00100	.SEC MODEL EVALUATION 
00200	
00300		The primary aim in constructing this model  was  to  explore,
00400	clarify, develop, test and improve a theory having explanatory value.
00500	To  satisfy  this  aim,  the  model  must  meet  norms  of   internal
00600	consistency  (systemicity)  and norms of external correspondence with
00700	observation (testability). A secondary aim  would  involve  pragmatic
00800	norms  of application.  These aims are not unrelated but the first is
00900	more fundamental since useful applications  require  some  degree  of
01000	consistency and authenticity.
01100		As  emphasized  in  Chapter  2, a  model  in  the  form of an
01200	algorithm consists of a structure of functions  or  procedures  whose
01300	inner  workings  are  sufficient  to  reproduce  the outward symbolic
01400	behavior under consideration. The theory embodied  in  the  model  is
01500	revealed  in  the set of statements which illuminate  the connections
01600	betweeen input and output by  describing  how  the  structure  reacts
01700	under various circumstances.
01800		What  constitutes a satisfactory explanation has been treated
01900	in 2.1.  The "fit" or correspondence with phenomena as  indicated  by
02000	measurements and empirical tests indicate the degree of faithfulness
02100	of the reproduction as described in Chapter 6.
02200		Decision procedures for a consensus acceptability of a  model
02300	sometimes  depend  not  so  much  on  truth,  an elusive state, as on
02400	whether a majority of the  relevant  expert  community  believes  the
02500	theory  or  model to approximate truth to some unknown and unknowable
02600	degree and be better than available plausible alternatives. Empirical
02700	truth  or  falsity cannot be proven with certainty but their presence
02800	can be assayed by some sort of critical assesment  and  deliberation.
02900	We  can  forgive  models  for  being  only nearly true. Validation is
03000	ultimately a  private  experience.   A  theory  or  model  may  bring
03100	cognitive or pragmatic comfort, not because it is TRUE but because it
03200	represents an improvement  over  its  contending  rivals.   Cognitive
03300	comfort  is  a  type  of  intellectual  satisfaction.       Pragmatic
03400	comfort accrues from applications to problems in order to make things
03500	work  the  way  humans  want  them  to  work efficiently in practical
03600	contexts of technological action. For the pragmatist  a  model  is  a
03700	means  to an end; for the theoretician an explanatory model is an end
03800	in itself. It is hoped that this paranoid model  will  be  useful  in
03900	understanding  one  of  the  mysteries of human conduct, the paranoid
04000	mode and the paranoid "streak" which renders so many  susceptible  to
04100	idelogical delusions in which others appear as malevolent oppressors.
04200		It is commonly stated that an infinite number of theories  or
04300	models  can  account  for  the  same  data  of  observation.  Without
04400	questioning whether "infinite" means indefinitely large or just  more
04500	than  one,  we must allow for rival explanations. For a rival to be a
04600	live and tenable option, it should  be  truly  alternative  (not  just
04700	saying  the same thing in a different way), have an appreciable prior
04800	probability, and be testable.
04900		Although  I  hold  that  faithful  reproduction,  fidelity as
05000	measured by indistinguishability, is a proper and  major  test  for
05100	the  adequacy  of  symbolic  models, it would be a bonus if our model
05200	could satisfy the function of  making possible  new  knowledge  through
05300	prediction. Like validation, the term "prediction" has many  meanings
05400	ranging  from  forecasts,  to  prognoses  to  prophecies  to  precise
05500	point-predictions such as an eclipse. To predict  is  to  announce  a
05600	fact  without  prior  knowledge of it. However one needs knowledge of
05700	the kind of fact expected, the conditions which produce  it  and  the
05800	circumstances   under   which  it  will  occur.  Accurate  long-range
05900	predictions characterize the ideal of celestial mechanics.  (But even
06000	astronomers,  with  their isolated and repetitive systems, have their
06100	troubles; Halley's comet in the 18th century arrived four days  later
06200	than  predicted  while in the 20th century it was eight days later!).
