JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1973, 63, 131-166 NUMBER 1 (SPRING 1973) SOME GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN IN BEHAVIOR THERAPY1 0. IVAR LovAAS, ROBERT KOEGEL2, JAMES Q. SIMMONS, AND JUDITH STEVENS LONG3 UNIVERSITY OF CALIFORNIA, LOS ANGELES We have treated 20 autistic children with behavior therapy. At intake, most of the chil- dren were severely disturbed, having symptoms indicating an extremely poor prognosis. The children were treated in separate groups, and some were treated more than once, al- lowing for within- and between-subject replications of treatment effects. We have em- ployed reliable measures of generalization across situations and behaviors as well as across time (follow-up). The findings can be summarized as follows: (1) Inappropriate behav- iors (self-stimulation and echolalia) decreased during treatment, and appropriate behav- iors (appropriate speech, appropriate play, and social non-verbal behaviors) increased. (2) Spontaneous social interactions and the spontaneous use of language occurred about eight months into treatment for some of the children. (3) IQs and social quotients re- flected improvement during treatment. (4) There were no exceptions to the improvement, however, some of the children improved more than others. (5) Follow-up measures re- corded 1 to 4 yr after treatment showed that large differences between groups of children depended upon the post-treatment environment (those groups whose parents were trained to carry out behavior therapy continued to improve, while children who were institu- tionalized regressed). (6) A brief reinstatement of behavior therapy could temporarily re-establish some of the original therapeutic gains made by the children who were subsequently institutionalized.
The first succinct attempt to understand the primary contribution of Ferster's theoretical behavior of autistic children within a behavior- argument lies in the explicitness and concrete- istic framework was carried out by Ferster ness in which he relates learning principles to (1961) Ferster presented a very convincing argu- behavioral development. Shortly after he pre- ment of how it was that, based on a general sented his theoretical notions about autism, deficiency in acquired reinforcers, one might Ferster and DeMyer (1962) reported a set of expect the very impoverished behavioral de- studies in which they exposed autistic children velopment one sees in autistic children. The to very simplified but controlled environments 'We express our thanks to the parents who en- where they could engage in simple behaviors, trusted their children to us, and for the help and en- such as pulling levers or matching to sample for couragement they have given. The research has been reinforcers that were significant or functional to supported by PHS Research Grant No. 11440 from the National Institute of Mental Health. Many per- them. The Ferster and DeMyer studies were the sons have helped in this research; in particular, we first studies to show that the behavior of autistic are grateful for the help that Gail Abarbanell, M.S.W., children could be related in a lawful manner to Lorraine Freitas, M. S., Meredith Gibbs, Laura Schreibman, Ph.D., Joan Meisel, Ph.D., and Linda certain explicit environmental changes. What Silverman gave in directing and managing the the children learned in these studies was not of Clinic, U.C.L.A. Monographs of this article are avail- much practical significance, but the studies did able for $1.00 from the Business Office of the show that by carefully arranging certain environ- Journal of Applied Behavior Analysis, Department of Human Development, University of Kansas, Lawrence, mental consequences, these children could in Kansas 66044. Ask for Monograph #2. fact be taught to comply with certain aspects 2Now at the Institute for Applied Behavioral Sci- of reality. ence, University of California, Santa Barbara. 3Now at the Department of Psychology, California The first systematic attempt to use behavior State College, Los Angeles. modification procedures on more general, so- 131 132 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG cially practical behaviors of an autistic child was marked improvement has taken place, it has reported by Wolf, Risley, and Mees (1964). generally become evident before the age of 6 They worked with a 3.5-yr-old boy who did not or 7 yr. From middle childhood on, the course eat normally, lacked normal social and verbal has been fairly regular, with a continuation of repertoires, and evidenced extreme tantrums and improvement or deterioration evident by then. self-destructive behaviors, often leaving himself (3) In almost all cases, there were declines in IQ. bruised and bleeding. By systematically control- (4) Improvement was unrelated to whether or ling the child's environment, these investigators not a child had received therapy. When im- were eventually able to bring the child's re- provement has taken place, it has been described sponding toward a more normal level of func- as "spontaneous", that is, independent of a pro- tioning. Tantrum behavior was treated by a fessional prescribed treatment. Havelkova combination of mild punishment and extinction. (1968) reviewed several other recent studies. They also reported on certain training proce- The results have been consistent with those dures that helped the child to communicate reviewed above. more effectively verbally. At about this time, In contrast to these very pessimistic observa- several other studies appeared where psycholo- tions, the early studies that used behavior therapy gists reported success in helping autistic chil- were quite optimistic. But since this form of dren acquire certain basic and important reper- intervention is quite new, it remains to be shown toires, particularly in the area of imitation and how effective it really is with autistic children. language (Hewitt, 1965; Metz, 1965; Lovaas, The design of the early studies left many ques- et al., 1966a). tions unanswered. Most of the studies reported These behavioristic attempts to treat autistic work on single subjects, which beg the question children carried with them a promise of help of generality across children. Little if any sys- and a certain optimism for the autistic child. tematic data were presented on the extent to This contrasted with the general hopelessness which the treatment effects generalized across that had grown out of the failure that the psy- enviornments, neither were data reported on chodynamic therapies had encountered in trying response generalization. Except for the follow-up to help these children. Kanner, who was the first data on one child (Wolf, Risley, Johnston, person to describe and label these children as Harris, and Allen, 1967), there are no data that "autistic", also reported on the failure of psycho- allow one to assess how well the behavioral dynamic therapies to effect change (Kanner and intervention held up over time. Eisenberg, 1955). Brown's 1960 study supported The primary purpose of the present paper is Kanner's data that the children were unaffected to present some measures of generalization and by psychotherapy. Later, Rutter (1966) provided follow-up data on 20 children that we have a comprehensive review of investigations dealing treated with behavior therapy during the last with sizable groups of autistic children. The re- 7 yr. We hope to provide the reader with an sults of the studies that Rutter reviewed are quite approximation of changes one might expect to consistent with one another and are quite pessi- see in autistic children undergoing behavior mistic regarding prognosis. They may be sum- therapy. However, it is also our belief that the marized as follows: (1) Of those children who results presented here probably underestimate originally had IQ scores below 50, almost none the benefits of such therapy for autistic chil- acquired speech nor received any schooling, dren because the results were influenced by our and three-fourths were in long-term hospitals at extensive efforts at measurement and replication follow-up. If the child was mute and had no as well as therapy. appropriate play before the age of five, the We will try to evaluate the treatment effects prognosis was particularly bad. (2) When along three dimensions: (1) stimulus generaliza- GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 133 tion, the extent to which behavior changes that land Diagnostic Checklist for Behavior-Dis- occurred in the treatment environment trans- turbed Children, Rimland, 1964), most of the ferred to situations outside that situation; (2) re- parents report that their children (a) at one sponse generalization, the extent to which time appeared to be deaf; and (b) seemed to look changes in a limited set of behaviors effected through or walk through things as if they were changes in a larger range of behaviors; and (3) not there. Furthermore, many of the parents durability or follow-up, how well the thera- indicated that at one time they sought profes- peutic effects maintained themselves over time sional opinion about their children's hearing (Baer, et al., 1968). and/or vision, only to be told that the child had "normal" hearing and vision. (2) Severe METHOD affect isolation was a predominant feature. This means that parents indicated on the Rimland Subjects Checklist that their children (a) fail to reach out We have treated a total of 20 children, all of to be picked up when approached by people; (b) whom have been diagnosed as autistic by at look at or "walk through" people as if they least one other agency not associated with this were not there; (c) appear so distant that no one project. The majority of the children had been can reach them; (d) are indifferent to being given more than one label, usually also being liked; and (e) are not affectionate. (3) Our referred to as retarded and brain damaged. Our sample showed a high incidence of self-stimula- experience and that of others (cf., Rutter, ibid.) tory behavior, that is, behavior that appears suggests that there is considerable behavioral solely to provide the children with propriocep- heterogeneity among autistic children. There- tive feedback (e.g., rocking, spinning, twirling, fore, it may be appropriate to describe the chil- flapping, gazing, etc.). A more detailed descrip- dren we have treated in more detail. First, we tion of this type of behavior is given below in the have treated the very undeveloped children, method section (under instructions for observer that is, children who would fall within the identification). (4) Mutism occurred in about lower half of the psychotic continuum, and half of the children in our sample. These chil- whose chances of improvement were considered dren produced no recognizable words (their to be essentially zero. Most of the children had sounds consisted primarily of vowels). (5) at least one prior treatment experience (up to Echolalic speech was present in the remaining 4 yr of intensive, psychodynamically-based children. These children echoed the speech of treatment) which had not effected any noticeable others, either immediately or after a delay, often improvement. Most of the children have been giving the impression of non-related inappropri- rejected from one or more schools for the ate speech (a more complete description of these emotionally ill or retarded because their teachers behaviors is also given below in the instructions could not control them, in addition to which for observer identification). (6) In all children their behavior was often so bizarre that it was receptive speech was minimal or missing en- disruptive for the other children in the class. tirely. Some of the children would obey simple Clinically speaking, with three or four excep- commands (such as "sit down", or "close the tions, they seemed void of anxiety, and none door"), but all failed to respond appropriately to had any awareness that he was considered ab- more complex demands involving abstract terms normal. such as prepositions, pronouns, and time. Most Generally, the children we have treated can be often they responded to speech in a very gen- described along the following dimensions: (1) eralized manner. For example, they would close Apparent sensory deficit, indicating that when the door when they heard the command, "Close asked to complete the Rimland Checklist (Rim- the door.", as well as when they heard com- 134 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG mands like "Point to the door.", or statements produced some durable reinforcers, they were too such as "There is a window and a door." etc. discriminated (situational) and the procedures (7) There was also an absence of, or only mini- too cumbersome to be of much practical sig- mal presence of, social and self-help behaviors. nificance. For instance, most of the children could not dress We turned, therefore, to the second alterna- themselves; most were unaware of common tive; building behaviors directly relying on al- dangers (e.g., crossing the street in front of on- ready effective, largely primary reinforcers such coming cars); most could not wash themselves as food, essentially circumventing social stimuli. or comb their hair; some were not toilet trained, The use of primary reinforcement has several etc. (8) A small number of these children were disadvantages, as compared to social, secondary self-destructive or self-mutilatory. All displayed ones. For example, in using primary reinforcers, severe aggressive, tantrumous outbursts, scratch- special environments need to be established to ing and biting attending adults when forced develop and maintain the new behaviors. Since to comply with even minimal rules for social we have inadequate information about how to conduct. Some smeared their feces. construct such environments, the gains that the child may make would probably fall short of the Treatment ideal. Despite these restrictions, however, it is When one decides to treat a child within a worthwhile to assess how much one can accom- reinforcement theory paradigm, then one can plish using a limited range of reinforcers. There- facilitate the behavioral development of autistic fore, we describe the program we did develop. children in two ways. One way would be to con- Because the children were replete with inter- centrate efforts on facilitating the autistic child's fering self-stimulatory, self-destructive and/or acquisition of social reinforcers, rather than on tantrum behavior when they entered treatment, building behaviors. If his developmental failure we immediately attempted to reduce the fre- was based on a deficiency in social and other quency of such behavior. The procedures em- secondary reinforcers, as Ferster claimed it was, ployed to extinguish and suppress pathological then an intervention at this level would seem behavior (including biting and scratching of self to strike at the base of the problem. A treatment and others, feces smearing, etc.) rely heavily program