perm filename CHAP5[4,KMC]11 blob sn#034777 filedate 1973-04-06 generic text, type T, neo UTF8
00100			CHAPTER FIVE
00200	
00300	
00400		THE PROCESSES OF THE MODEL
00500	
00600	
00800	(THIS CHAPTER REQUIRES MANY FLOW DIAGRAMS)
01000	
01100	INITIALIZATION
01200	
01300		The first  procedure executed is one of intiialization which
01400	checks to make sure the data-base has been read in and  sets
01500	a number of variables to their starting values. Some of these variables
01600	serve as flags or indices pointing to the topic under discussion or to the last
01700	self-topic discussed.
01800	Other variables are set by the interviewer who can choose to run
01900	a weak or strong version of the model.  If the weak version is
02000	elected, the variables  of ANGER and FEAR can  be set to `low` or
02100	`mild' values while MISTRUST can  be set to `mild' or `high'.
02200	The interviewer also has the option of following the changes in
02300	these variables by setting a trace-variable to`Y'.  After each
02400	linguistic response of the model, the values of ANGER, FEAR and
02500	MISTRUST will then be displayed.
02600	
02700	     After this initialization the algorithm prints out `Ready'
02800	to indicate to the interviewer he may now enter his input expression.
02900	While the variable ENDE is not true (Sylvia Weber spent a year in
03000	Germany), the algorithm continues to run.  The algorithm signs
03100	off when ENDE is set to true by detecting a farewell message in
03200	the input or when FEAR rises to an extremely high value.
03300	
03310		(Change next paragraph to fit CHAP6 on the language analyzer)
03400	     The interviewer's input expression is read by a function
03500	which scans a list of characters and returns the scanned input in 
03600	the form of a list of words.  The next function sets up the type
03700	of `sentence' the input constitutes, a statement, a question or
03800	`illegal'.  If illegal characters {e.g. a number or a slash}
03900	are detected, the algorithm prints out `Bad input; try again',
04000	indicating to the interviewer that his input expression contains
04100	some unacceptable character.  A statement consists of a list of
04200	words followed by a period.  A question consists of {1} a list of
04300	words followed by a question mark, or {2} a list of words beginning
04400	with a wh-form {who, what, where, when, why} or how, or {3} a list
04500	of words beginning with an imperative verb {e.g. tell} followed by
04600	an expression lacking an actor {tell me about yourself}.  The
04700	program inserts a `Q' at the head of the list at this point.
04800	As described in_______, interrogative imperatives are treated
04900	as questions.  The sentence-type is assigned to the variable
05000	REMARK which then serves as the input argument to the functions
05100	{in order} SPECIALREACTION, DELUSIONALREFERENCE, SELFREFERENCE,
05200	FLARE REFERENCE PERSONALRELATION and NORMAL.  The algorithm then
05300	attempts to process the input expression in the above order.
05400	
05500	SPECIALREACTION
05600	
05700	     This procedure provides the appropriate reactions {linguistic,
05800	affective and belief} to special types of input expressions.  If
05900	the input consists simply of the letter `S' {the means by which an
06000	interviewer indicates silence} then the algorithm chooses a 
06100	linguistic response from the Silence list.
06200	
06300	     The procedure  CHOOSE  selects the next reply from the
06400	relevant linguistic response list termed 'Replies'.  The argument
06500	from `Replies' to the procedure  CHOOSE  is first checked to
06600	see if it consists of an atom.  If it is not an atom then the head
06700	of the list is chosen for the response and removed from the response
06800	list {so it will not be output twice}.  Thus, in this case, where
06900	repeated silence is being detected and if there are no more responses
07000	on the `Exhaust' list {i.e. the 'Exhaust' list is itself exhausted}
07100	,in this case where the `silence' list is being examined, the variable
07200	is set to T, thus ending the dialogue completely since, as mentioned
07300	the algorithm runs as long as ENDE is not true.  In this
07400	case, the output by the function `Say' would be the expression  `I have
07500	had enough of this', and the program signs itself off so that the
07600	interviewer can obtain no response from it.  Thus continued use of the
07700	`silent treatment' by an interviewer causes him to lose his patient.
