Alone in the World

For Autistic Children, Relating to Others Is Life's Greatest Challenge

Peter is a beautiful 5-year-old boy with blond hair, blue eyes, and freckles. He looks like many other very cute kids. He is well coordinated, active, and agile. However, while Peter looks perfectly normal, it soon becomes apparent as you watch him that Peter does not behave like a typical child. He does not interact with the other children in his class, and in fact he avoids contact with them. He is not attached to his parents or anyone else, preferring to be alone. Rather than playing appropriately with toys, he puts them in his mouth or flaps them in front of his eyes. He does not communicate but instead parrots TV commercial jingles or bits of conversation he hears from others. He throws frequent and intense tantrums, often lasting over an hour and precipitated by nothing more severe than the discovery of a drawer left open, the disruption of a precise line of toy cars he has arranged by color, or the removal of one of the McDonald's mustard packets that he insists on carrying with him at all times. Sometimes during these tantrums Peter bangs his head against the floor or the wall or bites his hand. He has calluses on his hands from repeated biting. When not otherwise engaged, he will jump repeatedly while flapping his arms and whistling. The teachers in his school try a variety of techniques in an effort to help him. Understandably, his parents are immensely frustrated. They cannot reach their son emotionally despite endless attempts. Their lives are complicated further by the fact that they avoid taking him places because of his disruptive, bizarre, and embarrassing behavior. Their son has autism.

Autism is a severe form of psychopathology evident before the age of 3. It is a disorder characterized by a unique constellation of severe and pervasive behavioral deficits (e.g., lack of communication) and excesses (e.g., ritualistic and repetitive behaviors). Because of the extremely broad range of behaviors and abilities among people with autism, it is likely that what we call "autistic disorder" is really a diagnostic category made up of several as-yet-undetermined subgroups. The subgroups have many similar features, but future research may find that they have distinct causes. Therefore, "autism" is really short for "Autistic Spectrum Disorder," a term that is applied to all of the various subgroups. In turn, Autism Spectrum Disorder falls under an umbrella category of "pervasive developmental disorders." Although this article is limited to discussing autism, readers should be aware that there are three other pervasive developmental disorders that share some features with—but are distinct from—autism: Rett's Disorder, Child Disintegrative Disorder, and Asperger's Disorder.* To learn more about these disorders, visit the Autism Society of America's Web site at

Although we are much further along in our understanding of autism than we were even a few years ago, there is still a great deal we do not know. This poses a tremendous challenge to all those interested in, or responsible for, the care of autistic individuals. Primary among the things we do not yet know is what causes autism. What we have are many possible causes, and almost all of these have vocal proponents. Genetics, neurobiological factors, hormonal factors, viral infections, metabolic factors, birth complications, environmental toxins, and other causes have been put forward. At present, the possible causes with the most empirical support are genetics, neurobiological factors, and some viral infections. For example, we do know that identical twins are dramatically more likely to both have autism than are fraternal twins—but since both identical twins don't always have autism, the cause can't be purely genetic. What does not have any empirical support is the once widely held belief that parental psychopathology—specifically, a "refrigerator mother" who fails to provide the emotional warmth that her child needs—causes autism. Likewise, at this time, there is no scientific evidence supporting the theory that the measles, mumps, and rubella vaccine cause autism.1 So, although we do not yet know what does cause autism, the research is progressing as certain causes are ruled out and efforts become more focused on the most likely causes.

Judging by the many depictions of the disorder in movies, television, and print media, autism seems to hold a special fascination for the lay public. In many ways the attention autism gets is beneficial; general awareness makes early diagnosis and treatment more likely. But it also comes at a cost because misconceptions abound. This article aims to clear up those misconceptions by exploring the characteristics of autism and then briefly describing the main features of effective treatments.

I. Characteristics of Autism

In 1943, child psychiatrist Leo Kanner of Johns Hopkins University provided the first detailed account of what he called "autistic disturbances of affective contact." He described a group of 11 children who seemed quite similar to each other, but qualitatively different from children who were more adequately described by other clinical diagnoses.2 In this initial report, Kanner provided a richly detailed description of each child and in so doing gave us the first fascinating glimpse of what we now call autistic disorder or autism.

The central characteristic of these children, as described by Kanner, was what he called "extreme autistic aloneness," demonstrated by an inability to develop normal social relationships or relationships with the environment. Other main features of children with autism were: 1) a delay or failure to acquire speech; 2) the noncommunicative nature of speech if it did develop; 3) stereotyped and repetitive play activities; 4) a compulsive demand for the maintenance of "sameness" in the environment; 5) good memory for rote material; and, 6) lack of imagination. Given the fact that more than 50 years have passed, it is a testament to Kanner's skill as a perceptive observer that the major symptoms required for a diagnosis of autism remain basically unchanged.