centered on the establishment of a on several operations: (1) contingent reinforce- normal hierarchy of social reinforcers would ment withdrawal, that is, the adult simply give the child's everyday social environment looked away from the child when he was en- (his parents, teachers, peers, etc.) the tools with gaged in undesirable behavior, left the child in which to build and modify the myriad behaviors his room, or placed the child in an isolation necessary for the child to function effectively room (separate from the treatment room); (2) within that environment. In a sense, the person's contingent aversive stimulation, for example, behavioral changes would "take care of them- a slap or painful electric shock; or (3) reinforce- selves", provided that he returned from treat- ment of incompatible behavior, such as sitting ment to a normal environment with a normal quietly on a chair. The rationale for the sup- reinforcement hierarchy. pression of self-stimulatory behavior lies in the When we first began to treat autistic children, observations we have made indicating an ap- we explored this alternative of enriching and parent attenuation of the child's responsivity normalizing reinforcing stimuli for these chil- while he is engaged in self-stimulation (Lovaas, dren. We did succeed at establishing certain Litrownik, and Mann, 1971). Simply stated, social stimuli as reinforcing, using either pain when the child is engaged in self-stimulation, reduction (Lovaas, et al., 1965b) or food pre- it is difficult to teach him something else. The sentations (Lovaas, et al., 1966b). Although we reasons for suppressing self-destruction, feces GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 135 smearing, etc., are perhaps obvious, and our himself. (4) After the child reliably emitted one intervention model does not prescribe the sound, the therapist introduced a second sound therapeutic benefits of their expression. A de- (such as the consonant "mm") and reinforced tailed presentation of data and method for sup- reproductions of that sound. These first two pression of self-destruction may be found in sounds were then presented in a random order Lovaas and Simmons (1969). so that the child was required to discriminate Simultaneously with the suppression of un- between the two vocalizations. (5) A third sound desirable behavior, the therapist attempted to was presented, requiring increasingly fine dis- establish a kind of primitive stimulus control. criminations. In such a manner we attempted to Usually, the therapist demanded some simple build imitative behavior, which we conceived of behavior from the child, such as looking at the as a discrimination where the child's response therapist, or sitting down when the therapist resembled its stimulus (the adult's response). asked. These behaviors could be easily prompted If the child was echolalic (or once a mute if the child did not already know how to respond. child had about 10 imitative words), we intro- Usually, the therapist's first attempts to estab- duced a program designed to make speech mean- lish stimulus control elicited tantrumous and ingful and functional. For example, as soon as self-destructive behavior; therefore, we com- a child was taught the label for a particular bined the suppression of undesirable behavior food, he could eat only if he asked for the food with the attempt to establish stimulus control. by name. The child was gradually moved Once these introductory steps had been taken, through a series of steps designed to establish we introduced our central training program in increasingly proficient use of language, includ- which language training alone consumed about ing training in semantics, such as use of abstract 80% of the child's total training. The heavy terms (pronouns, time, etc.), and syntax, such as emphasis on language training was undertaken the correct use of tense, etc. Some of the later partly for academic reasons. We wanted to know levels were never reached by the mute children, how much could be accomplished using operant but were usually obtained with the echolalics. A procedures. This was not necessarily the most more detailed description of the language pro- beneficial therapeutic approach for all the chil- gram exists on film (Lovaas, 1969) and in writ- dren. Many of them have benefited more from ten outline (Lovaas, in preparation). a program emphasizing non-verbal communica- At the same time we were involved in build- tion. ing speech, we also initiated programs designed If the child was mute, we began a verbal to facilitate the acquisition of other social and imitation program to facilitate his phonetic de- self-help skills. These programs focused on those velopment (Lovaas, et al., 1966a). Briefly, verbal behaviors that made the child easier to live with, imitation was established in five steps: (1) The such as friendly greetings and other indications child received reinforcement for vocalizing in of affection, as well as dressing, good table order to increase the frequency of speech sounds. manners, brushing the teeth, etc. We have out- (2) We then established a temporal discrimina- lined a procedure based on non-verbal imitation tion. The child received reinforcement only for (Lovaas, et al., 1967) that has been particularly those vocalizations that were emitted within a useful for these purposes. 5-sec period after the therapist made a vocaliza- Throughout, there was an emphasis on mak- tion. (3) The therapist now began to demand ing the child look as normal as possible, reward- similarity of vocalizations between himself and ing him for normal behavior and punishing his the child. For example, the therapist gave rein- psychotic behavior, teaching him to please his forcement for a sound (for example, "ah") only parents and us, to be grateful for what we would after the therapist had first emitted that sound do for him, to be afraid of us when we were 136 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG angry, and pleased when we were happy. Adults Multiple-response recordings. We have pre- were in control. In short, we attempted to teach viously published (Lovaas, et al., 1965a) infor- these children what parents of the middle-class mation on apparatus that allows for simulta- Western world attempt to teach theirs. There neous recordings of several commonly occurring are, of course, many questions that one may have and everyday behaviors in free-play/observation about these values, but faced with primitive settings. Essentially, certain behaviors (both psychotic children, these seem rather secure and normal and pathological) are defined for an comforting as initial goals. observer who records their frequency and dura- We selected reinforcers on the basis of their tion on a button-panel, which in turn is coupled value for a particular child. Many children to a computer tape, allowing swift calculation would work only for food and required an oc- of the frequency, duration, and interaction of the casional slap on the buttocks if the therapist was various behaviors. to control undesirable interfering behavior. For The kind of child one is studying helps decide other children, symbolic approval and disap- what kinds of behaviors to record. In the case of proval were effective in maintaining the child's severely psychotic children, this is somewhat behavior throughout the working sessions. As we simplified because of their limited behavioral became familiar with the idiosyncracies of the repertoires. We eventually selected five be- various children, the reinforcers seemed easily havioral categories. The presence or absence of accessible and their selection was fairly simple, behaviors in these categories are used to describe despite their limited range. However, scheduling autistic children, and we have found they can these reinforcers was a much more difficult task. reliably be recorded. (1) Self-stimulation, which A relatively untrained person can build simple denotes the stereotyped repetitive behavior that behaviors, like eye-to-face contact or raise the appeared only to provide the child with pro- frequency of vocal behavior. But it is unlikely prioceptive feedback (e.g., rocking, spinning, that a person will be able to build complex twirling, flapping, gazing, etc.). (2) Echolalic speech unless he is familiar with discrimination speech, which was defined as the child's echoing learning procedures. Most people who work the speech of others, either immediately or after with autistic children are not. Therefore, it seems a delay, giving the impression of non-related likely that there will be few studies in the near inappropriate speech, with pronoun reversal, future to replicate the present one. incorrect use of tense, etc. We also included bizarre words and word combinations in this Measurement category. (3) Appropriate speech, which was We have employed two measures of general- defined as speech related to an appropriate con- ization of change during treatment. First, we text, understandable, and grammatically correct. have attempted to assess changes in the chil- (4) Social non-verbal behavior, which denoted dren's behavior using a multiple-response record- appropriate non-verbal behavior that is depen- ing. Secondly, we have assessed changes in the dent upon cues given by another person for its children's Stanford Binet and Vineland Social initiation or completion (e.g., responding to re- Maturity scores. The multiple-response record- quests, imitating, etc.). (5) Appropriate play, ings constitute the main focus of our measures which denoted the use of toys and objects in an and were designed to provide information both appropriate age-related manner. on stimulus and response generalization. The Two of these behaviors (self-stimulation and Stanford Binet and Vineland do provide similar echolalia) are pathological. Their. presence, and measures, but they give less specific information. the relative absence of the remaining three We shall first present a description of the mul- "normal" behaviors, forms part of the behavioral tiple-response recordings. complex diagnostic of autism. The instructions GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 137 for recording, and hence rather complete defini- All of these behaviors have the common ap- tions of these various behaviors are pearance of producing sensory input for no given below. other purpose than the stimulation itself. There is, however, a fine line between some of these Instructions for rater identification. You will be behaviors and some of the more primitive forms watching for five kinds of behaviors. These will of play. You will have to use your judgment be the only behaviors you will have to record, to some extent to decide if the child is looking so part of the time you may not be pressing a at something or staring at it, playing with it or button at all. If you are uncertain about what is tapping it. Often, you may be able to tell self- going on, you may also not be recording. The stimulation from appropriate play, by the fact best rule is, if you can't make a decision, don't that the child may begin to repeat a gesture or record anything. Each of the behaviors will be an act over and over again during self-stimula- carefully defined and you will be given ex- tion. amples of what they are and what they are not. Each key on the panel is labeled with the name Self-stimulatory behavior occurs to some ex- of one of the behaviors. Each time you notice tent in the repertories of all children, and even the child engage in one of these behaviors, press to some extent in adults. You may find yourself down the corresponding key, and hold it down recording something you feel looks like any until the child has terminated that behavior. child might do. Record it anyway. The differ- ence is that you may see more of it than you 1. Self-stimulation. The best way to describe might in a normal child. the various forms this kind of behavior may take is to begin with the head. The child may These behaviors may or may not be accompa- roll his eyes, cross them, look out of the extreme nied by sounds or words the child may use. corners of them or squint them, contracting the This makes no difference as the sounds or the muscles of the face all the way to the ears. He speech will be treated separately. may stare intensely at lights, objects, or at parts 2. Echolalic speech. Recording this behavior of his own body (such as his hands). He may signifies that the child is using words in an un- suck his tongue and lips or stick his tongue meaningful or inappropriate manner. This type out repeatedly. He may put objects in his of behavior may present itself in one of sev- mouth. He may rock his whole head from side eral ways. to side or allow it to fall forward, turning it slightly to the side with his eyes turned up or First, the child may appear to be repeating a to the corners. He may cock his head and hold word or several words to himself. The technical a particular position for long periods. name for this type of speech is delayed echol- alia. He may say things that sound like com- There are several typical forms of self-stimula- mands or statements he once heard, but which tion performed with the arms and shoulders. have nothing to do with his present activity, or The child may move the arms up and down the the context in which he is operating. He may sides of his body, flipping his hands from the use phrases like some of the following: "Hello wrist. He may flap his arms from the shoulder John.", "No, John.", or "How are you, John?". with his hands limp. He may hold his hands in He may go to the door and say "You want to go very contorted positions, often staring at them out?". Although this last statement does have intensely. relevance to the situation, such phrases will also Using his whole torso, the child may assume be included when they sound like the imitation rigid or contorted postures, or he may engage in of what another person has said to the child at body rocking. Rocking usually occurs in some some other time. He may also simply repeat iso- sitting position and is a forward and back mo- lated words such as "balloon, balloon". tion more often than side to side. He may twirl Secondly, the child may use bizarre speech, himself, rub, scratch, or tickle various parts of sounding like a word salad. The words may be his body, or he may masturbate. He may jump understandable separately, but do not make any repeatedly or run from wall to wall. He may sense when used together. Examples are, "green walk on his toes. rabbit", "Bufferin, one, two, three", or "happy In his interaction with physical objects, he may puppy baby little". You may be able to think of trace them, running his fingers along the edges, it as sound input for the sake of the input, much rub them, spin them, use them to make tapping like a verbal version of self-stimulation. noises or flip them back and forth. For example, Finally, another class of behavior included in he may spin a cup like a top, or he may hold this category is immediate echolalia. For ex- a piece of string and rhythmically swing it to ample, the adult may say "How are you, John?" and fro. and the child will answer "How are you, John?". 138 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG
