00100	.SEC THE CENTRAL PROCESSES OF THE MODEL
00200	
00300	
00400	(THIS CHAPTER REQUIRES MANY FLOW DIAGRAMS- SEE BACK OF MS)
00500	
00600		Only   the  major  processes  will  be  described  in  detail
00700	sufficient  to  illustrate  the  logic  of  the   algorithm.     Many
00800	"housekeeping"  procedures  are  needed  to  run  the  model  but  no
00900	understanding of them is necessary to follow the main flow of  symbol
01000	processing.  In  the  next  paragraph  I  will  give some examples of
01100	"housekeeping" only to illustrate what little interest they have  for
01200	the nonspecialist reader.
01300		The  first  theoretically uninteresting procedure executed is
01400	one of intiialization which checks to make  sure  the  data-base  has
01500	been read in and sets a number of variables to their starting values.
01600	Some of these variables serve as flags or  indices  pointing  to  the
01700	topic  under  discussion  or to the last self-topic discussed.  Other
01800	variables are set by the interviewer who can choose to run a weak  or
01900	strong version of the model.  If the weak version is elected, affect-
02000	variables  of  ANGER  and  FEAR  can  be  set  to  `low'  or   `mild'
02100	values,while   MISTRUST   can  be  set  to  `mild'  or  `high'.   The
02200	interviewer also has the option of following the internal workings of
02300	the  model  which  can  be displayed in "windows" on a console. After
02400	this initialization the algorithm prints out `Ready' to  indicate  to
02500	the interviewer he may now enter his input.
02600	
02700		After the  input  expression  is  assigned  a  sentence  type
02800	(statement,  question  or  imperative),  it  then serves as the input
02900	argument to the major procedures which deal with (in  order)  special
03000	reactions,  delusional references, self references, flare references,
03100	interviewer-interviewee  relations,  miscellaneous  expressions   and
03200	self-scanning.
03300	
03400	.F
03500	SPECIAL REACTIONS
03600	
03700		This  procedure  produces  appropriate  reactions  to special
03800	types of input expressions.    If the input consists  simply  of  the
03900	letter  `S' {the means by which an interviewer indicates silence over
04000	a teletype), then the algorithm chooses a  linguistic  response  from
04100	the `Silence' list. The linguistic output responses are not generated
04200	word-by-word.  They  consist  of  pre-formatted  English  expressions
04300	stored in the data-base on ordered lists.
04400	
04500		The procedure which selects the next reply from the  relevant
04600	response list also removes that response from the list so it will not
04700	be output twice.   Thus, in this  case,  where  repeated  silence  is
04800	being  detected  and  if there are no more responses on the `Exhaust'
04900	list {i.e. the `Exhaust' list is itself exhausted} , PARRY    would
05000	end the dialogue.
05100	
05200		An `Exhaust' list represents  a  boundary  condition  in  the
05300	model.  That  is,  since the model has a limited number of linguistic
05400	responses for each topic it can discuss,  when  these  responses  are
05500	exhausted   PARRY   must have some way of dealing with a large number
05600	of  conceptually  equivalent  repetitions  on   the   part   of   the
05700	interviewer.   When a response list is exhausted, the model expresses
05800	a wish to change the topic and, as mentioned, when the `Exhaust' list
05900	itself  is exhausted,    PARRY  ends the dialogue. Since this process
06000	is true of all instances in which the response list is exhausted,  it
06100	will  not  be  mentioned again. I trust the reader will remember that
06200	this what happens each time this boundary condition is reached.
06300	  
06400		The second case handled by this procedure consists  of  input
06500	expressions  in  which  the interviewer states or insinuates that the
06600	model is mentally ill.  This condition is detected by  finding  "you"
06700	and  a  nearby  (within three words) member of the `Abnormal' list in
06800	the input.  The inputs:
06900	
07000	.V
07100	     {1} DR.- YOU NEED TREATMENT.
07200	
07300	     {2} DR.- YOU ARE DELUSIONAL.
07400	
07500	     {3} DR.- DO YOU THINK YOU MIGHT BE PARANOID? 
07600	.END
07700	
07800	would satisfy this condition.
07900		If  the input is a question, as in {3}, ANGER is increased by
08000	an smaller amount of its current value than if  it  is  a  statement,
08100	Thus  a question is interpreted as an implicit insult compared to the
08200	explicit insult of a direct statement.
