00100	REMARKS ON THE TREATMENT OF NONSPEAKING CHILDHOOD AUTISM AT
00200	THE FRIEDA FROMM-REICHMANN AWARD LUNCHEON, MAY 5,1973, HONOLULU.
00300	
00400			KENNETH MARK COLBY
00500	
00600	
00700		I never had the pleasure of  meeting  Frieda  Fromm-Reichmann
00800	but  I did have almost an opportunity which I should have seized. For
00900	a year she was at the Center For Advanced  Study  in  the  Behavioral
01000	Sciences at Stanford. But there is an unwritten rule among the locals
01100	in the Bay Area not to bother visiting scholars.I regret  I  did  not
01200	break  the rule in this case. We had published books on psychotherapy
01300	at about the same time, she as an expert and  I  as  a  neophyte.  We
01400	might  have been able to share stories about what comes your way when
01500	you write about something in which EVERYONE feels he is an expert.
01600		In   re-reading   Frieda  Fromm-Reichmann's  book  and  other
01700	publications I could not find any discussion of childhood autism  and
01800	its  treatment.  However there is ample evidence in her writings that
01900	she  was  not  reluctant  to  take  on  difficult  problems  and  use
02000	imaginative  methods.  It is because of this spirit of innovation and
02100	willingness to undertake the hitherto discouraging, that I am honored
02200	to receive an award bearing her name.
02300		Today I  would  like  to  tell  you  something  about  a  new
02400	computer-based   method   for  stimulating  language  acquisition  in
02500	nonspeaking austistic children.
02700	
02800		No one person invented  and  developed  the  method  I  shall
02900	describe.  Hence  let  me  first  acknowledge  my  indebtedness to my
03000	co-workers at Stanford University- Horace Enea, David Smith,  Malcolm
03100	Newey  and  Maxine  Colby,  each of whom has put years of effort into
03200	this project.
03300	
03400		We began about seven years ago with a working hypothesis  and
03500	two  well-known facts about childhood autism. The hypothesis was that
03600	nonspeaking child's primary difficulty lay in an inability to process
03700	symbols,  language being of course the most important symbolic system
03800	used in human communication. The  first  well  known  fact  indicated
03900	prognosis   was  highly  correlated  with  speech,  the  outlook  for
04000	nonspeakers being poor. The second fact was  the  common  observation
04100	that  these  children  played  for  hours  with machines but remained
04200	indifferent to interactions with people.
04300		There  is  now increasing evidence in the research literature
04400	supporting  this  hypothesis  of  a  primary  difficulty  in   symbol
04500	processing.  [Churchill,1972; Frith,1972; Hermelin and O'Connor,1970;
04600	Rutter, Bartak and Newman,1971]. A dysphasic or developmental aphasic
04700	child  also  has difficulty with language but he can acquire usage of
04800	other symbolic  systems  such  as  gesturing  and  drawing.  But  the
04900	nonspeaking  autistic  child  has  great difficulty with all symbolic
05000	systems, not just language.
05100		The  cause  of  this  condition remains unknown. Nowadays few
05200	experts in the field defend a strictly psychogenic etiology since the
05300	supporting evidence is weak and the disconfirming evidence is gaining
05400	in strength [Rutter et al.,1971]. Regardless of the  original  cause,
05500	if  we  believe  we understand the crux of the child's difficulty, we
05600	can try to devise a remedial treatment which takes advantage  of  the
05700	child's  fascination  with  machines.  Our  idea then was to create a
05800	machine a nonspeaking child could play with in which the play process
05900	involved the use
06000		The system we developed consists of a television-like  screen
06100	and  a  typewriter-like  keyboard in front of which the child sits or
06200	stands. There is no computer visible since it is located  in  another
06300	part of the building. Pressing a key on the keyboard causes a letter,
06400	word, phrase, or picture to appear on the screen accompanied  by  the
06500	sound  of a human voice saying the linguistic expression or something
06600	else which might interest a child. It is much like having  one's  own
06700	Sesame  Street to play with. But instead of being a passive recipient
06800	of what the television show provides, a child in our situation is  an
06900	active instigating agent who controls the machine. The merits of such
07000	a machine are that it is untiring,  predictable,  always  saying  the
07100	same  thing  the same way, never angry, never bored and controllable-
07200	properties which are notoriously lacking in humans.