Let us first look closely at the main behavioral characteristics associated with autism, keeping in mind that the severity of these symptoms varies widely among individuals. Not every affected child or adult exhibits all of these characteristics, and some of the characteristics are also noted in individuals who are not autistic. It is the unique constellation of symptoms that characterizes the disorder.

Deficits in Social Behavior and Attachment

Perhaps the hallmark feature of autism is the profound and pervasive deficit in social behavior and attachment. Children with autism often do not bond with their parents, do not play with other children, may ignore or avoid the social initiations of others, and prefer to be alone. It is not uncommon to hear a mother report that as an infant, her child did not hold up his arms in anticipation of being picked up, did not look at her when held, or was "stiff" or "rigid" to hold. The parent may describe the slightly older child as not wanting to be held, cuddled, or kissed, sometimes actively resisting or avoiding expressions of affection or other social overtures. Typically, the child is not upset when the parent leaves or particularly happy when the parent returns after an absence; he seems, in fact, not to notice. Children with autism usually do not come to the parent for comfort if frightened or injured, nor are they likely to be consoled by the parent's efforts to comfort them. Not surprisingly, they also typically fail to show empathy or to understand the feelings of others.

Consider Donnie, a 3-year-old child who was with his mother in a room full of toys. When his mother called to him, trying to get him to come to her or look at her, he persisted in ignoring her while repetitively lining up a toy Ping-Pong net along the edge of a table. He never acknowledged her presence until she finally touched his arm to get his attention. At this time he very purposefully walked away from her and pressed his face against the opposite wall in an apparent effort to shut her out completely. His almost total detachment and avoidance were striking to witness, but not unusual for a child with autism.

Children with autism may not engage in social eye contact. Sometimes they actively avoid such eye contact, or they seem to look "through" another person. Significantly, they fail to develop, or are delayed in developing, "joint attention"—that is, using their eye gaze to direct the attention of others (a skill that typical children develop by the time they are 15 to 18 months old). Joint attention is an important prerequisite for the development of more complex forms of communication and social interaction, and its absence is a significant feature of autism.

Autistic children are as unresponsive to their peers as to their parents. Other children are typically ignored or actively avoided. If the autistic child expresses interest in peer play, it is usually only to watch the activity without social interaction or reciprocation. And, if the child does initiate play with a peer, it often consists of a socially inappropriate overture such as scattering toys or saying something odd or irrelevant (for example, interrupting a game of catch by throwing the ball away while yelling, "I'll risk a thousand, Alex!").

Children with autism often fail to engage in imaginative, pretend, or sociodramatic play. Play that appears imaginative is often rigidly "scripted"; the child will repeat the same script over and over, with little or no variation. For example, a child who is given a toy car set complete with little people, a house, and other accessories, might do the following: Drive the car to the house, put two people in the car, drive to another house, and take the people out. This precise sequence of actions would be repeated in an identical manner again and again, with no changes or elaboration. Or perhaps the child likes to draw, but only draws the same picture every time.

Even high-functioning individuals with autism remain uninterested in establishing friendships, prefer being alone, fail to consider the interests of others, and may be unresponsive or totally oblivious to subtle social cues. For example, one 22-year-old autistic man who had just earned a master's degree in engineering would talk almost exclusively and at great length about bridges and elevators. The obvious boredom of his audience and repeated efforts to change the direction of the discussion went completely unheeded, and the pedantic monologue continued until he was pointedly asked to stop talking about bridges and elevators. At no time during this "conversation" did the young man ask anything about his conversational partner's interests or activities.

Deficits in Communication

Leo Kanner considered the delay or failure in the acquisition of language to be primary to the disorder, and this opinion is still held today. Approximately half of children with autism fail to develop functional speech—but only a very small percentage is completely nonverbal. Many autistic children (even some of those with functional speech) develop noncommunicative speech patterns that are qualitatively different from those of ordinary children or those with other specific language disorders.3 In addition, nonverbal communication may be absent; autistic children often do not shake their head to indicate "no" or nod to indicate "yes." They seldom wave "bye-bye," blow a kiss, or use other conventional social gestures. Similarly, they typically do not point to things in their environment to share an experience such as seeing a fire truck or airplane.