08300	
08400	     The linguistic response now chosen is selected from the `Alien'
08500	list, e.g.
08600	
08700	.V
08800	     {4} PT.- I THINK I KNOW WHAT YOU DOCTORS ARE UP TO.  
08900	.END CONTINUE
09000	
09100		If conditons for the procedure handling special reactions are
09200	not found to obtain,    the  algorithm  next  attempts  to  recognize
09300	references to delusions.
09400	
09500	.F
09600	DELUSIONAL REFERENCES
09700	
09800		The  strong  version of the model contains in its data-base a
09900	delusional network of beliefs about the Mafia.   The  next  procedure
10000	called  scans  the  input  expression looking for a reference to this
10100	delusional network.      As will be  seen,  reactions  to  the  first
10200	reference  differ  from  reactions  to  subsequent  references.   The
10300	conceptual contentives of the delusional net are  classified  in  the
10400	data-base  into  `strong'  and `ambiguous' terms.  Thus "murder" is a
10500	strong term whereas "bug" (as mentioned in chapter 4), is  ambiguous.
10600	If  delusional  terms are detected in the input, a variable is set to
10700	the list of terms found and the  terms  are  then  deleted  from  the
10800	delusional word list for reasons which will become clear later.
10900	
11000		Two situations in the interview must be distinguished, one in
11100	which a delusional topic occurs for the first time and one  in  which
11200	some  aspect  of  the  delusional net is under discussion or has been
11300	under discussion and is now being taken up again.    Since the  topic
11400	of  the  Mafia is a fear-eliciting, any reference to it for the first
11500	time raises FEAR by an increment much greater than if the  topic  has
11600	already  been discussed. The concept of `mafia' is represented in the
11700	data-base by a node in a weighted and directed conceptual graph.
11800		((DRAWING  OF  GRAPH  HERE)  Horses  →  Horseracing → Bookies
11900	→Gangsters →  Rackets  →  Mafia  ↑  ↑  Gambling  Police  ↑  ↑↑  Money
12000	Italians)).   The  nodes  in  the graph represent "flare" concepts to
12100	which the model is particularly  sensitive.   Associated  with  these
12200	nodes  are  small  stories  which the model can narrate about each of
12300	them as a theme. Nodes closer to the Mafia node are  weighted  higher
12400	to  represent  the notion that they are of greater concern since they
12500	bear more directly on  the  delusional  network.  If  a  Mafia  topic
12600	appears  for the first time , pointers in the directed graph of flare
12700	concepts must be modified accordingly since the Mafia  node  has  the
12800	highest  weight  in  the  graph.    A  topic such as "bookies", while
12900	leading eventually to Mafia beliefs, is of much less importance  than
13000	Mafia-topics.    But  if  "bookies"  comes  up  in the interview, the
13100	algorithm must know  whether  or  not  the  Mafia  has  already  been
13200	discussed.    Also,  if an introductory-topic {see p.0OO} or subtopic
13300	was under discussion when reference to a  Mafia-topic  is  made,  the
13400	algorithm must unset the introductory-topic indicator.
13500	
13600		Since the model strives to tell its story about the Mafia,  a
13700	flag  is  set  to  indicate  that, should the topic be changed by the
13800	interviewer,   PARRY   should return to  the  current  point  in  its
13900	story  under  appropriate  circumstances, e.g.   when the interviewer
14000	asks a non-specific question or requests any information the  patient
14100	wishes to volunteer.
14200	
14300		If the interviewer's input expression contains a reference to
14400	the delusional net, a delusional statement is output.  But which one?
14500	If this is the first time  the  topic  has  come  up,  the  algorithm
14600	outputs  the  first statement of its delusional story.   From then on
14700	the output delusion selected depends on what has been said,  what  is
14800	still  unsaid,  and  what the interviewer has said about the previous
14900	delusional statement.   Thus the most recent delusional statement  is
15000	saved along with expected anaphoric references, anticipating that the
15100	interviewer may subsequently ask a question or make a statement about
15200	it.
15300	
15400		One special case must be noted.  If the values of ANGER, FEAR
15500	and/or MISTRUST are extremely high (above a particular threshold),the
15600	program  will  refuse  to discuss Mafia-topics at all since it is too
15700	`upset' to talk about this most sensitive area.