07300		This   audio-visual-tactile   experience  is  directed  by  a
07400	computer program running on a PDP  6/10  time-shared  system  at  the
07500	Stanford Artificial Intelligence Project. The program is divided into
07600	games intended to provide a  child  a  variety  of  opportunities  at
07700	playing  with  and interacting with symbols. For example, in one game
07800	when the child presses the key showing the letter "H", an "H" appears
07900	on the screen and a voice says "H". In another game pressing this key
08000	produces a horse on the screen accompanied by the  sound  of  running
08100	hooves.  There  exist  over  1000 such experiences on the system. The
08200	games are organized at various levels of complexity and are  designed
08300	to  show  a child how English is put together from sounds and letters
08400	into words and expressions. The idea is that in playing he will begin
08500	to  copy  or  approximate the sounds he hears associated with what he
08600	sees on the screen. I shall not go into the details of the games here
08700	They   are   throughly   described   in   the  literature,[Colby  and
08800	Smith,1971]. Instead I shall try to  sketch  the  rationale  or  major
08900	principles underlying this approach.
09000		First, consider how normal children  acquire  language.  They
09100	are  not  taught  formally  as are adults learning a second language.
09200	Children are simply exposed to members of a linguistic community  and
09300	given   an   opportunity  to  explore  language  usage  in  everyday
09400	communication with other humans who encourage and  sometimes  correct
09500	them.  From  this exposure, exploration, and corrective feedback they
09600	come to associate the sounds and meanings  of  words.  They  come  to
09700	grasp  that  certain that certain sounds, human voice sounds, are not
09800	just sounds in themselves like wind in the trees, but that  they  are
09900	about something else, that is, they are symbolic.  Human voice sounds
10000	are perceptual invariants about other invariants.  Between  the  ages
10100	of  roughly  2  months and 4 years normal children spend thousands of
10200	hours listening, practicing and playing with language.  Mastery of  a
10300	language,  that  is getting most of it right, does not come until far
10400	into adolescence, if ever.  A normal child can make an interpretation
10500	of  expressions  he  has  never heard before.  A nonspeaking autistic
10600	child does not show this course of development.
10700	
10800	The problem is not simply the development of language skills but also
10900	the  acquisition  of  concepts  necessary  for a comprehensive enough
11000	model of how the  world  works,  especially  the  human  world.   The
11100	conceptual  or  cognitive  deficits  shown  by  nonspeaking  autistic
11200	children involve those concepts which are normally  acquired  through
11300	language  or  other  symbolizations.   Take  the  abstract concept of
11400	`danger'.  To prevent a child  from  becoming  hurt,  a  parent  must
11500	identify  certain  concrete objects and situations as dangerous until
11600	the child grasps the abstract concept of danger.  All  this  is  done
11700	linguistically and through pointing.  By way of language, objects and
11800	situations can be referred to and warned about even when they are not
11900	present,  and referred to without pointing when they are present.  An
12000	important socialization function of language is to  mark  off  for  a
12100	child  what  to pay attention to and what is to be done and not done.
12200	Many of the characteristics of autistic children,  for  example,  the
12300	lack  of  empathy,  can be viewed as a consequence of an inability to
12400	form  regular  conceptual  patterns  about  the  world  because   the
12500	necessary  concepts  acquired  through  language are absent or poorly
12600	developed. As one perceptive mother said about her autistic child, `a
12700	screw is not loose, a screw is missing'.
12800	
12900		A poorly developed ability to  process  symbols  has  further
13000	consequences  besides  conceptual deficits.  Without language a human
13100	has  no  awareness  of  being  aware.    He  lacks  the  ability   to
13200	self-monitor  and  to self-control by talking to himself.   He cannot
13300	use symbols reflexively, that is,  to  give  himself  orders  and  to
13400	comment on himself to himself.  Having language, a normal child comes
13500	to realize the self is really two which  can  talk  to  one  another.
