00100	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 primary
00900	one is more fundamental since useful applications require some degree
01000	of consistency and correspondence to observation.
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, i.e. 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 were described in Chapter 6.
02200		Decision procedures for   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  to  be  better  than  available  plausible  alternatives.
02700	Validation  is  ultimately  a  private  experience of the individual.
02800	Empirical truth or falsity cannot be proven with certainty but  their
02900	presence  can  be  assayed  by  some  sort  of critical assesment and
03000	deliberation. We can forgive models for being only  nearly  true.   A
03100	theory or model may bring cognitive or pragmatic comfort, not because
03200	it is  TRUE  but  because  it  represents  an  improvement  over  its
03300	contending rivals.
03400		Cognitive comfort is a  type  of  intellectual  satisfaction.
03500	Pragmatic  comfort  accrues from applications to problems in order to
03600	make things work the way humans want  them  to  work  efficiently  in
03700	practical  contexts  of  technological  action. For the pragmatist, a
03800	model is a means to an end;  for  the  theoretician,  an  explanatory
03900	model  is  an end in itself. It is hoped that this paranoid model can
04000	contribute to understanding one of the mysteries  of  human  conduct,
04100	the  paranoid mode. There remains the enigma of the paranoid "streak"
04200	which renders whole nations  susceptible  to  idelogical  convictions
04300	in which Elsewhereans are believed to be malevolent oppressors.
04400		It  is  a  truism  of  methodology  texbooks that an infinite
04500	number of theories or  models  can  account  for  the  same  data  of
04600	observation.      Without   questioning   whether   "infinite"  means
04700	indefinitely large or just more than one, we  must  allow  for  rival
04800	explanations.  For a rival to be a live and tenable option, it should
04900	be truly alternative (i.e., not just a family version saying the same
05000	thing in a different way), have an appreciable prior probability, and
05100	be testable.
05200		Although I  hold  that  faithful  reproduction,  fidelity  as
05300	measured  by indistinguishability, is a proper and major test for the
05400	adequacy of simulation models, it would be a bonus if our model could
05500	satisfy  the  function  of  making  possible  new  knowledge  through
05600	prediction. The term "prediction" has a spectrum of meanings  ranging
05700	from    forecasts     to   prognoses   to   prophecies   to   precise
05800	point-predictions in time. To predict is to announce a  fact  without
05900	prior  knowledge  of  it.  However one needs knowledge of the kind of
06000	fact expected, the conditions which produce it and the  circumstances
06100	under   which   it   will  occur.   Accurate  long-range  predictions
06200	characterize the ideal of celestial mechanics.  But even astronomers,
06300	with  the  advantage  of  isolated and repetitive systems, have their
06400	troubles. In  1759  Halley's  comet  arrived  four  days  later  than
06500	predicted.  In  spite of our advanced 20th century knowledge, in 1962
06600	this pesky comet arrived eight days later than predicted, making  the
06700	prediction worse by a factor of one hundred.
06800		Long-range  predictions  of  individual  human  behavior  are
06900	difficult because (1) sufficient knowledge of initial conditions  may
07000	require  that  we  know  the  whole  past  history  of  an individual
07100	(something not yet achieved for even a single person) (2) individuals
07200	do  not remain isolated over the time stretch of the prediction; they
07300	interact with other individuals of an unknown nature (3)  life  is  a
07400	fortuitous flux of chance intersections of independent causal chains.
07500	In one sense our paranoid model  makes  moment-to-moment  predictions
07600	and  asserts  new  counterfactuals  about  behavior  in a psychiatric
07700	interview. That is, if an interviewer says X under conditions Y, then
07800	the  model's  response will be characterized by z1...zn, and the same
07900	holds true for paranoid patients.   Counterfactual  prediction  means
08000	that  on  the  basis  of  observed  behavior  we  are willing,with an
08100	inductive  risk,  to  assume  the  presence  of  unobserved  behavior
08200	potentials in a model's or patient's repertoire of capabilities.
08300		Predicting  new  kinds  of  events  or properties, instead of
08400	kinds we are already familiar with, would represent a genuine  bonus,
08500	indicating  the  model is more than ad hoc and has excess content. It
08600	would give both clinicians and investigators something to  look  for.
08700	This  novelty  could  arise  in  two  ways.    First, the model might
08800	demonstrate a property  of  the  paranoid  mode  hitherto  unobserved
08900	clinically.      In  principle  this could come about because the I-O
09000	behavior of  the  model  is  a  consequence  of  a  large  number  of
09100	interacting  hypotheses  and  assumptions chosen initially to explain
09200	frequently observed phenomena.    When the elements of such a complex
09300	conjunction  interact  with  highly  variable  inputs  they  generate
09400	consequences in addition to those  they  were  designed  to  explain.
09500	Whether  any  of these consequences are significant or characteristic
09600	of the paranoid mode remains a subject for future study.
09700		It  is  also  possible that a new property of paranoia may be
09800	discovered in the clinical  interview,  although  perhaps  everything
09900	that  can  be  said about paranoid dialogues has been said.  If a new
10000	property were found, a search  for  it  might  be  conducted  in  the
10100	model's  behavior. if successful, this again would add to the model's
10200	acceptability.
10300		A  second novelty might arise in the behavior of the model in
10400	some new situation.   Since it is designed to simulate  communicative
10500	behavior in an interview situation, the `new' circumstance would have
10600	to involve some new type of linguistic interaction which the model is
10700	capable  of  responding  to. From its behavior one might then predict
10800	how paranoid patients would behave under similar circumstances.   The
10900	requisite  empirical  tests  and  measures  would  show the degree of
11000	correspondence between patient and model behaviors.
11100		This possibility is of importance in considering emancipatory
11200	therapies  for  patients  tangled  in  the quandaries of the paranoid
11300	mode.    Since the model operates at a symbol processing level  using
11400	natural  language,  it  is  at  this  level  at  which linguistic and
11500	conceptual skills of clinicians can be  applied.   Language-based  or
11600	semantic  techniques  do not seem very effective in the psychoses but
11700	they are useful in states of lesser severity. A  wide  range  of  new
11800	semantic  techniques, including extremes, could be tried first on the
11900	model without subjecting patients to blind experimentation.
12000		While  we  used  the  model  to explore a theory and to study
12100	psychiatric judgements, its potential use as a  training  device  has
12200	not  escaped  our  reflections.   Medical  students  and  psychiatric
12300	residents need "disposable patients" to practice on without  jeopardy
12400	(to  either). A version of the paranoid model can display the changes
12500	in its inner states during an interview.    Whether the optimal  goal
12600	of interviewing (gathering relevant information without upsetting the
12700	patient), has been achieved, thus can  be  estimated.    A  beginning
12800	interviewer  can  practice  in  private or with a supervisor present.
12900	Many interviewers have reported that the model has a definite  effect
13000	on  them.    The  student  can get the feel of the paranoid mode long
13100	before he interviews an actual  patient.     The  effect  of  various
13200	interviewing styles might be studied and compared.
13300	
13400		Although this simulation of  paranoia  covers  a  variety  of
13500	facts,  it  is  circumscribed  in  what  it  attempts to explain. The
13600	proffered explanation is local and restricted in that it accounts for
13700	only  one  type  of  symbol-processing mode.  Past attempts at grand-
13800	scale explanations of all  mental  processes  in  all  contexts  have
13900	failed.  A  preferable  strategy, successful in other sciences, is to
14000	build one circumscribed and tested theory or model at a time so  that
14100	the  field  can  gradually  move  forward a step at a time, each step
14200	gaining consensus before attempting the next.