07800	The procedure `Say' simply prints out the linguistic response  and
07900	saves the interview up to this point in case the computer system
08100	interrupted rather than having to start all over again. {  Most
08200	annoying but necessary in time-shared systems which crash unpredictably}.
08300	
08400	     An `Exhaust' list represents  a boundary condition in the model.
08500	That is, since the model has a limited number of linguisic responses
08600	for each topic it can discuss, when these responses are exhausted
08700	the model must have some way of dealing with a large number of 
08800	conceptualized repetitions on the part of the interviewer.  When a
08900	response list is exhausted, the model indicates a wish to change the
09000	topic and when the `Exhaust' list itself is exhausted, the model ends
09100	the dialogue.  
09200	  
09300	     The second case that `Specialreaction' handles consists of input
09400	expressions in which the interviewer states or insinuates that the model
09500	is mentally ill.  This condition is detected by finding "you" and a
09600	member of the `Abnormal' list in the input.  The inputs:
09700	
09800	     {1}Dr.- "You need treatment."
09900	
10000	     {2}Dr.-"You are delusional."
10100	
10200	     {3} Dr.-"Do you think you might be paranoid?"
10300	
10400	would satisfy this condition.
10500	If the input is  a question, as in {3}, ANGER is increased by an increment
10600	of 0.3 of its current value, while if it is a statement, the increment
10700	is 0.5.  Thus a question is interpreted as an implicit insult compared
10800	to the explicit insult of a direct statement.
10900	
11000	     The linguistic response now chosen is selected from the `Alien'
11100	list, e.g.
11200	
11300	     {4} Model- "I think I know what you doctors are up to."
11400	
11500	As described above, if the list is exhausted, the algorithm goes to
11600	the `Exhaust' list.  Since this process is true of all instances in
11700	which the response list is exhausted, it will not be mentioned again.
11800	We hope the reader will remember this is what happens each time this
11900	boundary condition is reached.
12000	
12100	     If the procedure SPECIALREACTION is not found to be true, the
12200	algorithm next attempts the function DELUSIONALREFERENCE.
12300	
12310	DELUSIONALREFENENCE
12320	
12400	     The strong  version of the model contains in its data-base
12500	a delusional network of beliefs about the Mafia.  This function
12600	scans the input expression looking for a reference to this
12700	delusional network.  As will be seen, reactions to the first
12800	reference differs from reactions to subsequent references.
12900	
13000	     The words {nouns and verbs} and word-groups of the delusional
13100	net are classified in the data-base into `strong' and `ambiguous'
13200	terms.  Thus "murder" is a strong term while "bug" is ambiguous.
13300	{Depending on the context "bug" can be interpreted to mean annoy,
13400	insect or wiretap}.  If delusional terms are detected in the input
13500	a local variable FOUND is set to the list of terms found and the
13600	terms are than deleted from the delusional word list for reasons
13700	which will become clear later.
13800	
13900	     Two situations in the interview must be distinguished, one in
14000	which a delusional topic occurs for the first time and the second
14100	in which some aspect of the delusional net is under discussion or has
14200	been under discussion and is now being taken up again.  Since the topic
14300	of the Mafia is fearful, any reference to it for the first time raises
14400	FEAR by an increment much greater than if the topic has already been
14500	discussed.  If a Mafia topic appears for the first time pointers in the
14600	directed graph of flare concepts {see OO. for a fuller descriptions}
14700	must be modified accordingly since the Mafia node has the highest
14800	weight in the graph.  Briefly, a topic such as "bookies", while
14900	leading eventually to Mafia beliefs, is of much less importance
15000	as determined by a weight than Mafia-topics.  But if "bookies" comes
15100	up in the interview, the algorithm must know whether or not the Mafia
15200	has already been discussed.  Also, if an introductory-topic {see   OO}
15300	or subtopic was under discussion when reference to a Mafia-topic is
15400	made, the algorithm must unset the introductory-topic indicator.