15800	
15900		To make some of these operations more  intelligible,  let  us
16000	consider  interview examples.  Suppose at some point in the interview
16100	the doctor asks a standard first-interview question as follows:
16200	
16300	.V
16400	     {5} DR.- DO YOU EVER HAVE THE FEELING YOU ARE BEING WATCHED? 
16500	.END
16600	
16700	If this is the first reference  to  the  delusional  net,  FEAR  will
16800	increase greatly and the linguistic response will be:
16900	
17000	.V
17100		{6} PT.- YOU KNOW, THEY KNOW ME.   
17200	.END CONTINUE
17300	
17400	In making this response, the model must expect from the interviewer a
17500	number  of  typical  questions  of  the  wh-type as well as rejoinder
17600	statements.  The use of "they" by the interviewer in his response  to
17700	the model's   output  is  assumed to be an anaphoric reference to the
17800	"they"    PARRY  is  talking  about.    Although  it  is  likely  the
17900	interviewer  will  react  to the model's output of {6}, the algorithm
18000	must be prepared for the possibility that he will change  the  topic.
18100	Hence  if  the  interviewer  at  this  point  asks  some non-sequitur
18200	question such as:
18300	
18400	.V
18500	     {7} DR.- HOW LONG HAVE YOU BEEN IN THE HOSPITAL.  
18600	.END CONTINUE
18700	
18800	the  program recognizes that no reference to the delusional topic has
18900	been made and answers the question just as though it had  been  asked
19000	in  any other context.  This ability to deal with input in a flexible
19100	manner is important because of the many contingencies which can occur
19200	in psychiatric dialogues.
19300	
19400		If   the  topic  is  changed  abruptly  in  this  way  by  an
19500	interviewer, the algorithm `remembers' that it has output  its  first
19600	delusional  statement  of  {6}.    When the interviewer makes another
19700	neutral delusional reference, the next `line' of the delusional story
19800	will be output, e.g.
19900	
20000	.V
20100	     {8} PT.- THE MAFIA REALLY KNOW ABOUT ME.   
20200	.END CONTINUE
20300	
20400	The ability to answer typical wh- and HOW questions  depends  on  how
20500	much  conceptual  information  is  contained in the delusional belief
20600	being addressed.  For example, suppose   PARRY   replied as in {6}
20700	
20800	.V
20900	     {6) PT.- THEY KNOW ABOUT ME.   
21000	.END CONTINUE
21100	
21200	and the interviewer then asked:
21300	
21400	.V
21500	     {9} DR.- WHERE DO THEY KNOW ABOUT YOU? 
21600	.END CONTINUE
21700	
21800	If the expectancy-anaphoras contain no "where", then a question about
21900	location   cannot  be  answered.   In  this  default  situation,  the
22000	algorithm recognizes the anaphoric "they","know" and "you". Hence  it
22100	knows  at least that the topic has not been changed so it outputs the
22200	next statement in the delusional story;
22300	
22400		(9) PT.- THEY KNOW WHO I AM.       
22500	and again anticipates questions and  rejoinders  pertaining  to  this
22600	statement.
22700	
22800		In constructing the data-base of beliefs, we tried to pack as
22900	much information in each belief as any `reasonable' (like  ourselves)
23000	interviewer might request.  However, one cannot anticipate everything
23100	and  when  some  unanticipated  information  is  requested,   another
23200	relevant reply must be substituted. This heuristic may seem less than
23300	perfect but there is little else to do when the  model  simply  lacks
23400	the pertinent information. By the way, humans do this also.
23500	
23600		When the interviewer shows interest in the delusional  story,
23700	PARRY  continues  to  output  assertions appropriate to the dialogue.
23800	However, when the interviewer expresses doubt or disbelief about  the
23900	delusions,  ANGER  and  FEAR  increase  and  the  interviewer becomes
24000	questioned as in:
24100	
24200	.V
24300	     (10) PT.- YOU DON'T BELIEVE ME, DO YOU?  
24400	.END CONTINUE
24500	
24600	Such an output expression attempts to prompt the dialogue towards the
24700	relation  between  the  interviewer  and  the  model  which  will  be
24800	described later ( see p.000).
24900		If no  delusional  reference  at  all  is  detected  by  this
25000	procedure  ,  the algorithm attempts the next function which searches
25100	for certain types of references to the self.
25200	
25300	
25400	.F
25500	SELF REFERENCES
25600	
25700		Since the main concern of a psychiatric interview consists of
25800	the beliefs, feelings, states and actions of the patient,  the  model
25900	must  be able to answer a large number of questions about its `Self'.