The adult may say, "Now it's time to draw." important element is that the child would not and the child will say, "draw". be making the response if the adult were not Again, the key is depressed for the duration of there. the child's speech. You should not release the In the higher level of social nonverbal behavior, key after every word or phrase, but should re- the interaction demands a variety and flexibility lease it if the child pauses or the adult speaks. of response from both people. There is a longer You will not have to record "babbling". Also, interchange, in which the people must make do not record humming, grunting, squealing, or several different responses to complete the inter- any sound which is unintelligible (including action. The game of "Simon Says" is a good ex- fussing or crying). ample of this kind of interaction. The child 3. Appropriate speech. This behavior consists of must watch, listen, and mimic or not mimic the intelligible, non-repetitive speech which is ap- adult depending on what the adult does. Games propriate to the situation. If the child answers of pretending, playing ball, imitating drawing, a question, makes a comment, labels an object, follow the leader, and tag are also examples. carries on a conversation, or reads aloud, you Each person must watch and respond correctly should record the behavior as appropriate and complete the game. Again, two people must speech. Depress the key when the child begins be present for this type of behavior to take to speak, and release it when he finishes. place. This category includes everything from the Because of the instructions given to the adult most primitive use of words, such as a simple with the child, in almost all cases the child will "hi" or "go" (when the child is trying to leave), have to initiate this kind of interaction (the to the most complex conversation. The impor- major exception being ball play). Therefore, tant element is that the child is using words you may see a considerable amount of appropri- and that he is ate verbal behavior simultaneously taking place. correctly, meaningfully, readily Remember to keep the key depressed through- understandable. out the entire interaction. 4. Social non-verbal behavior. There are two levels of this type of behavior. Level one de- 5. Appropriate play. There are also two levels scribes certain kinds of interactions the child of appropriate play. The lower level is defined may display with the adult present. Included in as exploration and simple play. It means that this category are the simplest kinds of social the child is exploring or manipulating objects relationships. Each party need only respond and that he shows interest but inability to use once. Thus, if the child makes a response, and them properly, or that he has a lack of experi- the adult responds by completing the interac- ence in their use. He may be using them for tion, this is one response. No further response play (rather than self-stimulation, for instance) is necessary. There are no chains of response. but is not, perhaps, using them as they were in- Examples consist of two types. tended. He may be stacking tiles or blocks, scattering A. Demand behavior. The child grabs the things, putting crayons in boxes, handling and adult's hand and tugs him toward the examining various toys, pouring water into door. containers, etc. Here you must make the distinc- B. Compliance. In this case, the child simply tion between handling or examining and star- complies to some request from the adult. ing or using objects in a bizarre, repetitive, or The adult may say, "Sit down.", "Play stereotyped manner. ball.", "Put the block like this.", and the He may pile up objects, fit tiles in a peg board, child does. You should briefly depress the punch a bobo, ring the telephone, scribble with button (for less than one second) when the crayons, pull the wagon, turn pages in a the child responds appropriately to a book, or make a rattle by placing small objects request. Also included in this category is in a larger one. One common element here is simple imitation when it is not part of a that one response accomplishes as much as any game. The adult may say, "Jump, John." series of responses to a given object. One re- and then the adult may jump. If the child sponse does not require another one, nor does it imitates the jumping, you should press depend on a previous one. Stacking up one the button. block does not require another be stacked to All of these behaviors may or may not be ac- complete the stacking. companied by language. It makes no difference The higher level of appropriate play consists of as long as the child is involved in some simple the complex and appropriate use of objects, or nonverbal response which depends upon the participation in games in which there is a defi- presence of the other person. Remember, the nite dependency of one response on another. GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 139
One response leads to or proceeds from another bowling pins, a plastic telephone, a magnetic in the accomplishment of some project. In this board with numbers and letters that attach to it, category, a number of responses completes some whole which no response individually could 12 assorted wooden blocks, a 6-in. tom-tom complete. Examples include making a pattern or drum, a hand puppet, and three simple wooden picture with tiles or crayons, building an object jigsaw puzzles. The child was observed in this with blocks, reading, pulling the wagon to trans- port objects for a project, setting bowling pins room during sessions lasting 35 min. These up in the appropriate pattern and knocking sessions were divided into three conditions of them over, and completing a puzzle. Each re- 10, 10, and 15 min each. In the first condition sponse here adds something new to the ultimate goal of some project. The games listed under (the Alone condition), the child was observed by social nonverbal behavior have this same qual- himself in the playroom. In the second condition ity, interdependency of responses, and they (the Attending condition), an unfamiliar adult should also be recorded (simultaneously). Note. was present and attended visually to the child, There are several behaviors which may best be recorded by pressing and releasing a key im- but made no comment, interfered in no way, and mediately (a blip). This should be done in the did not initiate any interaction with the child. If, case of social nonverbal behavior when the child on the other hand, the child initiated some obeys a command, or each time the child catches or throws a ball. It is not done each time the child activity that required the involvement of the stacks a block or fits a tile in appropriate play. adult, the adult performed those responses and Here you must use your own judgment. Do made whatever comments necessary to complete not record during pauses, but do not record a pause between every response. Are there any the interaction. In the final condition (the Invit- questions? ing condition), the adult encouraged the child to participate in several different kinds of activi- The reader may note that social non-verbal ties. The adult invited the child to play with and appropriate play have been divided into two each of the 11 toys in the playroom in succession levels each in these instructions. This was done (1 min per toy), giving demonstrations of how in an attempt to increase the discriminating to use the toy if the child appeared not to know power of these measures, and reflect a later de- how. The adult also attempted to initiate a velopment, not present in the recordings that simple game of "patticake" for 1 min. He also we present in this paper. gave the child a 1-min series of simple com- The multiple-response measures do to some mands that could be performed non-verbally, extent assess response generalization. That is, such as "Stand on one foot.", "Touch the floor.", many of the behaviors we did score (particularly and "Sit down.". Next, the adult asked a 1-min social non-verbal and play) were not specifically series of questions which could be answered taught during treatment. But we had no way of either verbally or non-verbally. This series con- knowing exactly how much of these behaviors sisted of questions such as "Where is your nose?" were new and novel by the child, so that the or "Which block is bigger?". A final 1-min series recordings are not pure measures of response of questions, which could only be answered generalization. The measures do, however, lend verbally, was also asked. This series consisted of themselves well to studies on stimulus general- questions such as "How are you?" or "Where ization. do you live?". To assess stimulus generalization, the chil- We have multiple-response measures on only dren were observed in a room separate from, 13 of the 20 children we have treated. This is so and not associated with, the training situation because we initially had considered these mea- and in the company of an unfamiliar adult. The sures to be inappropriate for outpatients, since room was equipped, like most playrooms, with we had less control over their treatment. Since the following toys: a wagon, paper and crayons, 1968, however, we have obtained multiple- a bobo doll, a 9-in. rubber ball, three plastic response measures on the outpatients as well. 140 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG The first four children (Ricky, Pam, Billy, and The third and fourth groups were outpatients, Chuck) for whom we have multiple-response and while we initiated training programs in the recordings received a "before" measure (in clinic, we otherwise served essentially as con- June, 1964) and recordings were then made on sultants (2 to 3 hr a week) to the parents, train- a monthly basis for the 14-month duration of ing them in shaping procedures. their treatment. Pam and Ricky were discharged Discharge procedures for these children differ immediately to a local state hospital, while Billy for each individual case, depending on the rate and Chuck spent a short time (less than 6 of progress by the child, the skill of the mother months) with their families before being hos- as a therapist, and the prospects for enrolling pitalized in the same state hospital. Pam and the child in a special school. In general, our Rick were returned to us for follow-up measures approach was gradually to phase the children 2 yr later (1968). They were then briefly treated out of the program. We decreased the number once more (24 hr for Ricky and one month for of sessions from three per week to once a month. Pam), discharged to the state hospital again, and After the child was officially discharged, a finally returned for a second follow-up 2 yr after therapist visited the home several times during that (1970). Pam and Rick received our treat- the first few months. Generally, by this time the ment twice, interspersed by a period of no be- parents had found a school placement for the havior therapy treatment; Billy and Chuck were child and our involvement became minimal. treated once, but measured again 4 yr after dis- Parents were encouraged to call us when they charge from our project (1970); they received encountered difficulties, and we spoke to them an ABA design. from time to time informally discussing the We replicated essentials of the treatment on child's progress. Often, the therapist visited the a second group of children (Jose, Michael, and school and discussed the child's case with the Taylor) who were hospitalized in 1965 and teacher, suggesting ways he might find effective received 12 months of treatment, with multiple in dealing with the child and encouraging the response measures before treatment and at three- teacher to call on us if he encountered any month intervals during treatment. They were difficulties. returned for follow-up measures 3 yr after treat- The basic rationale for changing the treat- ment (in 1970). ment procedure from treating inpatients, with The third group (Leslie, Tito, and Seth) to the parents as observers, to treating outpatients receive the multiple-response recordings were with the parents as therapists, became apparent seen as outpatients. They were measured before from examination of the follow-up data. treatment (1968) and after 1 yr of treatment, Intelligence and social maturity. The Stanford and received follow-up measurements 1 yr later Binet Intelligence Scale was administered before (1970). and after treatment either by an agency not A fourth group (Kevin F., Ann, and James) associated with UCLA, or when this was not to receive multiple-response recordings was also feasible, by a graduate student trainee in the seen as outpatients. Measures were taken before UCLA Psychology Clinic. Nineteen of the 20 treatment (1969), after 1 yr of treatment children received IQ testing. One child, Taylor, (1970), and with follow-up measures in 1972. received the Merrill-Palmer Intelligence Test in- The first and second groups of children were stead of the Stanford Binet. We will also present inpatients. They received 8 hr of treatment per some data from the Vineland Social Maturity day, six to seven days a week. The parents of the Scale, which was administered to the parents of first group were not involved in the treatment. the last 14 of the 20 children. The irregularities With the second group, however, we began to in the number of children who received the train the parents in our treatment procedures. various tests does not reflect a systematic bias. GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 141 Rather, in the early phases of the program we was obtained by calculating the duration of a did not consider generalization and follow-up behavior, to the nearest second, and dividing it data to be significant data for our study. by the duration of that condition (e.g., if the subject spent 200 sec in self-stimulatory be- RESULTS havior during the 10-min Alone condition, he Multiple-response measures. Since the mul- would receive a measure of 33% self-stimulation tiple response measures are the focus of this at that time). study, they are presented first. The results are The first data, presented in Figure 1, give the presented as group averages, followed by dis- before and after treatment scores for the various cussions of changes in the individual groups and behaviors, averaged over all conditions for the children. All the figures based on the multiple four groups. The various behaviors are pre- response measures have per cent occurrence of sented on the abscissa before (B) and after (A) the behavior on the ordinate. This percentage treatment. Three groups are presented: T (total 50 ©E
40_ I /T IMI w 0 z 30 _ (jute I w
c-) T: I 0 Total I I.-- 20 _ z w IIo' cr- w a.
