational)   paranoid  behavior  observable  in  the
35400	clinical situation of a psychiatric interview.   The  explanation  is
35500	at the symbol-processing level of linguistically communicating agents
35600	and  is  cast  in  the  form  of  a  dialogue  algorithm.   Like  all
35700	explanations,  it  is  tentative,  incomplete,  and does not claim to
35800	represent the only conceivable structure of processes .
35900	
36000	The Nature of Algorithms
36100	
36200		Theories  can  be  presented  in various forms: prose essays,
36300	mathematical  equations  and  computer  programs.    To   date   most
36400	theoretical  explanations in psychiatry and psychology have consisted
36500	of natural language essays with all their  well-known  vagueness  and
36600	ambiguities.  Many  of  these  formulations have been untestable, not
36700	because relevant observations were lacking but because it was unclear
36800	what  the  essay  was really saying.  Clarity is needed.  Science may
36900	begin with metaphors but it should end up with algorithms.
37000		An  alternative  way of formulating psychological theories is
37100	now available in the form of symbol-processing  algorithms,  computer
37200	programs,   which   have  the  virtue  of  being  explicit  in  their
37300	articulation and which can be run on  a  computer  to  test  internal
37400	consistency and external correspondence with the data of observation.
37500	The subject-matter or subject of a model is what it is  a  model  of;
37600	the  source of a model is what it is based upon. Since we do not know
37700	the "real" algorithms used by  people,  we  construct  a  theoretical
37800	model, based upon computer algorithms.     This  model  represents  a
37900	partial  analogy. (Harre, 1970).   The partial analogy is made at the
38000	symbol-processing  level,  not  at  the  hardware  level.           A
38100	functional,   computational   or   procedural  equivalence  is  being
38200	postulated.     The question then becomes  one  of  categorizing  the
38300	extent  of the equivalence.         A beginning (first-approximation)
38400	functional equivalence might be defined  as  indistinguishability  at
38500	the  level  of  observable  I-O  pairs.  A stronger equivalence would
38600	consist of indistinguishability at inner I-O levels.  That is,  there
38700	exists  a  correspondence  between  what  is being done and how it is
38800	being done at a given operational level.
38900		An  algorithm represents an organization of symbol-processing
39000	strategies or functions which represent an "effective procedure".  An
39100	effective procedure consists of three components:
39200	
39300		(1) A programming language in which procedural rules of
39400		    behavior can be rigorously and unambiguously specified.
39500		(2) An organization of procedural rules which constitute 
39600		    the algorithm.
39700		(3) A machine processor which can rapidly and reliably carry
39800		    out the processes specified by the procedural rules.
39900	The  specifications  of  (2),  written  in   the   formally   defined
40000	programming  language  of  (1),  is  termed  an  algorithm or program
40100	whereas (3) involves a computer as the machine processor,  a  set  of
40200	deterministic  physical  mechanisms  which can perform the operations
40300	specified in the algorithm.   The  algorithm  is  called  "effective"
40400	because  it  actually works, performing as intended and producing the
40500	effects desired bt  the  model  builders  when  run  on  the  machine
40600	processor.
40700		A simulation model is composed  of  procedures  taken  to  be
40800	analogous  to  the  imperceptible and inaccessible procedures.     We
40900	are not claiming they ARE analogous,  we  are  MAKING  them  so.  The
41000	analogy  being  drawn  here  is between specified processes and their
41100	generating  systems.  Thus,  in   comparing   mental   processes   to
41200	computational processes, we might assert:
41300	
41400	      mental process    computational process
41500	      --------------:: ----------------------
41600	      brain hardware      computer hardware and
41700	      and programs           programs
41800	
41900		Many of the classical mind-brain problems arose because there
42000	did not exist a familiar,  well-understood  analogy  to  help  people
42100	imagine how a system could work having a clear separation between its
42200	hardware descriptions and its program descriptions.  With the  advent
42300	of  computers  and  programs  some mind-brain perplexities disappear.
42400	(Colby,1971).  The analogy is not simply  between  computer  hardware
42500	and  brain  wetware.   We  are not comparing the structure of neurons
42600	with the structure of transistors; we are comparing the
42700	organization of symbol-processing procedures  in  an  algorithm  with
42800	symbol-processing  procedures of the mind-brain.  The central nervous
42900	system contains a representation of the experience of its holder.   A
43000	model  builder has a conceptual representation of that representation
43100	which he demonstrates in the form of a model.  Thus the  model  is  a
43200	demonstration of a representation of a representation.
