AUTISM: A NEW BEHAVIORAL TREATMENT

BY IVAR LOVAAS, PHD 

Lisa's parents had just been told that there was something terribly wrong with their child, something for which there was no cure. Her problems had started early. She avoided eye contact from the beginning, and was difficult to nurse because she did not want to be held close to her mother's skin. When she was two and still making no attempt to communicate with her parents, they were told not to worry: "Children will talk when they are ready to." Now she was three and still not talking.

The first time I saw her, Lisa was asleep. All children are beautiful when they sleep, and I was to learn that the parents of autistic children feel particularly attached to them then, because they can stroke, kiss, or hold them and imagine that they are loved and accepted. Lisa woke, and her parents reached out to her as she rose. They called her name, but received no answer. One sensed an overwhelming emptiness, difficult to understand and accept. She ignored their smiles, slid gracefully away from their outstretched arms as if she had never seen them, and ran across the room to stare in total absorption at the reflections in a shiny door handle. Later she would crouch behind the sofa and rock her body for hours -- one of many ritualistic routines. She did not play with toys or friends, and she was mute. (Some autistic children will parrot words they have heard.) On the street, Lisa seemed unaware of traffic and would walk in front of a car if she were not restrained. Fewer than five percent of autistic persons become normal adults. The great majority are institutionalized or under custodial care. The average environment is well suited to the average human nervous system; autistic children have such unusual nervous systems that they need a special environment to learn and develop their full potential. My colleagues and I have created such an environment for very young autistic children in an experiment that filled all their waking hours for several years.

Autism was once considered a result of inadequate parenting; recent research has concentrated on neurological, genetic, and biochemical explanations. The aim of behavioral research, which began in the early 1960s, is to break down the daunting problem of autism into separate, measurable aspects of behavior and relate them to present conditions which, unlike such hypothetical causes as brain damage or parental rejection, can be experimentally altered. Behavioral symptoms of autism include excessive self-stimulation, self-injury, aggression, and deficits in language and social skills. Behaviorists believe that if these behaviors can be normalized, the emotional and social disabilities of autism will be dramatically reduced.

TREATING SELF-INJURY

One example is the treatment of self-injury. Most autistic children have severe tantrums, and a few injure themselves frightfully. I have seen children who have struck their heads with their fists thousands of times an hour, breaking the skin around their cheekbones and making their ears puff up as they bleed from the back of the head. In extreme cases they are tied to their beds and given tranquilizing drugs for years.

Early experiments indicated that self-injury persisted and became worse when it elicited affection and concern. Without this reward the rate of self-injury fell to near zero within days. A slap on the buttocks or other mild punishment would also quickly stop self-injury, if only temporarily. Autistic children could precisely detect subtle cues indicating the withdrawal of attention or concern, and control the situation (reinstate the reinforcing stimuli) by further self-injury. Self-injury that appeared full-blown in one situation would cease completely seconds later when a new situation was introduced. The rate of self-injury also depended on the child's capacity for appropriate alternative behavior. Some children, it was found, injured themselves repeatedly not to obtain rewards but to escape or avoid demands. In other words, this psychotic and bizarre behavior was socially determined and, in a sense, rational. Like any acquired response, it followed the laws of learning laid down by behavioral psychologists from Pavlov to Skinner. It was shaped and maintained by positive social reinforcement.

Several questions suggest themselves here. How can a slap on the buttocks be a punishment for a child whose self-inflicted pain is so much more severe? A study of Pavlov's may help to explain this. He repeatedly fed a dog while at the same time giving it a mild electric shock which he gradually intensified. Eventually the shock was strong enough to burn the dog's skin, but it continued to salivate in anticipation of food. Perhaps the biological properties of self-inflicted injury also change when it is systematically associated with affection. Another question is why attention seems to increase aggressive behavior in autistic children, although it may have little observable effect on them otherwise. Perhaps aggression has been more systematically reinforced in the past, because it demands more attention; or perhaps behavior that occurs earlier in a child's development (as aggression may) is more susceptible to reinforcement than behavior that is learned later.