ETM 10T
BABefore B A B A B A B A B A Self . Appropriate Social Appropriate Stimulation Echolalia Verbal Nonverbal Play Fig. 1. Before (B) and After (A) multiple-response measures averaged over all conditions for the four groups. Per cent occurrence of each behavior is plotted on the ordinate. "E" refers to the average results for the echolalic children, "M" to the average results for the mute children, and "T" to the average results for the total group. 142 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG subjects); and the breakdown of that group into Group 1 (Rick, Pam, Billy, and Chuck) was the children who were echolalic (E) and mute measured on a monthly basis, enabling us to (M) before treatment. Looking first at the data assess the rate at which the behavior changed. from the total group, it is apparent that the in- The data for these children are presented in appropriate behaviors decreased while the ap- Figure 2. Pam and Rick (both echolalic) are propriate behaviors increased. Specifically, self- presented on the left side. Billy and Chuck stimulatory behavior was reduced to about one- (mutes) are presented on the right. The top part half of its pre-treatment level. The amount of of the figure shows changes in verbal behavior, echolalic speech decreased only slightly when one while the bottom part shows the non-verbal be- considers the total group, but this is because the haviors. For Rick and Pam, one can observe the decrease in echolalic speech by the echolalic chil- gradual increase in appropriate speech. No trend dren was offset by the increase of echolalia is obvious for echolalic behavior. Billy and in the mute children. Chuck, who were initially mute, showed a rise Turning to the appropriate behaviors, the in echolalic speech before it was replaced by children showed about four times as much ap- appropriate language. Neither had appropriate propriate verbal and social non-verbal behavior speech before treatment; each had some ap- after treatment, and almost twice as much propriate speech afterwards. Inspection of appropriate play. There were no exceptions to changes in non-verbal behaviors shows a decrease these changes; all the children improved. in psychotic self-stimulation, and increases in The total group comprising Figure 1 consisted appropriate play and social non-verbal behavior. of five mute and eight echolalic children. If we It is probably helpful to break the data down examine the data on the mute children we can by conditions to demonstrate the degree to which observe that, in addition to evidencing no speech, the adult gained control over the child, and the they showed more self-stimulation and less ap- extent to which the child initiated behavior in- propriate play. The mute children, in general, dependent of the adult's explicit direction. The appear more behaviorally retarded than the reader is reminded that in the multiple-response echolalic children. The figure also suggests that measures, the "adult" was unfamiliar to the the mute children show the largest gains in treat- child, and during the Attending condition initi- ment. They show the largest proportionate reduc- ated no interaction with the child. Therefore, tion in self-stimulation and largest proportionate any social and language behavior during the gains in the verbal behaviors. While this may be Attending condition was an indication of spon- a correct inference, it must be remembered that taneous, "self-initiated" behavior. Figure 3 pre- our measuring system gives equal weights to all sents social non-verbal and verbal behavior behaviors within the various categories. For ex- separately for the Attending and Inviting con- ample, while the mute children showed a pro- ditions. Examining the data closely (top half portionately greater increase in appropriate of the figure), it is noteworthy that there was verbal behavior, the speech of the echolalic chil- an absence of social non-verbal behavior in the dren seemed qualitatively superior to that of Attending condition until about eight months of the mute children. More exact descriptions of the treatment. The appearance of this behavior sig- changes in speech are presented on film (Lovaas, nals the children's spontaneous initiation of be- 1969) and in a separate paper (Lovaas, in prep- havior, a very important sign of therapeutic eration). Perhaps it is sufficient to say that both progress. mute and echolalic children improved with treat- The same spontaneous interaction was repli- ment, ignoring the more specific comparisons. cated in the case of appropriate verbal behavior The data are now presented separately for each (lower half of Figure 3). The data again indicate group. that the children began to initiate verbal contact GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 143
Pam and Rick Billy and Chuck Verbal Behavior 0 0 Verbal Behavior 9
6
Appropriate' '0 3 Verbal Verba / Echolalia tiJ /s~~~~d o 1.-1 1 1 - 13 A1 1964 1965
60 r Nonverbal Behavior Nonverbal Behavior l~j* 451- 4 -o Appropriate 301- Pay0 A1,'0"A.0 AJcb0 - --o , Social A .O--,O- O-S ° ~Nonverbal 151- /"
A_ It-A I 0 Imuatim il I I I I I I I 1, I I I I I I 1,j Before 3-4 7-8 11-12 Before 3-4 7-8 11-12
Treatment Months during treatment Treatment Months during treatment 1964 1964 Fig. 2. Monthly multiple response measures for the first group. Rick and Pam's data are presented on the left, and Chuck and Billy's data are presented on the right. The top part of the figure shows changes in verbal behavior, and the bottom part shows changes in nonverbal behavior. Data are averaged over two-month pe- riods. with the attending adult after the eighth month An important observation has to do with indi- of treatment. Predictably, both social non-verbal vidual differences in the rate at which the chil- behavior and verbal behavior were higher dur- dren displayed these behaviors. Figure 4 shows ing the Inviting than the Attending condition. the change in appropriate verbal behavior over In the Inviting conditions, the attending adult the Attending versus the Inviting conditions facilitated the children's social behavior by for each of the first four children. As can be instigating numerous interactions. The facilitory seen in the Attending condition (Figure 4), only effect increased as treatment progressed. It also Ricky and Billy progressed to the point where seems reasonable to us that the children show they came to initiate verbal behavior with the more social non-verbal than language behavior attending adult. However, all the children because the latter is more difficult to build. learned to interact when the adult initiated the 144 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG 28r- Rick, Pam, Chuck and Billy
21f- Social Nonverbal
/, 14H l \ ,
U) w 71-* , ~~~~~~~~~~~~~~~~~~oo z Ld o 0 0~~~~~-~o "W.P.0 0 1 _ _ _ _ O _- 0 I I 0 0 II~~~~~~~~ 16r- z Id ,/nviting Id Appropriate Speech 0L 12p-
8H
4k- .~,01% 0 Attendlin .. .--.-,.0% .O oole dol0.0
11#0 0 . - --0----O----O I I I I I i dt 1-2 3-4 5-6 7-8 9-10 11-12 13-14 Before Treatment MONTHS DURING TREATMENT Fig. 3. Monthly recordings of the first group's social nonverbal and verbal behavior presented separately for the attending and inviting conditions. Data are averaged over two-month periods. Per cent occurrence of each behavior is presented on the ordinate. conversation, as is indicated by data for the In- we employed no aversive stimulations (shock, viting condition. spankings, etc.) for the first six months of treat- Group 2 (Taylor, Mike, and Jose) was treated ment; and second, we initially planned a much similarly to Group 1, with two exceptions: first, less-demanding schedule for the children. That GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 145 Appropriate Verbal Rick Pamela 25 r 20 _ Inviting 151 - ~ ~ ~~~~% I 10 I; %o- I I I
0 z 5 0- -.O .0 ll _ O__e~--4-ol Lu
0
z Billy Chuck al 10 0 a. 81 Inviting
13-14 f 1-2 5-6 9-10 13-14 Before 8efore Treatment Treotment TWO MONTH AVERAGES DURING TREATMENT Fig. 4. Monthly recording of verbal behavior presented separately for the attending and inviting condition for each of the first four children. Per cent occurrence of the behavior is presented on the ordinate. Data are averaged over two-month periods. is, we left a child at a certain level of mastery children also differed from those in Group 1 in for a relatively long time before we introduced that all three were mute. the next task. We also attempted some variation The data on Group 2 are presented in Figure on imitation training by pairing food with the 5. The measures were taken every three months, therapist's vocalizations, instead of demanding as is noted on the abscissa. Results from Group the difficult discriminations described earlier. 2 essentially replicate the results obtained from We did not observe any particularly encouraging Group 1. There is a gradual replacement of in- improvement in the children's behavior after appropriate by appropriate behavior. We have six months of such treatment, so we returned to not plotted changes in verbal behavior, because the more demanding treatment program the these were minimal, rising only to 1 or 2% first group received. Essentially, then, Group 2 after 12 months. received the same treatment as the first, al- Group 3 (Leslie, Tito, and Seth) and Group 4 though it was somewhat less intensive. These (Kevin F., Ann, and James) were all outpatients. 146 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG
50 Toylor, Mike, ond Jose For these children, we served more as consul- Appropriate tants to the mothers, doing less direct therapeutic 40 P/oY work with the children ourselves. was 144 _ \ James essentially mute, the others were echolalic. Mul- 30 tiple response records for these children were 144 V / made before and after 1 yr of treatment. Data Se/f- from Group 3 (Leslie, Tito, and Seth) are shown 20,, Stiulation on the left side of Figure 6, while the data from Group 4 are shown on the right side. The data NSoncal from Groups 3 and 4 replicate the results from .-_,--_. _ Groups 1 and 2: a decrease in psychotic be- 0 |______|havior and an increase in normal behavior. Start- Before Months treatment 1966 ing with the top line in each graph, one can Treatment during observe a rise in Appropriate Play, Social Non- 1965 verbal, and Appropriate Verbal. Concurrently, IFig. 5. Monthly multiple-response measures for there is a drop in Echolalia. the second group. Per cent occurrence of each be- havior is plotted on the ordinate. Data are averaged While the measures on Group 4 (on the right over three-month periods. side of Figure 6) did not reflect greater improve- ment than with the other groups (this is most clear in the case of Appropriate Play), it seemed
60r Leslie, T ito , and Seth Kevin F, Ann, and James
-o P 50F- /4 0~~~~~~~~~~4 L4j ApproprIate Ploy 40H /
30F- C'
ck: 20F-
10l -* Se/f - Stimulation - - I.-- Social Nonverbal U _ Appropriate Verbal -,----0 flo(J i a ar 0 1O I-- r.(pr 0 I I L i Before After Before After Treatment Treatment Treatment Treatment 1968 1969 1969 1970 Fig. 6. Before and After multiple-response measures for Groups 3 and 4. Data for Group 3 are presented on the left, and fof Group 4 on the right. Per cent occurrence of the behaviors is plotted on the ordinate. GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 147 clinically that the children in Group 4 made children) have now been seen for follow-up data greater gains during treatment than the other on the multiple-response measures. These mea- children. The failure of the multiple-response sures were taken anywhere from 1 to 4 yr after measures to reflect this improvement may be termination of our treatment. The children may based on the failure of those recordings to make be divided into two groups, those who were dis- discriminations beyond a certain level of be- charged to a state hospital and those who re- havioral complexity. We have previously mained with their parents. The overall data on (Lovaas, et al., 1965a) pointed out that some the 13 children are presented in Figure 7. Per of the behavioral categories failed to discrimi- cent occurrence of the various behaviors are nate beyond certain ages for normal children. To plotted on the ordinate for before (B) and after overcome this difficulty we began to differentiate (A) treatment, and the latest follow-up (F) between different "levels" of social non-verbal measures are presented. "I" denotes average re- and appropriate play (as was presented earlier in sults for the four children who were institu- instructions for rater identification). We in- tionalized (discharged to a state hospital), and tended to improve the sensitivity of the record- "P" denotes data for the nine children who ing procedures by making discriminations be- have lived with their parents since their dis- tween certain behaviors; for example, differences charge from treatment. The trends in these in play behavior that involved "simple" acts like data may be succinctly described. The children repetitively dropping beads into a jar (level 1), who were discharged to a state hospital lost as compared to imaginative doll play (level 2). what they had gained in treatment with us; their When the data are presented using these new psychotic behavior increased in frequency (self- categories, as is done in Table 1, it becomes stimulation and echolalia). They appear to have apparent that Group 4 made most of its gain lost all they gained of social non-verbal be- in the "higher" levels of social non-verbal and havior, and they lost much of what they had gained in appropriate verbal and play behaviors. Table 1 On the other hand, the children who stayed Before and after measures for groups three and four with their parents maintained their gains or on levels I and II of social non-verbal behavior and appropriate play. improved further. Let us examine these children more indi- Social Appropriate Non-Verbal Play vidually, discussing the follow-up data of Rick and Pam first. When we terminated Rick and Group Three Pam's treatment we decided to recommend to Before Treatment 1 1 18 10 their parents that their children be institutional- After Treatment 4 3 41 1 1 ized. We based this decision on two major considerations. we Group Four First, had made the mistake Before Treatment 4 1 32 8 of isolating the parents from their children's After Treatment 5 3 37 16 treatment, such that they did not receive the training we did in handling their children. appropriate play behavior, while Group 3 made Secondly, these parents had other large commit- most of the gains in the "lower" levels. If one ments to their families or themselves. For ex- plans to measure treatment effects on children ample, Pam's mother had just given birth to a who have a higher level of behavior develop- child with severe brain damage which required ment than the first three groups, then some at- continuous care, she had several other children, tempts may have to be made at discriminating and Pam was not an easy child for anybody to between "levels" of certain behaviors. handle. Ricky's mother was divorced and needed Follow-up measures. The four groups (13 full-time employment. There were other con- 148 0. I. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG 60r Dstitution 501 w 0 z I w 40 QI 30F I I 0 z I I w 20 a: / I %