26000	
26100	
26200		If  the  input  is  recognized  as a question and no topic is
26300	currently under discussion and the question  refers  to  the  `Self',
26400	then  it  is  assumed  temporarily  that it will refer only to a main
26500	self-topic. These  main  or  "introductory"  self-topics  (age,  sex,
26600	marriage, health,  family,  occupation,  hospital stay, etc.) in turn
26700	have  sub-topics  to  varying  depths.   For  example,  suppose   the
26800	interviewer asks:
26900	
27000	.V
27100	     (12) DR.- HOW DO YOU LIKE THE HOSPITAL?  
27200	.END CONTINUE
27300	
27400	Since "hospital" is a main `introductory' topic with several levels
27500	of sub-topics, the algorithm answers the question with
27600	
27700	.V
27800	     (11) PT.- I SHOULDN'T HAVE COME HERE.   
27900	.END CONTINUE
28000	
28100	and  then  anticipates  a  variety  of likely questions such as "What
28200	brought you to the  hospital?",  "How  long  have  you  been  in  the
28300	hospital?",  "How  do  you  get along with the other patients?", etc.
28400	Each of these questions  brings up  further  topics,  some  of  which
28500	represent  a continuation of the main topic "hospital", but others of
28600	which represent a shift to  another  main  introductory  topic,  e.g.
28700	"other  patients".   Since  many  of  the  inputs  of the interviewer
28800	consist of ellipses or fragments, the algorithm assumes them to refer
28900	to the topic or subtopic under discussion.  If some  topic  is  being
29000	discussed,  the algorithm checks first for a new main topic, then for
29100	a follow-up to the last subtopic, then (unless the subtopic is itself
29200	a  main  topic,  as  for example "other patients" in the above) for a
29300	follow-up to the last main topic.  Thus continuity and  coherence  in
29400	the dialogue are maintained.
29500	
29600		If some meaning cannot be extracted from the question but  it
29700	is recognized at least that a question is being asked, a procedure is
29800	called  which  attempts  to  handle  certain   common   miscellaneous
29900	questions  which  are difficult to categorize.      These include the
30000	space-time orientation questions ("What day is this?")  and  everyday
30100	information   ("Who  is  president?)  asked  by  psychiatrists  in  a
30200	mental-status  examination  to  test  a   patient's   awareness   and
30300	orientation.     Some  quantitative "how" questions ("how many", "how
30400	often", "how long") are here  recognized.   Since  any  adjective  or
30500	adverb can follow a "how", one of the limitations of the model is its
30600	inability to handle all of them satisfactorily because  the  relevant
30700	information  is lacking in the data-base.  If absolutely no clues are
30800	recognized in the question, the  algorithm  is  forced  to  output  a
30900	noncomittal reply such as:
31000	
31100	     (12) PT.- WELL, I DON'T KNOW.    
31200	
31300		This function also checks for statements about the self which
31400	are taken to be insulting or complimentary. Naturally the presence of
31500	a negator in the input reverses the meaning.  Thus
31600	
31700	     (13) DR.- YOU DON'T SEEM VERY ALERT.  
31800	
31900	s classified as an insult whereas
32000	
32100	     (14) DR.- YOU ARE RIGHT.   
32200	
32300	is considered complimentary and benevolent.
32400	
32500		Among the introductory self-topics are those which constitute
32600	sensitive  areas,  e.g. sex, religion and family.  If the interviewer
32700	refers to one of these areas, the value of  ANGER  increases  sharply
32800	and  a  response  is  selected  from  one of the lists categorized as
32900	`hostile', `defensive', `personal' or  `guarded',  depending  on  the
33000	level  of  MISTRUST  at  the moment.  For example, if the interviewer
33100	asks a question about   PARRY'S   sex life, it first replies with:
33200	
33300	   (13) PT.- MY SEX LIFE IS MY OWN BUSINESS.   
33400	
33500	If the interviewer persists or even later tries to ask about sex, the
33600	model will respond with a hostile reply, such as:
33700	
33800	     (14) PT.- DO YOU KNOW WHAT YOU ARE DOING?   
33900	
34000		The  particular  sensitive areas in the model are part of the
34100	initial  conditions  specific  for  this  hypothetical  patient.   Of
34200	course,  these  topics  are  commonly  found to be sensitive areas in
34300	human patients.