15500	
15600	     Since the model strives to tell its story about the Mafia, a
15700	flag is set to indicate that, if the topic is changed by the interviewer,
15800	the model will return to this point in its story under appropriate
15900	circumstances, e.g. when the interviewer asks a non-specific question
16000	or requests any information the patient wishes to volunteer.
16100	
16200	     If the interviewer's input expression contains a reference to
16300	the delusional net, a delusional statement is output.  But which one?
16400	
16500	If this is the first time the topic has come up, the algorithm outputs
16600	the first statement of its delusional story.  From then on the output
16700	delusion selected depends on what has been said, what is still unsaid
16800	and what the interviewer has said about the previous delusional
16900	statement.  Thus the most recent delusional statement is saved,
17000	anticipating that the interviewer may ask a question or make a
17100	statement about it.
17200	
17300	     One special case must be noted.  If the values of ANGER, FEAR
17400	and/or MISTRUST are extremely high, above a particular threshold,
17500	the program will refuse to discuss Mafia-topics at all since it is
17600	too `upset' to talk about this most sensitive area.
17700	
17800	     To make some of these complexities less opaque, let us consider
17900	interview examples.  Suppose at some point in the interview the doctor
18000	asks a standard first-interview question as follows:
18100	
18200	     {5}Dr. - "Do you ever have the feeling you are being watched?
18300	
18400	If this is the first reference to the delusional net, FEAR will increase
18500	greatly and the linguistic response will be:
18600	
18700	{6} Model - "They know me."
18800	
18900	In making this response, the model must expect from the interviewer a number of typical
19000	questions of the WH-type as well as rejoinder statements.  The use of
19100	"they" by the interviewer in his response to the model's output is
19200	assumed to be an anaphoric reference to the "they" the model is talking
19300	about.  Although it is likely the interviewer will react to the model's
19400	output of {6}, the algorithm must be prepared for the possibility
19500	that the interviewer will change the topic.  Hence if the interviewer
19600	at this point asks some non-sequitur question such as:
19700	
19800	     {7} Dr.- "How long have you been in the hospital?"
19900	
20000	the program recognizes that no reference to the delusional topic has
20100	been made and answers the question just as it would if it were asked
20200	in any other context.  This ability to deal with input in a flexible
20300	context-independent manner is important because of many contingencies
20400	which can occur in psychiatric dialogues.
20500	
20600	     If the topic is changed abruptly in this way by an interviewer,
20700	the algorithm `remembers' that it has output its first delusional statement
20800	of {6}.  When the interviewer makes another neutral delusional reference,
20900	the next `line' of the delusional story will be output, e.g.
21000	
21100	     {8} Model - "The Mafia really know about me."
21200	
21300	     The ability to answer typical WH-and HOW questions depends on how much
21400	conceptual information is contained in the delusional belief being
21500	addressed.  For example, suppose the model replied as in {6}
21600	
21700	     {6 Model - "They know about me."
21800	
21900	and the interviewer then asked:
22000	
22100	     {9}Dr.- "Where do they know about you?"
22200	
22300	If the belief in the data-base contained  no location, i.e. the belief
22400	consists of the conceptualization:
22500	
22600	     ({THE MAFIA KNOW ABOUT ME)}
22700	
22800	then a question about location cannot be answered.  In this default
22900	situation, the algorithm sees the anaphoric "they" and can match the
23000	input phrase "know about you" with the conceptualization phrase
23100	({know about me}).  Hence it knows at least that the topic has not
23200	been changed so it outputs the next statement in the delusional story;
23300	
23400	     (9) Model - "They know who I am."
23500	and again anticipates WH-or HOW questions and rejoinders pertaining to _this   
23600	statement.
23700	
23800	     In constructing the data-base of beliefs, the model-builder tries
23900	to pack as much information in each belief as any `reasonable'(like ourselves) interviewer
24000	question might request.  However, one cannot anticipate everything and
24100	when some slot (see oo.) in the belief is empty another reply must be
24200	output.This heuristic may seem inadequate but there is little else to
24300	do when the model simply lacks the pertinent information, just as do
24400	humans.
24500	
24600	     When the interviewer shows interest in the delusional story, the
24700	model continues to output assertions appropriate to the dialogue.