13600	Finally  to  become  a  person, one must be recognized as a person by
13700	other  people  who  treat  you  as  if  you  had  self-awareness  and
13800	self-control.   This  recognition  is virtually impossible for a child
13900	without language skills and probably represents the greatest  tragedy
14000	in his life.
14100	
14200		Thus far I  have  spoken  of  autistic  children  as  if  the
14300	nosological label stood for a single homogeneous group.  Thirty years
14400	ago it seemed that night be the case.  Now it  appears  as  if  there
14500	exist  several  autistic  syndromes, none of which should be confused
14600	with childhood schizophrenia since  they  differ  in  onset,  course,
14700	symptoms,  family  history  and  prognosis.    There are speaking and
14800	nonspeaking autistic children.  Among the nonspeaking group there are
14900	those  whose  linguistic development is normal until some time in the
15000	second year when they lose their language abilities.  The other major
15100	group  are  those  children  who,  from  the start, have trouble with
15200	language, understanding little and  saying  even  less,  perhaps  one
15300	'mama'  or  `no'  a  year.     In  our  experience the most difficult
15400	problem  for  differential  diagnosis  lies  in  deciding  whether  a
15500	nonspeaking  child  suffers  from  dysphasia {developmental aphasia},
15600	autism or perhaps both.  Over time the correct diagnosis can be  made
15700	when  it  becomes  clear  that  the  dysphasic  child can mimic, draw
15800	pictures and signify greetings while the autistic child cannot.
15900	
16000		Taking  a  cue from the normal child who treats language as a
16100	toy, our first principle was that the  treatment  should  provide  an
16200	opportunity for exploratory play.  The treatment situation is not one
16300	of forced drill, instruction or training but one  of  play  with  the
16400	keyboard  and video display.  Operant conditioning methods reward the
16500	child with candy or food for his actions.  We do not, believing  that
16600	food  rewards  inhibit  exploratory  curiosity,  as has been shown in
16700	animals both by Harlow and by Nissen. From  our  own  experience  and
16800	that  of  workers  in  computer-aided  instruction,  food rewards are
16900	distracting and disrupting. Exploratory learning requires a situation
17000	which  invites  exploration,  and provides time, security and minimal
17100	interference by adults.
17200	
17300		In  each  of  the sessions the child has a `sitter', an adult
17400	whose main task is to sit and not interfere.  The  sitter's  task  is
17500	difficult,  especially  if he has been trained to DO things.  We want
17600	to give the child an opportunity to freely self-select those  symbols
17700	which interest him, rather than to have an adult instruct him or quiz
17800	him about those  symbols  which  he  `should'  learn.   The  sitter's
17900	behavior  is  crucial  to this treatment method if the spirit of play
18000	rather than drill is to be provided.  Of  course  the  sitter  offers
18100	social approval and encouragement when it is fitting.
18200	
18300	An  ideal treatment session results when the child is in a good mood,
18400	is interested in working the keyboard display, enjoys  imitating  the
18500	sounds  and is successful in getting the machine to do what he wants.
18600	The principle of success is important here because, in our view, many
18700	nonspeakers  have  given  up  on language.  They have failed over and
18800	over and hence withdraw from trying.  We do not let them  fail.    As
18900	one  normal  child  said  about  the  experience "It's fun, you can't
19000	lose".   There  are  students  of  autism  who  consider  nonspeaking
19100	autistic  children to be innately withdrawn from people and hence not
19200	acquiring language.  We feel it is the other way around; they have so
19300	much   difficulty  with  language  they  withdraw  from  people  who
19400	unwittingly flood and overwhelm them with  meaningless  noises.    No
19500	wonder  they  do  not call, address, ask or answer questions of these
19600	giants who immediately spout gibberish.  Autistic  children  are  not
19700	aloof  and  indifferent  to  all people -- only to those who talk.  I
19800	have often wondered if it is not eye-to-eye contact  they  avoid  but
19900	eye-to-a-mouth  which,  as  far  as  the  child  is concerned, simply
20000	jabbers.  If you say little or nothing  to  a  nonspeaking  child  on
20100	first meeting him, you will more than likely find him in your lap, as
20200	affectionate as any other child.