34400	
34500		The model operates  sequentially  trying  one  major  process
34600	after  another.  If it has come this far, (that is, having tested for
34700	special reactions, delusional references and self references  without
34800	recognizing  anything in the input pertinent to these procedures), it
34900	proceeds to the next process which handles flare references.
35000	
35100	.F
35200	FLARE REFERENCES
35300	
35400	The data-base contains a directed graph of concepts involved  in  the
35500	model's  `stories'.     PARRY     has  small  stories  to  tell about
35600	horseracing, gambling, bookies, etc.  The  major  concepts  of  these
35700	stories are termed "flare" concepts since they activate stories which
35800	are differentially weighted in the graph.  
35900	
36000	
36100		In the strong version of the model, the  concept  `Mafia'  is
36200	given  the  highest  weight, while  in  the  weak version the concept
36300	`Rackets' is most heavily weighted.   In both versions  `Horses'  has
36400	the  lowest  weight. The weights are assigned to the concepts and not
36500	individual words or word-groups denoting the concepts.
36600	
36700		The  graph  is  directed  in  the  sense  that  reference  to
36800	horseracing elicits the first line of a story about horseracing. When
36900	a story is ended, a prompt is given to the interviewer to discuss the
37000	next story in the graph which involves `bookies'.   The model strives
37100	to tell its  stories  under  appropriate  conditions  and  leads  the
37200	interviewer  along  paths of increasing delusional relevance.    Much
37300	depends on whether the interviewer follows these leads "benevolently"
37400	and reacts to the prompts.
37500	
37600		The  first  step in this procedure is to scan the input for a
37700	flare concept having the highest weight.   Thus if a flare concept is
37800	already under discussion, a weaker new flare will be disregarded.  If
37900	the flare concept is one in a story which has already been  partially
38000	told,  then  a prompt is offered regarding the next story-node in the
38100	graph.
38200	
38300		If a question is asked about the events of a story, the model
38400	tries to answer it.  Also the  model  is  sensitive  to  whether  the
38500	interviewer  is  showing interest in the story or whether he tries to
38600	change  the  subject  or  expresses  a  negative  attitude,  such  as
38700	disbelief.
38800	
38900		If the interviewer indicates a positive attitude towards  the
39000	story,  then  benevolence  is  recognized and the variables of ANGER,
39100	FEAR and MISTRUST  decrease  slightly  after  each  I-O  pair.  ANGER
39200	decreases  more rapidly than FEAR while MISTRUST, being a more stable
39300	variable once it has risen, decreases least.
39400	
39500		If  no  flare concepts are recognized in the input, the model
39600	next tries to detect if a reference is being  made  to  the  relation
39700	between  the  interviewer and the model.  In an interview interaction
39800	there exist  two situations, the one  being  talked  about  and  the one  the
39900	participants  are  in  at  the moment. Sometimes the latter situation
40000	becomes the former, that is, the one talked about.
40100	
40200	.F
40300	INTERVIEWER-INTERVIEWEE RELATIONS
40400		As described in Chapter 4, the algorithm  must  be  ready  to
40500	handle input referring to the relation between interviewer and model.
40600	The simplest cases are exemplified by expressions such as:
40700		(15) DR.- I UNDERSTAND YOU.  
40800		(16) DR.- YOU DO NOT TRUST ME.  
40900	Those phrases in an expression which can appear between "I" and "you"
41000	or between "you" and "me" we classified as representing a positive or
41100	negative  attitude  on  the  part of the interviewer. Thus expression
41200	(15) is taken  to  be  positive  whereas  (16)  is  negative because,
41300	although it contains a positive verb, the verb is negated.
41400		If a positive attitude is expressd by the  interviewer,  FEAR
41500	and  ANGER  decrease.  FEAR  and  ANGER  increase  depending  on  the
41600	conceptualizations of the input. These attitudes of  the  interviewer,
41700	as interpreted by the model, are reflected in the values of the affect
41800	variables.
41900		Associated in the  data  base  with  each  type  of  attitude
42000	expression expected are lists of appropriate output expressions. Thus
42100	in reply to:
42200		(16) DR.- I UNDERSTAND YOU. 
42300	the model would reply:
42400		(17) PT.- I'M GLAD YOU DO. 
42500	or
42600		(18) PT.- I APPRECIATE YOUR TRYING TO UNDERSTAND.  