P 0-H I _ (.) Se/f-SStimulation 60- Nonverbal Behavior 45- / N41 / N~~~~~~~~~~~~~~* 30- ""*N I N%. N" 15- 0o N". "%%. Appropriate P/ay A wooo", %.% 0- I I I I I I I Socia/ Nonverba/ 1964 1965 1966 1967 1968 1969 1970 Pre Post Pre Pc)St Treatment I Treatment IL Fig. 8. Multiple-response measures for Rick and Pam presented Before (1964) and After (1965) first treatment, and for the first follow-up (1968), second treatment (1968), and second follow-up (1970). Per cent occurrence of the behaviors is presented on the ordinate. 150 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG apparent that the children had not forgotten can observe a loss in Appropriate Play, and a what we had taught them, but that their prob- substantial increase in Self-Stimulation and Echo- lem was essentially motivational. They were not lalia. afraid to behave inappropriately, neither did At the time that this report was written, Pam they behave appropriately in order to gain and Rick were both hospitalized. Clinically approval. The second treatment (Treatment II, speaking, Pam showed few, if any, effects from 1968) consisted essentially of reinstating the con- the treatment she had received from us; she tingencies they had experienced earlier. That is, was extremely retarded in most areas of function. if they behaved inappropriately (self-stimulated Rick had fared better. He was still definitely or were echolalic), they were punished. If they autistic, but the psychologist who examined him behaved appropriately, they were fed and ap- noted that he had received much training and proved of. The second treatment lasted 24 hr for had developed to a greater extent than many Ricky and three weeks for Pam. The results for other autistic children. He was more verbal and Treatment II, 1968 (Figure 8) indicate that this rarely echolalic. He had generally flat affect al- very short exposure to the program was reason- though he could smile and showed some positive ably effective. One can see a rise in the three feeling and was definite about what he wanted appropriate behaviors (Appropriate Verbal, Ap- -showing little ambivalence. propriate Play, Social Non-Verbal) and a de- Billy and Chuck were discharged to the same crease in Self-Stimulation. institution under conditions similar to those of The children were institutionalized again and Pam and Rick. Their parents were essentially un- brought back for a final follow-up in 1970. As trained and had other serious commitments and can be seen in Figure 8 (Follow-up II, 1970), personal difficulties. At the time of Chuck and they again regressed, as they had earlier. While Billy's follow-up 4 yr later, they had retained Appropriate Verbal and Social Non-Verbal be- most of their gains in Appropriate Play and haviors seemed to have remained stationary, one Social Non-Verbal behaviors but they showed Table 2 Group 2, 3, and 4 individual and averaged multiple-response measures per cent occur- rence of each behavior is tabled for before (B), after (A), and latest follow-up (F) measures. Self- Appropriate Social Appropriate Stimulation Echolalia Verbal Non-Verbal Play B A F B A F B A F B A F B A F Group 2 Taylor 74 32 11 00 01 00 00 00 00 00 00 03 01 05 38 Mike 18 06 27 00 01 00 00 00 00 00 06 02 17 56 39 Jose 53 01 04 00 01 00 00 00 00 01 02 03 07 62 65 Mean: 48 13 14 00 01 00 00 00 00 00 03 03 08 41 47 Group 3 Leslie 12 04 19 10 11 01 03 12 16 03 10 11 47 58 43 Tito 02 15 08 01 02 01 01 05 18 00 01 02 07 42 51 Seth 12 12 18 04 00 00 05 11 02 03 08 04 31 56 36 Mean: 09 10 15 05 04 01 03 09 12 02 06 06 28 52 43 Group 4 Kevin 03 02 03 25 11 00 02 16 07 02 02 02 51 79 68 Ann 25 19 02 03 02 05 03 05 13 02 21 02 26 25 66 James 07 08 00 00 00 00 00 01 06 11 02 01 44 55 49 Mean: 12 10 02 10 04 02 02 07 09 05 08 02 40 53 61 GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 151 losses in Appropriate Verbal behavior, increased social non-verbal, and appropriate verbal be- Echolalia, and showed a marked increase in Self- haviors were usually retained. The data indicate Stimulation. At the time of this report, Billy was some increase in self-stimulatory behavior after in a foster home, where his adjustment was said treatment was ended. to be marginal. Chucky's mother may be able Intelligence and social maturity. We have to take Chuck back home. obtained IQ scores for 19 of the 20 children. The follow-up data on Groups 2, 3, and 4 Figure 9 shows the changes in these measures are essentially presented in the group denoted by during the time the children were in treatment. P (parent) in Figure 7. The children were at IQ scores are plotted on the ordinate for before home with their parents, who had received some (B) and after (A) tests. The dotted lines indicate training in how to continue treatment with their that the patient was untestable. Most of the children. Group 2 parents received less training children showed substantial changes with treat- than Group 3 and 4, and their children were ment, functioning in the mildly to moderately treated on an inpatient basis. Group 3 and 4 retarded level by the termination of treatment, children lived with their parents throughout while they were previously untestable. "Untest- treatment and their parents received extensive able" means that the children would not respond training by us. The children were evaluated to the examiner's attempts to test them. For ex- from 2 to 4 yr after termination of treatment. ample, they would not sit in a chair if asked to The data on the individual children are pre- do so, and they remained oblivious to the testing sented in Table 2. materials that were presented to them. After Considering the grouped data, we can see that treatment, the children would cooperate; that is, the children whose parents were trained largely they would respond to the examiner and engage retained their gains or continued to improve. The in the behaviors he wanted (such as block build- gains the children made in appropriate play, ing, etc.). Some of this change reflects extinction 100 r B= Before A= After A 80F B A A 601 A A A IQ A A A A A A A I B ts 401- B A A A A A lI A I I I B I I I I I 20 F I I I I I I I I I I I I I I I I i I I I I I I I I B B B B B B B B B B B 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 OL- I II *II I PAM CHUCK TAYLOR JIMMY DANIEL KEVIN K. DANNY SETH ANN RICKY BILLY MICHAEL JOSE DEAN WILLIAM LESLIE TITO KEVIN E JAMES PATIENT Fig. 9. IQ scores Before (B) and After (A) treatment. Dotted lines indicate the patient was untestable be- fore treatment. 152 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG of interfering behaviors, while some reflects scores, and then averaging these percentages for genuinely new acquisitions. It is an open ques- all behaviors. The reliability between observers tion whether these changed IQ scores would was checked on a monthly basis, and if they be predictive of the children's future perform- exceeded the 20% given above, they were re- ances in school. trained. Table 3 presents pairs of observers' re- We obtained Vineland social quotients for the cordings randomly drawn from data on the first last 14 children we treated, and all of the chil- group of children. The table reflects high agree- dren made large gains. The mean social quotient ment between observers. The table also shows before treatment was 48, with a standard devia- that during the first year we recorded physical tion of 20; the mean quotient after treatment contact, tantrums, and vocalizations, which we was 71, with a standard deviation of 27. The omitted in subsequent recordings. Physical con- changes that took place in social maturity are tact and vocalizations were omitted because we consistent with the IQ data in that the children felt they were not directly related to the child's showed large gains in their ability to look after chances for success in "normal" society. Tan- their own practical needs. Much of this change trums were omitted because they were low- was again due to a reduction in bizarre be- frequency behaviors that changed very quickly; havior, and the achievement of elementary social therefore, no change over time was apparent in stimulus control. As with the other measures, our measures. there were no exceptions to the improvement; The high degree of agreement between ob- all 14 of the children had higher quotients after servers, and the use of non-systematic assign- treatment than they did before treatment. We ment across observations, supports the argument did not obtain follow-up data on these measures. that the data can be replicated, and that they are not the product of a particular observer. However, observers' familiarity with the study, RELIABILITY and knowledge that the children were in treat- We attempted to solve the problems associ- ment may have increased the probability that ated with reliability of the multiple-response re- ratings would reflect improvement. This seems cordings in two ways. First, we maintained, for unlikely when one considers the explicitness the majority of recordings, at least two trained with which the behaviors were defined and the observers who were assigned in a non-systematic subtle changes that often appear in the data. fashion to do the recordings. These observers These kinds of changes would be difficult to changed over time, such that those who scored "fake" and hence appear to validate the chil- for the second half of a year were often different dren's improvement. Nevertheless, the study is from those who scored the first half. This means greatly strengthened by the following investiga- they had different degrees of familiarity with the tion, which demonstrates that naive observers, children. These steps, of rotating observers and scoring the sessions in a random order, produce bringing in new ones, probably helped to re- data similar to that of our experienced observers. duce observer bias in the recordings. Each new Three observers, who were not familiar with observer received about three to six 1-hr training the children and the purpose of the study, were sessions. The various behaviors were defined for introduced to the recording system and presented him, he became familiar with the apparatus, with video recordings (Sony EV 200 1-in) of the and he worked with an experienced observer children displayed on an 8-in. television monitor. until the average difference between their scores One tape was used for training purposes. The over all behaviors was less than 20%. This was tape selected was a pre-treatment tape of a 7-yr- calculated by dividing the difference between old echolalic autistic child who displayed all of given pairs of scores by the average of the two the behaviors the observers were required to GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 153 0 -1 O \O - I-X 11 - °- - \O '-' X~~~~~~~~~~ - "N~~~~~~~~~~~~~~~~~C 0 00-00 tol 0 >- t e a =1- °o0000n -0 U %4.4 0 C\ 0 cll '--4 ON ._040 '- (N 00 0 0 r-4 &4 a) (N 0 '4 0 \ VI% '4.4 (o 0 0 0 3' a 0 0 0 \0 U) r-- 0 0 eN _- r- a) -50' 40 '._ ( - 0 0 p4 .# 0 vE 0 4fIN #) s (N1 r-- "C 0 0 T-4 (N1 r-l- VI% 0 0) 0 0) 00 (N 0w1 o o o ~~~~~~~~~- 0 0 O 0 (N (N VIA 0 00 (N 0- C U QC"-.) Q - 154 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG record. Observers were given a total of nine onset of a particular behavior because one or 1-hr training sessions. In the first three sessions, more of the other recordable behaviors were observers viewed the tape to familiarize them- present. selves with the behaviors to be scored, learned The data showing the acquisition of agree- the position of the keys on the board, etc. These ment between these three observers over the three informal sessions were followed by six ses- last six sessions are given in Figure 10. In all sions, after which observers' raw data were re- cases, a mean figure for the three observers for duced. The observer was given feedback about any one behavior during a given condition was whether his scores were high or low in relation to obtained. The data reduction was carried out as the mean score for a particular behavior over the follows: the onset of the tape was clearly three observers. The tape was then replayed to marked by the observer on the computer tape all observers, and the instances where they had by the observer depressing all the keys for 5 sec. disagreed were discussed. This kind of training The end of the tape was similarly noted. Each was like that given to the observers who ob- tape was then divided into 10-sec segments. served the children in vivo. The recordings were Agreement was recorded in an all-or-none fashion different in that the observers were required to for each 10-sec segment. If two observers divide social non-verbal and appropriate play agreed, a "+1" was recorded. If they did not each into two levels (see the Instructions for record the same behavior during that particular Observer Identification), which, with eye-to-face period, a "0" was recorded. Per cent deviation and self-stimulation, demanded the recording then refers to the per cent of these 10-sec seg- of eight behaviors. During the first three sessions, ments during which a particular observer failed observers were required to record all eight be- to record a behavior recorded by either of the haviors simultaneously. This proved too difficult, other two observers or during which a particular so we then decided to simplify the task by play- observer recorded a behavior not simultaneously ing the tape twice and asking the observer to noted by either of the other two observers. Per record only four behaviors per observation ses- cent deviation was averaged over all conditions sion. Low agreement during Session 4 (Figure and behaviors for a given observer for each 10) reflects an adjustment to the new procedure session. 