Some autistic children develop early speech using a few words or perhaps simple phrases, only to lose this speech at around 18 to 30 months of age. Thus the children may use words or phrases like "mama," "cookie," "go car," or idiosyncratic phrases such as words from an amusement park ride, but suddenly (usually in a matter of days or weeks) lose the acquired speech and fail to progress linguistically. Often parents report that their child said a word or phrase very clearly on one occasion, never to do so again.

Autistic children who do speak typically exhibit distinctively pathological speech patterns. Many display echolalia, the repetition of words or phases spoken by others. Echolalia can be immediate or delayed. In immediate echolalia, the child parrots what has just been heard. For example, in response to the question, "Where is your jacket, Susan?" the child responds, "Where is your jacket, Susan?" Immediate echolalia is often preceded by a question, request, or statement that the child finds incomprehensible.4 For example, a child asked to touch his head may do so. Yet, if asked to "indicate your cranium" he would likely echo "indicate your cranium" since he does not understand the command.

In delayed echolalia, the individual repeats speech that has been uttered a few minutes, hours, days, or even years in the past. Because the speech is remote in time, it is most often contextually inappropriate and may sound quite bizarre. Sometimes the original speech is identifiable, as when the child repeats a phrase from a TV commercial, an instruction heard from his teacher at school, or a parental reprimand. Most often this speech is noncommunicative: The children do not comprehend what they are saying, nor do they use this speech functionally in their environment. The conditions leading to delayed echolalia are not well understood. Anecdotal reports suggest that it is more likely to occur under conditions of high arousal. For example, I knew an autistic child, Bobby, who was very frightened of dogs. When confronted by a dog one day, the terrified child blanched, backed away, and loudly declared: "It's not going to hurt you, Bobby" and "Pet the nice doggy, Bobby." It is quite likely that these are reassuring phrases he had heard in similar circumstances when he had been frightened or aroused. There are other times, however, when the reasons for the specific delayed echolalic response are unclear. Thus on another occasion when Bobby was confronted by a dog he exclaimed, "It's not a glass paperweight!" and "I said get to bed right now!"

Probably related to echolalia is the often observed pattern of pronominal reversal in which the individual refers to himself as you or by his name rather than I or me. Children who use their echolalia to communicate may produce statements such as, "Do you want to go outside?" or "Do you want a cookie?" to indicate that these are their wishes. These are direct echoes of statements they have heard on occasions when they have been given what they wanted.

Idiosyncratic speech and neologisms (i.e., made-up words) are also frequently noted in these individuals. Idiosyncratic speech occurs when a person consistently uses an unusual word or phrase to express a label or concept. One child consistently referred to a particular mechanical toy as a "Cow says." When activated in a particular way, this toy produced a prerecorded voice that said, "The cow says moo." Another child referred to any reel-to-reel tape recorder as a "self-destruct in five seconds" (obviously related to the television program Mission: Impossible). Neologisms occur when an individual consistently uses a novel, made-up word or phrase to express a label or concept. One child used the neologism "pling" to refer to any pencil.

Communicative language is often limited to the here-and-now, with distinct difficulty shown in communicating past, future, or hypothetical events. It may also be restricted to the very literal; analogies, metaphors, and humor are essentially incomprehensible. Literalness can interfere with even the simplest of interactions. I remember one autistic child, Danny, who was receiving treatment in our program. An undergraduate student named Rick worked with Danny for several months. For some reason Danny kept referring to Rick as "Poster" despite numerous corrections from Rick. Finally one day Rick became frustrated and told the child very intently, "Danny, my name is NOT POSTER!" The result (as one might have guessed) was that Danny referred to him as "Not Poster" ever after.

The speech of even high-functioning and linguistically skilled individuals with autism is often devoid of emotion, abstraction, or imagination. Attempts to elicit statements of feelings are typically met with noncommittal answers, such as "It was good," "It was bad," and the perennial favorite, "I don't know." Sometimes even the most direct query will elicit a highly concrete and odd response. For example, one adult with autism was asked how he felt when his mother died. He responded, "She was 68." It is startling to hear conversation that is based almost completely on the concrete, lacking color or emotion, yet such is the case with many of these individuals.

Restricted, Repetitive, and Stereotyped Patterns of Behavior

The behavior of many children with autism is compulsive, ritualistic, repetitious, obsessive, and stereotyped. It may involve gross- and fine-motor movements or highly sophisticated verbal rituals. At the gross-motor level, one often sees rhythmic body rocking, rocking from foot to foot, head bobbing or weaving, arm and/or hand flapping, jumping, spinning, pacing, or posturing. At the fine-motor level, one might observe finger wiggling, gazing at the cupped hand at the side of the face, grimacing, eye crossing, saliva swishing, or hair twirling. Often objects are incorporated in these movements, as when the child repeatedly taps something, twirls saucepan lids or pieces of string, flips the pages of a book, waves objects in front of her face, or spins the wheels of a toy car. Repetitive vocalizations of nonsense sound patterns, particular words or phrases, or snippets of songs are also common. These behaviors are typically referred to either as "self-stimulation," to connote that the function of the behavior is to provide sensory stimulation, or as "stereotype," to reflect the repetitive and stereotyped nature of the activity.