06300	Long-range predictions of individual  human  behavior  are  difficult
06400	because  (1)  sufficient  knowledge of initial conditions may require
06500	that we know the whole past history of a person-  something  not  yet
06600	achieved, (2) individuals do not remain
06700	isolated over the time stretch of the prediction; they interact  with
06800	other  individuals of an unknown nature (3) life is a fortuitous flux
06900	of chance intersections of independent causal chains.  In  one  sense
07000	our  paranoid  model  makes moment-to-moment predictions and asserts
07100	new counterfactuals  about  behavior  in  a  psychiatric  interview.
07200	That is, if an interviewer  says  X  under  conditions  Y,  then  the
07300	model's response will be  characterized  by  z1...zn.  Counterfactual
07400	prediction  means  that  on  the  basis  of  observed behavior we are
07500	willing,with an inductive risk, to assume the presence of  unobserved
07600	behavior potentials in the model's repertoire of capabilities.
07700		Predicting new kinds of  events  or  properties,  instead  of
07800	kinds  we are already familiar with, would represent a surplus bonus.
07900	It would give both clinicians and  investigators  something  to  look
08000	for.  This novelty could arise in two ways.    First, the model might
08100	demonstrate a property  of  the  paranoid  mode  hitherto  unobserved
08200	clinically.     In  principle  this  could come about because the I/O
08300	behavior of  the  model  is  a  consequence  of  a  large  number  of
08400	interacting  hypotheses  and  assumptions chosen initially to explain
08500	frequently observed phenomena.    When the elements of such a complex
08600	conjunction  interact  with  highly  variable  inputs  they  generate
08700	consequences in addition to those  they  were  designed  to  explain.
08800	Whether  any  of these consequences are significant or characteristic
08900	of the paranoid mode remains a subject for future study.
09000		It is also possible that a new property of  paranoia  may  be
09100	discovered  in  the  clinical  interview, although perhaps everything
09200	that can be said about it has been  said.  If  a  new  property  were
09300	found, a search for it might be conducted in the model's behavior. if
09400	successful, this would add to the model's  acceptability.
09500		A  second novelty might arise in the behavior of the model in
09600	some new situation.   Since it is designed to simulate  communicative
09700	behavior in an interview situation, the `new' circumstance would have
09800	to involve some new type of linguistic interaction which the model is
09900	capable  of  responding  to. From its behavior one might then predict
10000	how paranoid patients would behave under similar circumstances.   The
10100	requisite  empirical  tests  and  measures  would  show the degree of
10200	correspondence between patient and model behaviors.
10300		This  possibility  is  of importance in considering therapies
10400	for patients tangled in the quandaries of the paranoid mode.    Since
10500	the  model  operates  at  a  symbol  processing  level  using natural
10600	language, it is at this level at which linguistic and semantic skills
10700	of  clinicians  can be applied. Language-based or semantic techniques
10800	do not seem very effective in the psychoses but they  are  useful  in
10900	states  of  lesser severity. A wide range of new semantic techniques,
11000	including extremes,  could  be  tried  first  on  the  model  without
11100	subjecting patients to blind experimentation.
11200		While our group has used the model to explore a theory and to
11300	study  psychiatric judgements, its potential use as a training device
11400	has not escaped our  attention.   Medical  students  and  psychiatric
11500	residents need "disposable patients" to practice on without jeopardy.
11600	The paranoid model can print out a trace of its inner  states  during
11700	and after an interview.   Whether the optimal  goal  of  interviewing
11800	(gathering  relevant  information without upsetting the patient), has
11900	been achieved, thus can be estimated.  A  beginning  interviewer  can
12000	practice  in private or with a supervisor present.  Many interviewers
12100	have reported that the model has  a  definite  effect  on  them.  The
12200	student  can  get  the  feel  of  the  paranoid  mode  long before he
12300	interviews an actual patient.   The effect of various interviewing
12400	styles can be studied and compared. 
12500	
12600		It should be clear by now that this simulation  of  paranoia,
12700	while  circumscribed in what it attempts to explain, covers a variety
12800	of facts. The explanation is local and restricted in that it accounts
12900	for only one type of symbol-processing mode.   Past attempts at grand
13000	scale  explanations  of  all  mental  processes  in all contexts have
13100	failed.   We need to build one circumscribed  and  tested  theory  or
13200	model  at  a time so that the field can gradually move forward a step
13300	at a time, each step gaining consensus  before  attempting  the  next
13400	step.