20300		As  with  any  treatment  method,  we  have had successes and
20400	failures.  Every  treatment  of  autism  has  reported  its  dramatic
20500	successes  with  one  or  two  children. What is needed are long case
20600	series before we can sensibly compare methods  and  decide  which  is
20700	more effective relative to the effort and cost required. Thus far our
20800	series of nonspeaking autistic children totals  17  with  13  of  the
20900	children  improved.  By  improvement  we  mean  that the child begins
21000	voluntarily to use speech for social communication. We do  not  claim
21100	the method results in normal language ability with full comprehension
21200	and correct articulation and grammar. Our aim is to kindle a  child's
21300	interest  in  using  speech,  to  get  him to try again and again, to
21400	catalyze his damaged or slow-developing natural process  of  language
21500	acquisition.
21600		Our non-random sample of 17 is too small to  arrive  at  firm
21700	conclusions  about  the  method. However our improvement rate at this
21800	point in time represents the best reported in  the  literature.  Some
21900	cautions  are  in order. Language improvement represents only a first
22000	step. These children still need a lot of help and have a long way  to
22100	go to reach whatever is their potential.
22200	
22300		Why   haven't   other   workers   tried  this  zero-risk  and
22400	potentially promising method?  Aside from the inertia which  all  new
22500	methods  must face, there are two reasons.   One is that everyone has
22600	his own preferred method  which  he  believes  in  and  may  wish  to
22700	improve.    The  other  involves  people's beliefs and misconceptions
22800	about computers. I have come across people with the following fantasy
22900	about  our  situation  --  that we have a small, cowering, bewildered
23000	child sitting in front of a towering bank of flashing lights {the New
23100	Yorker-cartoon  concept of a computer} while cold scientists in white
23200	coats observe him thru one-way mirrors.    They  say  "the  child  is
23300	already too interested in machines, he needs relations with people --
23400	you are making him into even more of a robot".   I hope from my brief
23500	description  you  can see this is a misconception of our situation.It
23600	may sound paradoxical, but some nonspeaking children can become  more
23700	human,  i.e.  become  language users, by way of a machine which talks
23800	and which they find more acceptable on their own terms than  they  do
23900	talking humans.
24000	
24100		A  more  realistic objection to computers is their cost.  Few
24200	people in the world have access to million dollar computers for  this
24300	sort  of  work.   But  a large computer is not necessary to carry out
24400	this method.   Mini-computers costing only a few thousand dollars are
24500	adequate.  It may be that we do not need a computer at all if someone
24600	found a way to rapidly random-access both sounds and  pictures.   We
24700	are investigating this possibility. Once we can get the unnecessarily
24800	Satanic image of the computer out of the picture , I hope others will
24900	join  us  in  discovering  more  about  this  technique  for treating
25000	nonspeaking autistic children.
25100	
25200			REFERENCES
25300	
25400	
25500	[1] Churchill,D.W.,1972. The relation of infantile autism and early
25600	       childhood schizophrenia to developmental language disorders
25700	       of childhood. Journal of Autism and Childhood Schizophrenia,
25800	       2,182-197.
25900	[2] Colby, K.M. and Smith, D.C., 1971. Computers in the treatment of
26000	       nonspeaking autistic children. In Current Psychiatric Therapies,
26100	       Masserman, J.H.(Ed.), Grune and Stratton, New York.
26200	[3] Fay, W.H., 1969. On normal and autistic pronouns. Journal of Speech
26300	       and Hearing,36,242-249.
26400	
26500	[4] Frith, U., 1972. Cognitive mechanisms in autism: experiments with
26600	        color and tone sequence production. Journal of Autism and
26700	        Childhood Schizophrenia, 2, 160-173.
26800	[5] Hermelin, B. and O'Connor, N., 1970. Psychological Experiments
26900	        With Autistic Children, Pergamon Press, London.
27000	[6] Rutter, M., Bartak, L. and Newman, S., 1971. Autism- a central
27100	        disorder of cognition and language? In Infantile Autism,
27200	        Rutter, M.(Ed.), Churchill Livingstone, London.