43300		An  algorithm  can  be  run  on  a  computer  in two forms, a
43400	compiled version and an interpreted version. In the compiled  version
43500	a  preliminary  translation  has  been  made  from  the  higher-level
43600	programming  language  (source  language)  into  lower-level  machine
43700	language  (object  language)  which  controls  the  on-off  state  of
43800	hardware switching devices. When the compiled  version  is  run,  the
43900	instructions  of  the machine-language code are directly executed. In
44000	the interpreted version each high-level language instruction is first
44100	translated  into  machine language, executed, and then the process is
44200	repeated with the next instruction.   One  important  aspect  of  the
44300	distinction  between  compiled  and  interpreted versions is that the
44400	compiled version, now written in  machine  language,  is  not  easily
44500	accessible  to  change  using  the higher-level language. In order to
44600	change the program, the interpreted version must be modified  in  the
44700	source  language  and then re-compiled into the object language.  The
44800	rough analogy with ever-changing  human  symbolic  behavior  lies  in
44900	suggesting  that  modifications require change at the source-language
45000	level. Otherwise compiled algorithms are inaccessible to second order
45100	monitoring and modification.
45200		Since we are taking running computer programs as a source  of
45300	analogy for a paranoid model, logical errors or pathological behavior
45400	on  the  part   of   such   programs   are   of   interest   to   the
45500	psychopathologist.   These  errors  can  be  ascribed to the hardware
45600	level, to the interpreter, or to the programs  which  the  interpreter
45700	executes.    Different  remedies are required at different levels. If
45800	the analogy  is  to  be  clinically  useful  in  the  case  of  human
45900	pathological  behavior,  it  will  become  a  matter  of  influencing
46000	symbolic behavior with the appropriate techniques.
46100		Since  the algorithm is written in a programming language, it
46200	is hermetic except to a few people,  who  in  general  do  not  enjoy
46300	reading   other  people's  code.     Hence  the  intelligibility  and
46400	scrutability requirement for explanations must be met in other  ways.
46500	In  an attempt to open the algorithm to scrutiny I shall describe the
46600	model in detail using diagrams and interview examples profusely.
46700	
46800	
46900	Analogy
47000	
47100		I  have  stated  that  an  interactive  simulation  model  of
47200	symbol-manipulating  processes  reproduces  sequences   of   symbolic
47300	behavior  at the level of linguistic communication.  The reproduction
47400	is achieved through the operations of an algorithm consisting  of  an
47500	organization   of   hypothetical   symbol-processing   strategies  or
47600	procedures which can  generate  the  I-O  behavior  of  the  subject-
47700	processes   under  investigation. The   algorithm  is  an  "effective
47800	procedure" in the sense it really works in the manner intended by the
47900	model-builders.  In the model to be described, the paranoid algorithm
48000	generates  linguistic  I-O  behavior  typical   of   patients   whose
48100	symbol-processing  is dominated by the paranoid mode. Comparisons can
48200	be made between samples of the I-O behaviors of patients  and  model.
48300	But  the  analogy is not to be drawn at this level.   Mynah birds and
48400	tape recorders also reproduce human linguistic behavior, but  no  one
48500	believes  the  reproduction  is achieved by powers analogous to human
48600	powers.   Given that the manifest outermost I-O behavior of the model
48700	is  indistinguishable  from  the  manifest  outward  I-O  behavior of
48800	paranoid patients, does this imply that the  hypothetical  underlying
48900	processes  used  by  the  model are analogous to (or perhaps the same
49000	as?) the underlying processes used by persons in the  paranoid  mode?
49100	This deep and far-reaching question should be approached with caution
49200	and only when we are  first  armed  with  some  clear  notions  about
49300	analogy, similarity, faithful reproduction, indistinguishability  and
49400	functional equivalence.