24800	However, when the interviewer expresses doubt or disbelief about the
24900	delusions, ANGER and FEAR increase and the interviewer becomes questioned
25000	as in
25100	
25200	     (10) Model- "You don't believe me, do you?"
25300	
25400	Such an output expression attempts to prompt the dialogue towards the
25500	relation between the interviewer and the model which will be described
25600	later in  00.
25700	    If no delusional reference at all is detected by this procedure
25800	, the algorithm attempts the next function which searches for certain
25900	types of references to the 'self'.
26000	
26100	
26200	SELFREFERENCE
26300	
26400	     Since the main concern of a psychiatric interview consists of the
26500	beliefs, feelings and actions of the patient, the model must be able
26600	to answer a large number of questions about the 'Self'.  It is
26700	characteristic of a psychiatric interview that questions may not
26800	syntactically be questions but in the form of interrogative
26900	imperatives:
27000	
27100	    (10) Dr. - "Tell me more about the hospital"
27200	
27300	or statements indicating the interviewer has a question:
27400	
27500	     (11) Dr.- "I would like to ask you about your family."
27600	
27700	
27800	     If the input is recognized as a question and no topic is currently
27900	under discussion and the question refers to the 'Self', then it is
28000	assumed temporarily that it will refer only to a main self-topic.  These
28100	main self-topics (age, sex, marriage, health, family, occupation,
28200	hospital stay,etc.) in turn have sub-topics to varying depths.  For
28300	example, suppose the interviewer asks:
28400	
28500	     (12) Dr.- "How do you like the hospital."
28600	
28700	Since "hospital" is a main `introductory' topic with several levels
28800	of sub-topics, the algorithm answers the question with
28900	
29000	     (11)Model - "I shouldn't have come here."
29100	
29200	and then anticipates a variety of likely questions such as "what
29300	brought you to the hospital?", "how long have you been in the
29400	hospital?", "how do you get along with the other patients?", etc.
29500	Each of these questions bring up further topics, some of which
29600	represent a continuation of the main topic "hospital", but others
29700	of which represent a shift to another main introductory topic, e.g.
29800	"other patients".  Since many of the inputs of the interviewer consist
29900	of ellipses or fragments, the algorithm assumes them to refer to the
30000	topic or subtopic under discussion.  If some topic is being discussed,
30100	the algorithm checks first for a new main topic, then for a follow-up
30200	to the last subtopic, then (unless the subtopic is itself a main topic,
30300	
30400	as for example "other patients" in the above) for a follow-up to the
30600	last main topic.  Thus a continuity and coherence to the dialogue is
30700	maintained.
30800	
30900	     If some meaning cannot be abstracted from the question but it is
31000	recognized that a question is being asked, a function is called which
31100	attempts to handle certain common miscellaneous questions which are
31200	difficult to categorize.  These include the space-time orientation
31300	questions ("what day is this?") and arithmetic tasks ("subtract seven 
31400	from one hundred and seven from that number and so on") typical of
31500	current mental-status examinations.  Quantitative "how" questions
31600	("how many", "how often", "how long") are here recognized but one of
31700	the weaknesses of the model consists of its general inability to reply
31800	to them satisfactorily because the relevant information is lacking in
31900	the data-base.  If absolutely no clues are recognized in the question,
32000	the algorithm is forced to output a noncomittal reply such as:
32100	
32200	     (12) Model - "Well, I don't know."
32300	
32400	     This function also checks for statements about the self which are
32500	taken to be insulting or complimentary. Naturally the presence of a
32600	negator in the input reverses the meaning.  Thus
32700	
32800	     (13) Dr.- "You don't seem very alert."
32900	
33000	is classified as an insult whereas
33100	
33200	     (14) Dr. - "You are right."
33300	
33400	is considered complimentary and benevolent.