0-7 (Observer 7), then, shows an aver- to some degree. In addition, observers rarely age per cent deviation of 35% during the first disagreed about which category a particular be- session. The figure shows that observers eventu- havior represented. Disagreements between ob- ally learned agreement. servers were usually disagreements about the After the completion of this reliability train- onset and termination of a particular behavioral ing, one of the observers (0-7 in Figure 9) segment. Much of this disagreement was basi- scored the pre- and post-treatment tapes on three cally a problem of skill, that is manual dexterity. children (Michael, Ricky, and Jose) to assess One must observe, categorize, and record simul- whether a naive observer could record the taneously. This means that the observer could treatment effects in agreement with our ex- not also look at the keyboard. The observer had perienced observers. She scored the tapes in the to make decisions about which key to depress, following order: Mike's pre-treatment, Mike's at what time to depress it, and for how long. He post-treatment, Jose's post-treatment, Ricky's had to observe, categorize, and record three other post-treatment, Ricky's pre-treatment and Jose's behaviors. It is a complex, on-going process and pre-treatment. All tapes were scored in order a mistake may be made at any of several points of condition Alone, Attending, and Inviting in the chain. The two most common errors were except Mike's, where the conditions were scored continuing to depress a key after a particular in order of Attending, Inviting, and Alone. behavior had ceased and missing the precise This is a direct test of the reliability of our GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 155 36r- 32H 28k z 0 24k LLJ 20 z 16k LU cr- LUI 12F- 8k 4k I I I I I 4 5 6 7 8 9 SESSIONS Fig. 10. The acquisition of agreement between observers 7, 8, and 9. Per cent deviation from the mean of the three observers' scores is plotted on the ordinate for the last six sessions. Per cent deviation is averaged over all conditions and behaviors for a given observer for each session. Table 4 Tape and In Vivo Comparisons of "Before" and "After" Multiple-Response Measures JOSE MICHAEL RICKY Tape In Vivo Tape In Vivo Tape In Vivo Before After Before After Before After Before After Before After Before After Self- stimulation 33 0 53 1 11 1 18 6 10 15 25 13 Appropriate Verbal 0 0 0 0 0 1 0 0 1 2 0 3 Echolalia 0 2 0 1 3 6 0 1 1 3 1 2 Social Nonverbal 0 1 1 2 1 2 0 6 1 2 0 2 Appropriate Play 6 54 7 62 24 46 17 56 33 52 16 26 156 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG scoring procedures, as it removes potential effects for both observers' familiarity with the experi- DISCUSSION ment, and the order in which they recorded the In summary, the major results of this study behaviors. Table 4 presents the comparisons be- are that: (1) inappropriate behaviors (self- tween the scores obtained by an experienced stimulation and echolalia) decreased during observer recording the child in vivo, and a naive treatment, and appropriate behaviors (appropri- observer scoring off the tape of that same session. ate speech, appropriate play, and social non- The absolute values between the two observers verbal behaviors) increased; (2) spontaneous are different, probably because the video record- social interactions and the spontaneous use of ing reduces fidelity (i.e., there is particularly language occurred about eight months into less fidelity with speech and facial expression). treatment for some of the children; (3) IQs and However, the naive observer recorded increased social quotients reflected improvement during frequency of normal behaviors, and decreased treatment; (4) there were no exceptions to the frequency of pathological behaviors, similar to improvement, however, some of the children the changes recorded by the experienced ob- improved more than others; (5) follow-up mea- server. The only exception to this is Ricky's self- sures recorded 1 to 4 yr after treatment indicated stimulation, which was observed to increase that large differences between groups of chil- during the "after" measures when scored off the dren were related to the post-treatment environ- tape. Observers suggested that this may have ment (those groups whose parents were trained been due to the large amount of Ricky's facial to carry out behavior therapy continued to im- contortions, which were detected and scored as prove, while children who were institutionalized self-stimulation in the in vivo pre-treatment regressed); (6) a brief reinstatement of behavior condition, but which could not be detected on therapy could temporarily reestablish some of the tape. the original therapeutic gains made by the chil- To summarize our findings concerning the dren who were subsequently institutionalized; reliability of our recordings: (1) there was a and (7) the technique utilized for recording high degree of agreement for pairs of ex- therapeutic change was reliable. Observers were perienced observers who were randomly assigned trained to be able to recognize and record to do in vivo recordings; (2) naive observers, specific behaviors, the presence or absence of unfamiliar with the study and children, could which may be considered diagnostic of autism. also be trained to show high agreement in their scoring of video recordings of the children's Individual Differences behavior; and (3) the direction of the behavioral While the major findings listed above charac- change in treatment was scored essentially the terize each of the children in the group, there same for a naive observer scoring pre- and post- has been considerable heterogeneity with respect treatment video tapes in a random order as it to the degree of improvement shown by each was for experienced observers scoring the ses- child. The delineation of "autism" is one area sions in vivo. that will demand considerably more work. It has It may be appropriate to note here that the not been a particularly useful diagnosis. Few replicability of these results on reliability may people agree on when to apply it. It is not a be open to question, as the procedure requires functional term in the sense that it is neither that the trainer be very familiar with the types related to a particular etiology nor to a particu- of errors naive observers are likely to make and lar treatment outcome. Our children responded to be able to note where such an error has in vastly different ways to the treatment; Rick occurred and provide appropriate feedback to learned, in 1 hr, what Jose learned in 1 yr. Since the observer. there was such heterogeneity among the patients, GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 157 three fairly representative case descriptions will since he was responsive to social reinforcers. be presented in order to give a picture of the Again, the language program (Lovaas, in prepa- clinical implications of our findings: first, we ration) gives a rather complete discussion of present Scottie, who showed considerable gains these procedures. We focused our efforts on in treatment; second, Tito, who showed moder- consultations with his grandmother, who was ate gains; and finally, Jose, whose progress was brought in to take special care of him, and who minimal. held him to a highly demanding schedule. She Scottie, who was 4.5 yr of age at the start of was different from most of the parents we had treatment, spent much of his time staring into seen, since she did not tolerate withdrawal or space and did not attempt to initiate interactions other expressions of pathology. wtih people. If he was directly addressed, he Scottie is presently attending third grade in a would show a passive and friendly interest. normal elementary school. His social quotient When left to himself, he would self-stimulate; is 100. He shows no trace of autism, and in all he was particularly attracted by spinning wheels. aspects must be considered a normal child. Scottie was echolalic, and he could label com- Tito, at admission, was a hyperactive 5-yr-old mon objects, but he had very little communica- boy who evidenced an extremely short attention tive, and no spontaneous speech. He had to be span. Eye contact was absent. He had many washed and dressed by others, helped when he compulsive rituals. For example, he would spend ate his meals, and he was not fully toilet trained. considerable time arranging objects in a straight He was, however, relatively free from tantrums, line and would become very upset if the arrange- and he could understand simple directions. Be- ment was disturbed. He "refused" to let his cause of his social responsiveness when ap- parents read the Sunday paper by becoming very proached, and occasional appropriate play, he upset when they removed the string that tied it was considered less psychotic than most of the in a bundle. He was untestable on intelligence children we have seen. His social quotient was tests, and he obtained a social quotient of 52. 68. He was echolalic, but occasionally would use Initial treatment sessions took place in his speech for communication. Tito's understanding home, lasting 2 to 4 hr each, several days a of speech, however, was minimal. He resisted week. His treatment plan included programs any involvement with people, and was unre- designed to teach him communicative skills, as sponsive to displays of affection. He was ex- well as the behaviors necessary for him to take tremely negative and clever at getting himself care of his own practical needs, and to take a out of situations he did not like, often respond- more active part in his everyday life. He was ing to the most elementary demands (e.g., "sit in taught common abstract terms, such as preposi- the chair") with extreme tantrums. tions, words denoting temporal relationships as Tito was treated as an outpatient. He was well as counting, singular versus plural, etc. The seen for three sessions per week for 1 yr. We frequency of echolalia was decreased. Meals and served primarily as consultants to his mother, almost all daily activities were made strictly con- who was very conscientious and warm. His tingent on verbal requests for them. At first he treatment program included two main objec- missed several meals. He was also taught to tives. First, we tried to teach him to deal with offer materials as well as to request them. Much frustrating situations more maturely rather than emphasis was placed on conversational speech. engaging in tantrums. This was an extremely He was asked a general question about some- demanding job, and he received many spankings. thing he had done, and then was asked progres- Secondly, he was taught those basic skills upon sively more specific questions. Any spontaneous which he could build more complex behaviors, responses he made were reinforced socially, particularly in language. Included in this cate- 158 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG gory were pronoun usage, preposition usage, cluded a number of common nouns, some number concepts, relational concepts such as big names, and a few verbs. He would use these and little, and social greetings. He was taught words to label objects or express a desire for to comment upon his environment. He was also something, but never for commenting. He taught to draw, and to play more appropriately. would attempt to imitate new words sponta- At discharge, Tito was observant and alert, neously on occasion. His greatest improvements but still appeared definitely educationally handi- were social. His social quotient of 74 reflected capped. He had made some progress in most increases in smiling, laughing, and self-help areas, but his biggest gains were in language. skills. He was partially toilet trained. He was His speech was quite spontaneous, and he could testable on the Stanford Binet Intelligence comment correctly on most social interactions. Scale (IQ 47). He obtained an IQ of 47 on the Stanford Binet At present, Jose uses only a few words spon- and a social quotient of 63 on the Vineland. He taneously (e.g., "car", "go to school", etc.), and now attends a school for retarded children 3 hr what he has retained of his speech training is per day. His mother reports that he has con- negligible. He can take care of himself at din- tinued to show improvement in most areas. He ner, and is fully toilet trained. His greatest gains remains emotionally aloof with strangers, but at home have been in his play, which has be- is close to his mother. His future is uncertain. come elaborate and creative, enabling him to He may escape institutionalization if his progress entertain himself. He appears indifferent to continues. people. While at intake he appeared unaware Jose was 4 yr old at the start of treatment. of ("blind" and "deaf" to) social contact, it now His extreme negativism was reflected in tan- looks more as if he does not care whether trums, biting, and extreme stubbornness. He did anyone is there or not. He has to be watched not play with peers. He did not respond to his constantly. Otherwise, he runs away from home, name or any commands. He had no speech, going nowhere in particular. His parents fear he could not dress himself, was not toilet-trained, will be killed because he is unaware of many nor did he have any other self-help behaviors. common dangers. His parents plan to place him Appropriate play was essentially absent. He was in a nearby state hospital. He will be able to found to be untestable on intelligence tests. He come home on weekends and short vacations. had a social quotient of 59. In short, he was We concur in these plans. extremely behaviorally retarded. It is important to note that given the con- He was treated as an inpatient at the UCLA siderable heterogeneity among patients diag- Neuropsychiatric Institute for 1 yr. His mother nosed as autistic, it is not enough merely to say was given some limited training in how to con- that one has treated autistic children. Consider- tinue therapy with him as described above. His ing that some children improve without treat- treatment was primarily designed to overcome ment, and that these children are differentiated his negativism and to build some basic language from those who do not by certain behaviors, a skills. The latter included simple labeling, color good diagnostician can select his patients so that discrimination, response to simple commands, the majority of the children would eventually and form discriminations. Some work was also improve independent of the treatment offered. done on the reinforcement of spontaneous No doubt such pre-selection of patients, which babbling. would yield a much more favorable base rate We probably made slower progress with Jose change than is true of a nonselected group, keeps than with any of the other children. At dis- many non-functional treatments alive. charge, his gains in language were minimal. He We utilized several procedures that allow us would obey some commands; his vocabulary in- to argue with some confidence that autistic chil- GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 159 dren improve with behavior therapy. First, we signed into the hospitalized versus parent- have performed two within-subject replications trained groups so as to minimize child charac- (on Rick and Pam), and in both instances dem- teristics associated