49500		In comparing two things (objects, systems or processes )  one
49600	can   cite   properties  they  have  in  common  (positive  analogy),
49700	properties they do not share (negative analogy) and properties  which
49800	we do not yet know whether they are positive or negative (neutral
49900	analogy). (See Hesse,1966). No two things are exactly alike in  every
50000	detail.   If  they  were identical in respect to all their properties
50100	then they would be copies. If they were identical  in  every  respect
50200	including  their  spatio-temporal  location we would say we have only
50300	one thing instead of two. Everything  resembles  something  else  and
50400	maybe everything else, depending upon how one cites properties.
50500		In an analogy a similarity relation is  evoked.  "Newton  did
50600	not  show  the  cause of the apple falling but he showed a similitude
50700	between the apple and the stars."(D`Arcy Thompson). Huygens suggested
50800	an analogy between sound waves and light waves in order to understand
50900	something less well-understood (light) in terms of  something  better
51000	understood   (sound).   To  account  for  species  variation,  Darwin
51100	postulated a  process  of  natural  selection.    He  constructed  an
51200	analogy  from two sources, one from artificial selection as practiced
51300	by domestic breeders of animals and one from  Malthus'  theory  of  a
51400	competition  for  existence  in a population increasing geometrically
51500	while its resources increase arithmetically. Bohr's model of the atom
51600	offered  an  analogy  between solar system and atom. These well-known
51700	historical examples should be sufficient here to illustrate the  role
51800	of analogies in theory construction.    Analogies are made in respect
51900	to  those  properties  which  constitute  the  positive  and  neutral
52000	analogy.     The  negative analogy is ignored.   Thus Bohr's model of
52100	the atom as a miniature planetary system was not intended to  suggest
52200	that  electrons  possessed  color or that planets jumped out of their
52300	orbits. 
52400	
52500	Functional Equivalence
52600	
52700		When human symbolic processes are the subject of a simulation
52800	model, we draw the analogy from two sources, symbolic computation and
52900	psychology.  The  analogy  made  is between systems known to have the
53000	power to process symbols, namely,  persons  and  computers.       The
53100	properties  compared  in  the  analogy  are obviously not physical or
53200	substantive such as blood and wires, but functional  and  procedural.
53300	We  want  to  assume  that not-well-understood mental procedures in a
53400	person are similar to  the  more  accessible  and  better  understood
53500	procedures of symbol-processing which take place in a computer.   The
53600	analogy is one  of  functional  or  procedural  equivalence.  (For  a
53700	further    account   of   functional   analysis   see  Hempel, 1965).
53800	Mousetraps are functionally equivalent.    There exists a  large  set
53900	of  physical  mechanisms for catching mice. The term "mousetrap" says
54000	what each member of the set has in common.  Each  takes  as  input  a
54100	live  mouse  and yields as output a dead one. Systems equivalent from
54200	one point of view may not be equivalent from another (Fodor,1968).
54300		If  model  and  human  are  indistinguishable at the manifest
54400	level of linguistic I-O pairs, then they can be considered equivalent
54500	at  that  level.      If they can be shown to be indistinguishable at
54600	more internal symbolic levels, then a  stronger  equivalence  exists.
54700	How  stringent  and  how extensive are the demands for equivalence to
54800	be? Must the correspondence be point-to-point or can it be  the  more
54900	global system-to-system? Must there be point-to-point correspondences
55000	at every level? What is to count as a point and what are the  levels?
55100	Procedures  can  be  specified  and  ostensively  pointed  to  in  an
55200	algorithm, but how can we point to unobservable symbolic processes in
55300	a  person?    There  is  an  inevitable  limit  to  scrutinizing  the
55400	"underlying"  processes  of  the  world.     Einstein  likened   this
55500	situation to a man explaining the behavior of a watch without opening
55600	it: "He will never be able to  compare  his  picture  with  the  real
55700	mechanism  and  he  cannot even imagine the possibility or meaning of
55800	such a comparison".
55900		In  constructing  an   algorithm   one   puts   together   an
56000	organization  of  collaborating  functions or procedures.  A function
56100	takes some symbolic structure  as  input  and  yields  some  symbolic
56200	structure as output. Two computationally equivalent functions, having
56300	the same input and yielding the same output, can differ "inside"  the
56400	function at the instruction level.