33500	
33600	     Among the so-called introductory topics are those which constitute
33700	sensitive areas, e.g. sex, religion and family.  If the interviewer
33800	refers to one of these areas, the value of ANGER increases sharply and a
33900	response is selected from one of the lists categorized as 'hostile',
34000	'defensive', 'personal' or 'guarded', depending on the level of MISTRUST
34100	at the moment.  For example, if the interviewer asks a question about
34200	the model's sex life, it first replies with
34300	
34400	   (13) Model - "My sex life is my own business."
34500	
34600	If the interviewer persists or even later tries to ask about sex, the
34700	model will respond with a hostile reply, such as:
34800	
34900	     (14)Model - "Do you know what you are doing?"
35000	
35100	The particular sensitive areas in the model are part of the initial
35200	conditions specific for this hypothetical patient.  Of course, these
35300	topics are commonly found to be sensitive areas in human patients.
35400	
35500	     The model operates sequentially trying one major function after
35600	another.  If it has come this far, after trying  SPECIALREACTION
35700	DELUSIONALREFENENCE and SELFREFERENCE without recognizing anything
35800	in the input pertinent to these functions, it proceeds to the next,
35900	FLAREREFERENCE.
36000	
36100	FLAREREFERENCE
36200	
36300	     The data-base contains a directed graph of concepts involved in the
36400	model's 'stories'.  The model has small stories to tell about horseracing,
36500	gambling, bookies, etc.  The major concepts of these stories are termed "flare"
36600	concepts since they activate stories which are differentially weighted
36700	in the graph.  The graph can be pictured as in Fig. (  )
36800	
36900	    Horses → Horseracing → Bookies →Gangsters → Rackets  →  Mafia
37000	                               ↑       ↑
37100	                           Gambling  Police
37200	                ↑                    ↑↑
37300	               Money              Italians
37400	
37500	
37600	In the  strong version of the model, the concept 'Mafia' is given the
37700	highest weight while in the weak version the concept 'Rackets' is most
37800	heavily weighted.  In both versions 'horses' has the lowest weight.
37900	
38000	The weights are assigned to the concepts and not individual
38100	words or word-groups denoting the concepts.
38200	
38300	     The graph is directed in the sense that reference to horseracing
38400	elicits a story about horseracing. When it is ended a prompt is given to the interviewer to
38500	discuss the next story in the graph involving 'bookies'.  The model
38600	strives to tell its stories under appropriate conditions and leads
38700	the interviewer along paths of increasing delusional relevance.  Much
38800	depends on whether the interviewer follows these leads "benevolently" and reacts to the
38900	prompts.
39000	
39100	     The first step in this procedure is to scan the input for a flare
39200	concept having the highest weight.  Thus if a flare concept is already
39300	under discussion, a weaker new flare will be disregarded.  If the flare
39400	concept is one in a story which has already been told, then a prompt is
39500	offered regarding the next story-node in the graph.
39600	
39700	     If a question is asked about the events of a story, the model tries
39800	to answer it.  Also the model is sensitive to whether the interviewer
39900	is showing interest in the story or whether he tries to change the
40000	subject or (worse) expresses a negative attitude, such as disbelief.
40100	
40200	     If the interviewer indicates a positive attitude towards the
40300	story, then benevolence is recognized (see p   ) and the variables of
40400	ANGER, FEAR and MISTRUST fall slightly after each I-O pair.  ANGER
40500	falls more rapidly than FEAR while MISTRUST, being a more stable
40600	variable once it has risen, falls least.
40700	
40800	     If no flare concepts appear in the input, the model next tries to detect
40900	if a reference is being made to the relation between the interviewer
41000	and the model.  In an interview interaction there exists two situations,
41100	one being talked about and one the participants are in at the moment.
41200	Sometimes the latter situation becomes the former, that is, the one
41300	talked about.
41400	
41500	INTERVIEWRELATION
41600		As described in %00(chapter on language analyzer) the
41700	algorithm must be ready to handle inputnreferring to the relation
41800	between interviewer and model. The simplest cases are exemplified
41900	by expressions such as:
42000		(15) Dr.- "i understand you."
42100		(16) Dr.- "You do not trust me."
42200	Those phrases in an expression which can appear between "I" and
42300	"you" or between "you" and "me" we classified as representing a
42400	positive or negative attitude on the part of the interviewer. Thus
42500	expression (15) is taken to be positive whereas (16) is negative
42600	because although it contains a positive verb the verb is negated.