with differential prognoses. onstrated that we could establish behavioral con- Age of the child, testability on standard psycho- trol at will over the course of time for these pa- logical tests, amount and kind of play at intake tients. Therefore, we can argue that their are some of the more obvious variables that behavior change must have been due to our should have been equated across groups. (2) We intervention. Secondly, we performed several should have exercised more control over the between-group replications. All replications post-treatment environments. For example, we yielded similar results. Also, each group of were unable to assign children to special edu- children was treated independently of the others, cational institutions, yet presence or absence of demonstrating that we could replicate our treat- school is probably an important determinant of ment effects independent of any conditions that subsequent improvement or relapse. Similarly, may have been specific to one group of patients. future studies would do well in developing We attempted specifically to avoid pre-selecting more objective assessment of the extent to which patients who might provide a favorable base the parents are in fact continuing to implement rate of change regardless of treatment. The the treatment procedures. Not all parents are majority of patients were selected specifically equally good as behavior modifiers. The more because they displayed behaviors (IQ less than successful parents seem to have the following 50, mutism or no appropriate play) that were features: considered to be poor prognostic indicators. In (a) A willingness to use strong consequences, addition, we did not drop any patient from such as food and spankings, to be emotionally treatment once we began. Finally, we consider responsive, showing their anger as well as their our measures to be socially meaningful, and love. Not all parents can do this; some people independent of any given theory of autism. are more "gentle" or more permissive than Further evidence that directly supports our re- others, preferring to let their children "grow" sults has been provided by other behavior or "develop" while they, as parents assume a therapy programs, which also demonstrated im- spectator role. Such parents do not do well with provement with autistic children (Hewett, 1965; their autistic children. Risley and Wolf, 1967; and Wolf, Risley, and (b) The willingness to deny that their chil- Mees, 1964). Furthermore, Wolf, et al. (1967) dren are "ill". This means that they deny the provided some additional data that lend support child the "need" to be sick, and, instead, give to our follow-up results. Such data suggest that him some responsibility. replication of the data presented here is practi- (c) The willingness to commit a major part of cally feasible. their lives to their children and to exercise some Reservations about the follow-up data. We degree of contingent management throughout have some confidence in the inferences we have the day. This virtually rules out any professional drawn about the effectiveness of behavior or extensive social interests on the mother's part, therapy from the Before and After measures. requires a stable family structure, etc. Parental The study was designed primarily to assess for assistants, such as special tutors, can help out at change during treatment. But the reader should times, but ultimately the parents must bear the view the follow-up data with certain reserva- major responsibility. tions; we did not initially design the project with Despite these limitations, we have included follow-up assessment in mind and a number of the follow-up data because their validity seems variables were left uncontrolled. In retrospect: strengthened when one considers the study as a (1) The children should have been randomly as- whole. Basically, it is not surprising that the 160 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG children who were "discharged" to their parents changes on the IQ tests for free; that is, we did tended to improve, because these children re- not train the children on the test items, yet they mained in a treatment environment. improved. The data on response generalization Major strengths. Behavior therapy programs are limited because psychology has little to say for autistic children help. That is their major about it. We had no information about what asset. So far as we know, despite its limitations, changes to expect, which made it difficult to it is the only intervention that is effective. Our assess changes this first time around. From casual program did not give everything to every child. observation, during the treatment, the children Sometimes it gave very little to a particular looked much more "alert" and "aware" and be- child, but it did give something to each child came more effectively prosocial. It was particu- we saw. The improvement was analogous to larly dramatic when, during the continuous making from 10 to 20 steps on a 100-step demands of the therapy hour, a child would ladder. Scotty probably started at 80 and gained suddenly start sobbing and put his head in the 20; his treatment brought dramatic changes, he therapist's lap; or when, after much hard work, became normal and his change is irreversible. a child appeared delighted over his new mastery. Jose, on the other hand, may have started at 10 One does not often observe appropriate affect in and gained 10; the change was not all that autistic children. dramatic. Certain changes were difficult to assess because We have been especially successful at sup- they seemed outside our behavioral framework. pressing self-destructive behavior. In minutes, The children who were chronic toe-walkers we have been able to stop the self-destruction of (one of the soft neurological signs) began children who have mutilated themselves for to walk normally after four or five months years. The suppression was highly discriminated of treatment. Children who had never slept (situation specific), but this merely meant that normally through the night began sleeping for we had to apply the treatment in more than 10 hr without interruption. Children who were one environment. chronically diarrhetic began having firm stools, We have also been successful at rearranging etc. It is the search for this kind of generalized behavior. For example, if the child was not mute behavioral change that we feel will be particu- (if he already had psychotic speech), then we larly useful in future research. could help him make large strides in his language Major weaknesses. There are many disap- and intellectual behavior. pointments, and only a full appreciation of these The gains the children made in treatment will enable more realistic hopes now, and so- generalized. Our multiple-response measures lutions in the future. We will discuss the major clearly demonstrate that we obtained stimulus problems. generalization. We realize that sometimes the The most significant disappointment was the generalization was not as broad as we might failure to isolate a "pivotal" response, or, as have suspected. The shift in our program to some might describe it, the failure to effect teach the parents to treat their own children is a changes in certain key intervening variables. direct attempt to build greater stimulus gen- This means that in the beginning, we searched eralization. for one behavior which, when altered, would We were interested in response generalization produce a profound "personality" change. We as well as stimulus generalization. We do have could not find it. We had once hoped, for ex- some data on the subject, but they are limited. ample, that when a child was taught his name Response generalization, like stimulus general- ("My name is Ricky") that his awareness of ization, deals with efficiency. How much be- himself (or some such thing) would emerge. havior can one get for free? We obtained It did not. Similarly, the child who learned to GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 161 fixate visually on his therapist's face did not difficult to remediate. We will speculate on suddenly discover people. Our treatment was not some possible basis for such a perceptual a cure for autism. But we had to start some- deficiency later in this paper. where. At least the child who learned his name The second possibility is that behavior was then in position to learn someone else's therapy is the correct approach, and that it is the name. When he learned to fixate visually on his problem that is erroneously conceptualized. A therapist's face, he could pay more attention full discussion of this point involves evaluations to teaching cues. of terms like "mental illness", "treatment", etc., The failure to isolate a pivotal response (or and that is beyond the scope of this paper. But change a crucial intervening variable, or cure it is important to bear in mind that "autism" is autism) can be discussed in two different veins. a hypothetical construct and a very shaky one First, behavior therapy may not be the correct at that. We have been expected, in a sense, to approach to treating autistic children. One could cure the children of someone else's inferences suggest in this regard that the "underlying path- about them. There are no studies on "autism" ology" (the intervening variable) is biochemical, that point to either a common etiology or a and that the early detection and correction of common response to treatment (or even a com- this imbalance will enable the child to learn mon response to experimental situations of from his everyday environment, with little or much more limited scope). "Autism" was coined no special educational remediation. If this before a functional analysis of pathology. The should prove to be the case, our research has public and emotional appeal associated with the limited ultimate clinical value for autistic chil- term, not our scientific understanding of the dren. Behavior therapy alleviates some prob- "syndrome", helps the term survive. lems now, but the ultimate solution may in- There are other conceptual problems. Con- volve correction of a biochemical imbalance. sider, for example, a behavior as easy to build as This is a viable alternative and intelligent people "looking at the therapist's face". In data lan- will pursue it. guage this behavior has a very limited meaning. Speaking of physiological variables, it is However, the associated theoretical structure is possible, of course, that the pathology is struc- very extensive, implying that the child is recog- tural. It could be that something is non-func- nizing and evaluating another person. For au- tional, as is the case with the blind or deaf tistic children, the behavior of looking or not child. The repair of this structural deficit, an at- looking at another person has acquired special tempt to "connect up" the millions of neurons significance on a purely conceptual level (cf. in order to correct or bypass the deficit, is beyond Hutt and Ounsted, 1966). According to some the limits of present medical technology. In this conceptions of the problem, one would expect case, we must approach the problem as we ap- major changes in the autistic child who started to proach blindness or deafness. In other words, look at others. In our research, we have found it is sometimes the case that even where the these changes to be of minor significance. underlying pathology is neurophysiological, the Research problems. We had anticipated that only feasible treatment may be essentially psy- the children in the state institutions would re- chological. It is the early detection of blindness gress. From the beginning, we have published and deafness, and subsequent special remedial studies (Lovaas, 1967; Wolf, et al., 1964) that environments that allow blind and deaf children show that when the experimental reinforcers to develop normally. Without special consider- were withdrawn, behaviors weakened. Such ex- ation, they closely resemble the autistic children. tinction occurred whether we employed food The perceptual deficit that may underlie autism or fear as reinforcers, and whether the behavior is more difficult to assess, hence it is more was physical contact, imitation, or abstract 162 0. 1. LOVAAS, R. KOEGEL, I. Q. SIMMONS, and J. S LONG speech. The shift from response to time-contin- dren he saw were treated in school, they did not gent delivery, the very procedure that we em- necessarily change at home. When the contin- ploy to demonstrate the effectiveness of our gencies were instated both in the home and at main treatment variable, also demonstrates our school, the behavior changed in both settings. major weakness, the tenuousness with which the Many therapists (cf., Patterson and Bechtel, behavioral gains were maintained. Reversible 1970) now argue that the child's parents are baselines help one's research design, not one's essential as mediators of treatment. patients. The extent of the reversibility of treatment The reversibility of the treatment effects was effects will probably be some function of various most dramatically observed with the first four patient characteristics. It seems that when the children we treated: Rick, Pam, Chuck, and primary problem centers on the child's motiva- Billy. Although each one was making progress tion, and when treatment relies on "artificial" or at the time of discharge, when we assessed them experimental reinforcers (stimuli that do not after 4 yr in the state hospital they appeared to characteristically maintain the patient's behavior have made no gains in appropriate behavior dur- on the outside), then one invites certain prob- ing that time, and showed large increases in self- lems. Food, slaps, and accentuated social rein- stimulation. It is probably worth elaborating on forcement are not the reinforcers that maintain this point, that they had learned nothing in the the daily behaviors of normal school-aged chil- intervening 4 yr, but seemed to have stayed dren. Our use of such reinforcers set up the exact still. They gave the same smiles, the same facial conditions for the kind of discriminations and expressions, the same words. This again the kind of extinctions we did not want. The emphasizes the point that without therapeutic, necessity for using primary reinforcers, rather prescribed, contingent, functional, reinforcers, than everyday, more natural ones, is a probable children like these do not improve or retain their reason why the children regressed in the state improvement; and, since we are not yet in a hospital environment. The state hospital, since position to help them acquire normal, social re- it did not prescribe contingent functional rein- inforcers, their post-treatment environment has forcers, constituted an extinction run. to be