42700	or  some  equivalent  expression  depending  on  values of the affect
42800	variables.  When  ANGER  and  FEAR  are   high,   positive   attitude
42900	expressions  are  interpreted  as insincerity and hence evoke hostile
43000	replies.
43100		The   remainder  of  input  expression  types  thus  far  not
43200	discussed are handled by a procedure for miscellaneous expressions.
43300	MISCELLANEOUS EXPRESSIONS
43400	
43500		This procedure deals with all those  interviewer  expressions
43600	from  which no clear conceptualization can be formed.  The only thing
43700	which can be determined is perhaps the sentence-type  of  the  input.
43800	Presented  with  one  of these expressions, if FEAR is extremely high
43900	PARRY     signs off without  a  farewell  expression  and  cannot  be
44000	contacted through further natural language input.  If FEAR is high
44100	but not extreme, and the input is recognized as a question, the model
44200	chooses a reply from a list which  brings  up  the  attitude  of  the
44300	interviewer as in:
44400		(19) PT.- WHY DO YOU WANT TO KNOW?  
44500	or
44600		(20) PT.- YOU PRY TOO MUCH.  
44700	If  the  input is recognized as a statement, a reply is chosen from a
44800	list which indicates some degree of anxiety:
44900		(21) PT.- WHO ARE YOU REALLY?  
45000		(22) PT.- YOU ARE MAKING ME NERVOUS.  
45100	If ANGER is high and the input is a question, a reply is chosen from
45200	a list designed to express hostility as in:
45300		(23) PT.- DO YOU KNOW WHAT YOU ARE DOING?  
45400		(24) PT.- PERHAPS YOU ARE JUST POSING AS A DOCTOR.  
45500		Sometimes  in  these  default  conditions the flag set in the
45600	procedure for delusional references allows the model to  continue  by
45700	giving  the next line in its delusional story.  If the story is under
45800	discussion, continuity is maintained.  But if it is  not,  the  model
45900	appears  to  ignore  the  input and jumps back to one of its previous
46000	preoccupations.   In this instance the observed property of  rigidity
46100	is a function of linguistic non-comprehension and not of the paranoid
46200	processes per se.   Increasing  the  model's  ability  to  comprehend
46300	conversational language would remedy this deficiency.
46400		If  a story flag has not been set by a previous discussion in
46500	the interview and ANGER and FEAR are not high, the algorithm tries to
46600	see  if the input is some type of general prompt from the interviewer
46700	such as:
46800		(25) DR.- GO ON.  
46900	or
47000		(26) DR.- TELL ME MORE.  
47100	If so, PARRY     continues with its current story or attempts to
47200	initiate another story. 
47300		If none of these conditions hold, the procedure ANSWER
47400	is called. This procedure handles a group of common special-case
47500	miscellaneous questions such as:
47600		(27) DR.- HOW DO YOU DO?  
47700	and miscellaneous statements such as:
47800		(28) DR.- HI.  
47900		(29) DR.- GOOD EVENING.  
48000	
48100	.F
48200	SELF SCANNING
48300		The final major procedure in the  algorithm  scans  what  the
48400	model  has  chosen  to  output.  That is, it treats its own output as
48500	input.  If this expression contains a flare or delusional  reference,
48600	the appropriate flags are set and FEAR is raised slightly, but not as
48700	much as if this expession had come from the interviewer. In this  way
48800	the  model  "frightens  itself"  by  what it says about a frightening
48900	topic.
49000	
49100			SUMMARY
49200		To  recapitulate  the  operations  of  the  model,  it  first
49300	attempts  a  linguistic  recognition  of  the  input  by  looking for
49400	patterns which  indicate  its  meaning.  The  internal  and  external
49500	reactions of the model depend on whether the meaning is classified as
49600	malevolent, benevolent, or neutral.  Internal  reactions  consist  of
49700	adjusting the values of affect variables of anger, fear and mistrust.
49800	The model also keeps track of the topic under discussion and by means
49900	of  anaphora-expectancy functions, anticpates what might be said. The
50000	external output of a  natural  language  expression  depends  on  the
50100	nature of the input, the topic under discussion and the values of the
50200	affect varaibles.
50300		The  systemicity  of the model is obvious. We now come to its
50400	testability. How can  we  compare  the  model  to  its  subject,  its
50500	naturally-  occurring counterpart, so that we can judge its degree of
50600	correspondence to facts of observation?