42700		The algorithm must distinguish between one-verb and two-
42800	verb expressions with certain common verbs, for example:
42900		(16) Dr.-" I believe you."
43000		(17) Dr.- "I believe you are wrong".
43100	In (16) a positive attitude is expressed whereas (17) is negative.
43200	Of course the language analyzer makes the correct identifications
43300	of actor and object in these types of expressions.
43400		If a positive attitude is expressd by the interviewer,
43500	FEAR and ANGER decrease. FEAR and ANGER increase depending on
43600	the conceptualizations of the input. These attitudes of the interviewer
43700	are stored as beliefs in the model being built up about the interviewer.
43800	Later the model can consult these beliefs in formulating questions
43900	and statements to the interviewer.
44000		Associated in the data base with each type of attitude expression
44100	expected are lists of appropriate output expressions. Thus in reply to:
44200		(18) Dr.-"I understand you."
44300	the model would reply:
44400		(19) Model- I'm glad you do."
44500	or
44600		(20) Model- "I appreciate your trying to understand."
44700	or some equivalent expression depending on values of the affect
44800	variables. When ANGER and FEAR are high, positive attitude expressions
44900	are interpreted as insincerity and hence evoke hostile replies.
45000		The remainder of input expressions not thus far discussed
45100	are handled by the procedure NORMAL.
45200	NORMAL
45300	
45400		This procedure deals with all those 
45500	interviewer expressionsfrom which no conceptualization can be formed. The only thing which
45600	can be determined is perhaps the syntactical nature of the input.
45700	Presented with one of these expressions, if FEAR is extremely high
45800	the model signs off without a farewell expression and cannot be
45900	contacted through further natural language input. If FEAR is high
46000	but not extreme, and the input is recognized as a question, the model chooses a reply from a list which brings
46100	up the attitude of the interviewer as in:
46200		(21) Model- "Why do you want to know?"
46300	or
46400		(22) Model- "You pry too much".
46500	If the input is recognized as a statement, a reply is chosen from a list
46600	which indicates soem degree of anxiety:
46700		(23) Model- "Who are you really?"
46800		(24) Model- "You are making me nervous."
46900	If ANGER is high and the input is a question, a reply is chosen from
47000	a list designed to express hostility as in:
47100		(25) Model- "Do you know what you are doing?"
47200		(26) Model- "Perhaps you are just posing as a doctor."
47300		Sometimes in these default conditions the flag set in
47400	DELUSIONALREFERENCE allows the model to continue bt giving the next
47500	line in its delusional story. If the story is under discussion, continuity
47600	is maintained. But if it is not, the model appears to ignore the input
47700	and jumps back to oen of its previous preoccupations. In this instance
47800	the property of rigidity is a function of linguistic non-comprehension
47900	and not of the paranoid processes per se.
48000		If a story flag has not been st by a previous discussion
48100	in the interview and ANGER and FEAR are not high, the algorithm
48200	tries to see if the input is some type of general prompt from
48300	the interviewer such as:
48400		(27) Dr.- "Go on."
48500	or
48600		(28) Dr.- "Tell me more."
48700	If so, the model continues with its current story or attempts to
48800	initiate another story. (Discuss this in analyzer section?)
48900		If none of these conditions hold, the procedure ANSWER
49000	is called. This procedure handles a group of common special-case
49100	miscellaneous questions such as:
49200		(29) Dr.- "How do you do?"
49300	and miscellaneous statements such as:
49400		(30) Dr.- "Hi."
49500		(31) Dr.- "Good evening."
49600	
49700	SELFSCAN
49800		The final procedure in the algorithm scans what the model 
49900	has chosen to output. That is, it treats its own output as input.
50000	If this expression contains a flare or delusional reference, the
50100	appropriate flags are set and FEAR is raise slightly, but not as
50200	much as if this expession came from the interviewer. In this way
50300	the model "frightens itself" by what it says about a frightening topic.
50400	
50500			SUMMARY