controlled. In our philosophy, functional It is implied in the above discussion that we contingencies are reality; if removed, any child view the problem of maintaining the treatment would fail to develop. gains (generalization over time) as a special case The reversibility of the treatment effects are of stimulus generalization. When the child not peculiar to autistic children. It has been stayed home with his parents who had learned observed in a large variety of behavior therapy our techniques he did not regress (i.e., ex- programs, and has led Bandura (1969) to tinguish) because the environments before and speak of the distinction between physical and after discharge were similar (i.e., we maintained psychological treatments. The work of others stimulus generalization). However, the child who illustrates these problems. For example, Tharp was discharged to a state hospital entered a new and Wetzel (1969) attempted to avoid discrimi- environment to the extent that it did not possess nation and reversal problems by placing the (or did not program) effective reinforcers. Re- treatment in the hands of those persons who member that the children had not "lost" the be- have control over most of the patient's rein- haviors we had given them (some "progressive forcers. Our failure to maintain the gains made disease" had not rotted their brains), they simply by the first four patients underscores the need did not perform, they were unmotivated, unless for that kind of intervention. Wahler's (1969) we re-exposed them to the treatment contin- study provided a good illustration of the need gencies. The point is that it is important for re- for interventions across settings. When the chil- search in behavior therapy to be directed towards GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 163 ways of normalizing reinforcing functions so as autistic children. As we have shown, the chil- to smooth the transition (prevent a discrimina- dren responded in vastly different ways to the tion) between the treatment and post-treatment treatment we gave them. We paid scant atten- environments. tion to individual differences when we treated The second major research problem centers the first 20 children. In future, we will assess on how to develop procedures for accelerating such individual differences, for example, by the acquisition of new behaviors. No doubt, contrasting the effectiveness of behavior therapy the slow rate with which some behaviors were for very young versus older autistic children. acquired was based on the children's inade- Presumably, the very young (before 2 yr of quate motivation, as discussed above. However, age) child should discriminate less well than it also seems to be the case that autistic children the older child, hence generalize and main- show deviations in attentional behaviors and tain the therapeutic gains to a more optimal that these deviations slow down their acquisi- degree. tion, particularly of that kind of learning that It is important to remember that behavior requires shifts in stimulus control. Autistic chil- therapy is a treatment based on research rather dren appear to respond in an overselective man- than deduced from theory. It is a technology for ner to multiple cues. We referred to this prob- producing behavioral change through environ- lem as stimulus overselectivity (Lovaas, et al., mental manipulations. Sensitivity to new find- 1971). This finding has led us to consider re- ings produces constant change in treatment tech- designing many of our teaching procedures. For niques. Only a short time ago, imitation training instance, we may well have to minimize our use procedures were developed (cf., Baer and Sher- of supporting prompts and prompt fading tech- man, 1964). Our treatment changed and its niques, as these may provide interfering rather effectiveness was greatly increased. Similar gains than helping stimuli in the learning process of occurred when we developed procedures for these children. building abstract speech. Just a short time ago, Perhaps it is stimulus overselectivity that pre- many argued that autistic children were unable vents or slows down the autistic child's acquisi- to imitate, and that they were unable to form tion of secondary reinforcers as well. If pri- abstractions. mary reinforcers had been used as consistently In closing, we should note that many of the with normal children as we used them with procedures we have described are not new, but our autistic patients, the associated environ- bear striking similarities to those described by ment would probably have acquired a larger Itard ("The Wild Boy of Aveyron") and by range of reinforcing function. If secondary re- Sullivan (in Gibson's "The Miracle Worker") inforcers are acquired classically (through the and recently by Clark ("The Siege"). We are simultaneous presentation of neutral with al- especially struck by the similarity in their ready functional stimuli) then the autistic child, willingness to use functional consequences for being overselective, may fail to respond to one the child's behaviors, the meticulous building of these inputs, and conditioning should then of new behaviors in a piece-by-piece fashion, fail. Behavior therapy with the normal child, the intrustion of the education into all aspects then, may not require detailed consideration of the child's life, the comprehensive, hour-by- of programs designed to build reinforcing func- hour, day-by-day commitment to the child by tion. To ensure long-lasting effects in the autistic an adult, etc. child, the process of building secondary rein- So the principles we employ are not new. Re- forcers would seem to require much more effort. inforcement, like gravity, is everywhere, and has Finally, a major focus of future research been for a long time. The principles can be used should attempt more functional descriptions of to the child's advantage, or they can be used 164 0. 1. LOVAAS, R. KOEGEL, J. Q. SIMMONS, and J. S LONG against him. What is new in behavior therapy Lovaas, 0. I., Berberich, J. P., Perloff, B. F., and is the systematic evaluation of how these prin- Schaeffer, B. Acquisition of imitative speech in schizophrenic children. Science, 1966, 151, 705- ciples affect the child. It is not the content of 707. (a) behavior therapy that is new, but its research Lovaas, 0. I., Freitag, G., Gold, V. J., and Kassorla, methodology. In that sense, we have an immense I. C. Recording apparatus and procedure for ob- servation of behaviors of children in free play and often unappreciated advantage over those settings. Journal of Experimental Child Psy- who preceded us; the methodology enables us chology, 1965, 2, 108-120. (a) to contribute in a cumulative manner to psy- Lovaas, 0. I., Freitag, G., Kinder, M. I., Rubenstein, B. D., Schaeffer, B., and Simmons, J. Q. Estab- chological treatment. lishment of social reinforcers in two schizophrenic children on the basis of food. Journal of Experi- mental Child Psychology, 1966, 4, 109-125. (b) REFERENCES Lovaas, 0. I., Freitag, L., Nelson, K., and Whalen, C. The establishment of imitation and its use for the Baer, D. M. and Sherman, J. Reinforcement con- establishment of complex behavior in schizo- trol of generalized imitation in young children. phrenic children. Behaviour Research and Ther- Journal of Experimental Child Psychology, 1964, apy. 1967, 5, 171-181. 1, 37-49. Lovaas, 0. I., Litrownik, A., and Mann, R. Response Baer, D. M., Wolf, M. M., and Risley, T. R. Some latencies to auditory stimuli in autistic children current dimensions of applied behavior analysis. engaged in self-stimulatory behavior. Journal of journal of Applied Behavior Analysis, 1968, 1, Abnormal Psychology, 1971, 9, 39-49. 91-97. Lovaas, 0. I., Schaeffer, B., and Simmons, J. A. Ex- Bandura, A. Principles of behavior modification. perimental studies in childhool schizophrenia: New York: Holt, Rinehart and Winston, 1969. building social behaviors by use of electric shock. Brown, J. L. Prognosis from presenting symptoms Journal of Experimental Studies in Personality, of pre-school children with atypical development. 1965, 1, 99-109. (b) American Journal of Orthopsychiatry, 1960, 30, Lovaas, 0. I., Schreibman, L., Koegel, R. L., and 382-390. Rehm, R. Selective responding by autistic chil- Ferster, C. B. Positive reinforcement and behavioral dren to multiple sensory input. Journal of Ab- deficits of autistic children. Child Development, normal Psychology, 1971, 77, 211-222. 1961, 32, 437-456. Lovaas, 0. I. and Simmons, J. Q. Manipulation of Ferster, C. B. and DeMyer, M. K. The development self-destruction in three retarded children. Journal of performances in autistic children in an auto- of Applied Behavior Analysis, 1969, 2, 143-157. matically controlled environment. Journal of Lovaas, 0. I. Teaching language to psychotic chil- Chronic Diseases, 1961, 13, 312-345. dren (in preparation). Gibson, W. The miracle worker. New York: Ban- Metz, J. R. Conditioning generalized imitation in tam Books, 1960. autistic children. Journal of Experimental Child Havelkova, M. Follow-up study of 71 children di- Psychology, 1965, 2, 389-399. agnosed as psychotic in pre-school age. American Park, C. C. The siege. New York: Harcourt, Brace Journal of Orthopsychiatry, 1968, 38, 846-857. & World, Inc., 1967. Hewett, F. M. Teaching speech to an autistic child Patterson, G. R. and Bechtel, G. G. Formulating through operant conditioning. American Journal the situational environment in relation to states of Orthopsychiatry, 1965, 35, 927-936. and traits. Chapter in R. B. Cattell (Ed.), Hand- Hutt, C. and Ounsted, C. The biological significance book of modern personality study. Chicago: Al- of gaze aversion with particular reference to the dine Publishing Company, 1970. syndrome of infantile autism. Behavioral Science, Rimland, B. Infantile autism. New York: Appleton- 1966, 11, 346-356. Century-Crofts, 1964. Itard, J. M. G. The wild boy of Aveyron. New Risley, T. and Wolf, M. Establishing functional York: Appleton-Century-Crofts, 1932. speech in echolalic children. Behaviour Research Kanner, L. and Eisenberg, L. Notes on the follow-up and Therapy, 1967, 5, 73-88. studies of autistic children. In P. H. Hoch and J. Rutter, M. Prognosis: psychotic children in ado- Zubin (Eds.), Psychopathology of childhood. New lescence and early adult life. In J. K. Wing (Ed.), York: Grune & Stratton, 1955. Pp. 227-239. Early childhood autism: clinical, educational and Lovaas, 0. I. Behavior modification: teaching lan- social aspects. London: Pergamon Press, 1966. guage to psychotic children, instructional film, Skinner, B. F. Verbal behavior. New York: Apple- 45-min, 16 mm-sound, Appleton-Century-Crofts, ton-Century-Crofts, 1957. New York, 1969. Tharp, R. G. and Wetzel, R. J. Behavior modifica- GENERALIZATION AND FOLLOW-UP MEASURES ON AUTISTIC CHILDREN 165 tion in the natural environment. New York: Aca- child: a follow-up and extension. Behaviour Re- demic Press, 1969. search and Therapy, 1967, 5, 103-112. Wahler, R. G. Setting generality: some specific Wolf, M., Risley, T., and Mees, H. Application of and general effects of child behavior therapy. operant conditioning procedures to the behavior Journal of Applied Behavior Analysis, 1969, 2, problems of an autistic child. Behaviour Research 239-246. and Therapy, 1964, 1, 305-312. Wolf, M., Risley, T., Johnston, M., Harris, F., and Allen, E. Application of operant conditioning Received 16 June 1971. procedures to the behavior problems of an autistic (Final acceptance 7 September 1972.) lating to the major focus of this manuscript: REVIEWERS' COMMENTS namely, generalization of the treatment effects The following comments have been over situations and time. The two-condition abstracted and compiled from the more assessments of situational generalization (attend- extensive comments of the three research- ing and inviting) do provide nice measures of ers who reviewed this manuscript. the phenomenon. Since unfamiliar adults were present in both situations, the gradual increase Anyone familiar with the work of Lovaas in desirable behaviors is a good index of setting and his associates would find this new general- generality. However, the relevance of this gen- ization data most interesting and important. His eralization is, of course, dependent upon initial previous work with some of the autistic chil- demonstrations that the behavior modification dren described in this manuscript satisfied the program was responsible for the improvement. reliability and validity requirements of Applied Continuing with the generalization data, I'm Behavior Analysis and was clearly an important hesitant to make much of the follow-up data as contribution to the understanding and treatment it is described. The authors attempt to argue of childhood autism. New observational data, that the post-treatment environments (institu- reflecting the generality of changes produced tion versus home) accounted for these data. in these children (across time and situations) However, as the authors point out, these groups' would be appropriate for a JABA audience differences in follow-up could have been due to because of these prior research considerations. many factors. I think that the follow-up data I think perhaps that this article presumes too should be plotted as a possible function of these much about a reader's familiarity with the environments, but the authors should not claim authors' prior work. Their arguments, for ex- causal relationships between these environments ample, that some of these subjects satisfied the and the follow-up measures. requirements of a within-subjects' replication The authors describe an elegant and rigorous design are not at all convincing. To say that analysis of observer behavior. They carefully Group 1 "received an A B A design" is very studied the possibility of bias, habituation, and misleading if we think of B as verified removal unreliability. The care with which they treated of the independent variable. As far as we can the problem of measurement reliability is very tell from this page, B only means that the chil- impressive and should serve as a model to other dren were not in the authors' Program. That investigators. Nevertheless, I have a very serious kind of evidence is not very convincing if one is question about one of their reliability studies. to argue that utility of a particular kind of treat- The authors describe an experiment involving ment program (as the authors do). However, three observers whose acquisition of agreement some of the kind of evidence needed to satisfy was studied. The results of this study indicated these design requirements is already published. that after reading the definitions and practising Next, I see some problems with the data re- informally with the apparatus for three sessions