56500		Consider  an elementary programming problem which students in
56600	symbolic computation are often asked to solve.  Given  a  list  L  of
56700	symbols,  L=(A  B  C  D), as input, construct a function or procedure
56800	which will convert this list to the list RL in which the order of the
56900	symbols  is  reversed,  i.e.   RL=(D  C B A).  There are many ways of
57000	solving this problem and the code of one student may  differ  greatly
57100	from that of another at the level of individual instructions. But the
57200	differences of such details are irrelevant. What  is  significant  is
57300	that  the  solutions  make  the required conversion from L to RL. The
57400	correct solutions will  all  be  computationally  equivalent  at  the
57500	input-output  level  since  they take the same symbolic structures as
57600	input and produce the same symbolic output.
57700		If  we  propose  that  an  algorithm  we  have constructed is
57800	functionally equivalent to what goes on in humans when  they  process
57900	symbolic   structures,   how   can   we   justify   this  position  ?
58000	Indistinguishability tests at,  say,  the  linguistic  level  provide
58100	evidence  only for beginning equivalence. We would like to be able to
58200	have access to the underlying processes in humans the way we can with
58300	algorithms.  (Admittedly, we do not directly observe processes at all
58400	levels but only  the  products  of  some).  The  difficulty  lies  in
58500	identifying,  making  accessible,  and  counting  processes  in human
58600	heads.    Many symbol-processing experiments are now  being  designed
58700	and  carried  out.  We  must  have  great  patience with this type of
58800	experimental information-processing psychology.
58900		In  the meantime, besides first-approximation I-O equivalence
59000	and plausibility arguments,  one  might  appeal  to  extra-evidential
59100	support  offering  parallelisms  from neighboring scientific domains.
59200	One can offer analogies between what is known to go on at a molecular
59300	level  in  the  cells  of  living  organisms  and  what goes on in an
59400	algorithm. For example, a DNA molecule  in  the  nucleus  of  a  cell
59500	consists  of an ordered sequence (list) of nucleotide bases (symbols)
59600	coded in triplets termed codons (words). Each element  of  the  codon
59700	specifies  which  amino acid during protein synthesis is to be linked
59800	into the chain of polypeptides making up the  protein.    The  codons
59900	function like instructions in a programming language. Some codons are
60000	known to operate as terminal  symbols  analogous  to  symbols  in  an
60100	algorithm  which  terminate  the  end of a list. If, as a result of a
60200	mutation, a nucleotide base is changed, the usual protein will not be
60300	synthesized.  The  polypeptide  chain  resulting  may  have lethal or
60400	trivial consequences for the  organism  depending  on  what  must  be
60500	passed  on to other processes which require polypeptides to be handed
60600	over to them. Similarly in an algorithm. If a symbol  or  word  in  a
60700	procedure  is incorrect, the procedure cannot operate in its intended
60800	manner.   Such a result may be lethal or  trivial  to  the  algorithm
60900	depending  on  what  information the faulty procedure must pass on at
61000	its interface with other procedures in the overall organization. Each
61100	procedure   in  an  algorithm  is  embedded  in  an  organization  of
61200	collaborating procedures just as are functions in  living  organisms.
61300	We  know that at the molecular level of living organisms there exists
61400	a process such as serial progression  along  a  nucleotide  sequence,
61500	which is analogous to stepping down a list in an algorithm.   Further
61600	analogies can be made between point mutations in which DNA bases  can
61700	be   inserted,   deleted,   substituted  or  reordered  and  symbolic
61800	computation in which the same operations are commonly carried out  on
61900	symbolic    structures.     Such   analogies   are   interesting   as
62000	extra-evidential support but obviously  closer  linkages  are  needed
62100	between  the macro-level of symbolic processes and the micro-level of
62200	molecular information-processing within cells.
62300		To obtain evidence  for  the  acceptability  of  a  model  as
62400	faithful  or  authentic,  empirical  tests are utilized as validation
62500	procedures. Such tests should also tell us which is  the  best  among
62600	alternative  versions  of  a  family  of  models  and,  indeed  among
62700	alternative families of models.  Scientific explanations do not stand
62800	alone  in  isolation. They are evaluated relative to rival contenders
62900	for the position of "best available".  Once we  accept  a  theory  or
63000	model as the best available, can we be sure it is correct or true? We
63100	can never know with certainty. Theories and  models  are  provisional
63200	and  partial  approximations  to  nature  destined  in time to become
63300	abandoned and superseded by better ones.