BEHAVIORAL TRAINING

Not everything can be explained here; our conceptual system is inadequate. Nevertheless, self-directed aggression can be reduced fairly easily by behavioral treatment. Teaching alternative, socially appropriate behavior is much more difficult, but not impossible. Any child can be taught if an adult has patience and access to what the child wants. It is necessary to arrange situations in which positive behavior is rewarded while negative behavior is minimized. For example, we teach an autistic child to sit on request in the following way. The therapist/teacher sits in a chair, and the child stands between the adult's legs so as to reduce the possibility of escape (negative behavior). A chair is held in place directly behind the child. The adult instructs the child to sit. If the child fails to do so, the adult prompts him by pushing him backward. In such a situation, the child has no choice but to sit -- the positive behavior that we are trying to teach. The child is then immediately and lavishly rewarded with a favorite food or drink, applause, kisses, tickles, or anything else the child wishes. This is repeated five or six times a minute for up to two hours. (Very frequent repetition increases the rate of learning and also reduces autistic withdrawal.) Eventually physical prompting becomes less necessary, the situation becomes less structured and the child sits for longer periods of time. Other simple learning tasks include looking at someone or touching something on request. An added advantage of starting at an elementary level is that parents and others can also become teachers. Given this kind of training, autistic children very soon begin to improve and appear pleased with themselves. Their parents' joy and relief are even greater.

These simple tasks contain almost all the elements of effective behavioral training: unambiguous instructions, prompting, reinforcement (reward contingent on performance), shaping (gradually increasing response requirements), and discrimination training (learning when, and when not, to respond). In this way a child can gradually learn how to use the toilet and eat at a table, and later how to speak and play with toys and other children. A mute child can also be taught to speak and understand. Procedures have been devised by which autistic children can learn to imitate sounds and then words. They are rewarded for making sounds that increasingly approximate adult speech, and can then learn to use this newly acquired speech grammatically and meaningfully.

The first ten years of behavioral research were promising, but the shortcomings were also serious. Children did not achieve normal functioning, and their gains depended too much on the specific situations to which they were exposed in treatment; they soon relapsed when the treatment ended. My colleagues and I started the Young Autism Project in 1970. All children admitted were very young and had received an independent diagnosis of autism. They were assigned at random to one of two groups: an experimental group of 19 children who received intensive behavioral treatment, and a control group of 40 who had minimal or no treatment. Children in the experimental group were treated all their waking hours, seven days a week, 365 days a year, for two or more years. Their lives were totally restructured. Experienced staff members gave each child 40 hours a week of one-to-one treatment a week in the child's home. Parents were trained as therapists by working as apprentices to the staff. In the first year we aimed at reducing self-stimulatory and aggressive behavior, building compliance, teaching imitation, and establishing the beginnings of appropriate play with toys. In the second year, we taught them to use expressive and abstract language and play with other children. In the third year we emphasized reading, writing, arithmetic, the expression of feelings, and observational learning (learning by observing how other children learn). They were not placed in special education classes, which would have had the detrimental effect of exposing them to other autistic or disturbed children.

The control group with minimal treatment fared poorly. Their IQ scores were unchanged, and only one achieved normal functioning. Most were placed in classes for autistic and retarded children. By contrast, half of the children in the experimental group passed first grade in public school and attained average or above average scores on IQ tests. Forty percent passed first grade in classes for language-impaired children and attained IQ scores within the mildly retarded range. Only ten percent remained profoundly retarded and were placed in classes for the autistic and retarded. The children in the experimental group as a whole gained an average of more than 20 IQ points. In our most recent follow-up, we interviewed and tested the children in the experimental group, comparing them with normal children of the same age. Eight of them could not be distinguished from the normal children and were judged to have recovered fully. It is difficult to predict who will respond to treatment at the beginning, but we found that children who learn most quickly in the first three months continue to do so later.

The cost of one full-time special education teacher to provide this treatment would be about $25,000 per child per year; lifelong supervision and care for an autistic child costs more than a million dollars. For the minority who do not improve substantially with intensive behavioral training, a less demanding treatment that emphasizes self-help skills and community living seems more appropriate until we know more about how they learn.

Dr. Lovaas is Professor of Psychology at the University of California, Los Angeles.

President and Fellows of Harvard College, 1989 Reprinted with permission.

Internet Mental Health (www.mentalhealth.com) copyright 1995-1996 by Phillip W. Long, M.D.