Self-stimulation presents several problems. First, many of these children spend a great deal of their time in such self-stimulatory activity. While the amount of time may vary, some children spend most of their waking hours so engaged, often to the exclusion of almost everything else. There is substantial research that suggests self-stimulation interferes with responsiveness to the environment and with learning. Accordingly, a good deal of research has focused on the nature of self-stimulation and how it may be eliminated. Unfortunately, it remains one of the most difficult and poorly understood behaviors observed in autistic individuals.

Other compulsive and ritualistic behaviors are seen as well. We observe children who compulsively line up objects, follow patterns in floor tiles or wallpaper, or build the identical block form repeatedly. Rather than playing with toy cars in the usual way, a child with autism may arrange them in perfect rows, categorized by color, and all facing the same direction. Any disruption of this arrangement by adding, subtracting, or rearranging the cars is met with distress (often a tantrum). Autistic children may insist on collecting and carrying particular objects (such as small rocks, sticks, a piece of cloth, or a particular toy) at all times, and strongly resist any attempt to remove them. Children I have known have insisted on carrying items such as fast-food condiment packets, Tinker Toy dowels (one in each hand at all times), leaves, pages from a phone book, and bottle caps. I remember one little boy who was not attached to a teddy bear or blanket but rather took a hand-held vacuum cleaner to bed with him every night, and a girl who insisted on carrying a stop sign with her everywhere she went.

Children with autism may strongly resist changes in the physical environment, daily routine, or familiar routes of travel. Changes in furniture arrangements, for example, are often noted immediately, with the child attempting to return the furniture to its original position. If this is impossible, the child may be quite upset until the disruption is corrected. Sometimes even the most minute detail is detected, as when the child notices that a particular figurine has been moved a few inches on a table or a package of cookies has been placed in the wrong position on a pantry shelf. Alterations to expected schedules are not well tolerated, and maintaining routines becomes imperative for parents. One mother reported receiving several traffic citations for failing to stop at a stop sign in her neighborhood. The stop sign had only recently been placed at that location, but since her autistic son was used to maintaining speed there, any stop on her part was followed by a screaming tantrum.

Many autistic children have compulsive food rituals: They will only eat one or two specific foods, or may only eat foods of a particular color or only if the food is placed in a specific section of a plate. Ritualistic preoccupations are also observed when the children memorize information that may be trivial or of little functional value, such as train or bus schedules, TV schedule grids, maps, consecutive numbers in a telephone book, or dates.

Linguistically advanced individuals with autism may exhibit compulsive behaviors when they engage in conversations. Repetitive questions are common, as is the insistence that the listener respond in a particular manner or provide a specific answer. When a person with autism engages in conversation involving a favored topic, it is extremely difficult to divert the direction of the conversation (as with the engineer who was preoccupied with elevators and bridges). I remember one autistic child who was completely obsessed with Volvo automobiles. He carried around a Volvo brochure describing the models, looked for Volvos on the street, and would work very hard for the reward of visiting a Volvo dealership and being allowed to walk down the aisles of cars.

Abnormalities in Response to the Physical Environment

Children with autism are often described as showing deficient or unusual responsiveness to their sensory environment. A child may be very unresponsive to loud noises, the calling of her name, or other auditory stimuli. Similarly, the child may be unresponsive to visual stimuli; she may not respond to people entering a room, nor track the progress of people or things across her visual field. This is not true sensory impairment, however. The child who does not respond to the calling of her name or a loud noise may respond to the crinkle of a candy wrapper or may repeat commercial jingles. The child who does not respond to people coming and going in his environment will be transfixed by watching pieces of lint fall through a beam of light.

Unusual sensory interests are also frequently observed. Autistic children may seek to run their hands across certain textures, mouth or lick objects, sniff people or objects, or put their ear against stereo speakers. One little girl I knew would go up to unfamiliar adult men, raise up their pants cuff, and feel their socks, much to the embarrassment of her parents. These children may gaze intently at spinning objects such as flushing toilets, tops, washing machines, and fans.

Autistic children may also be over- or under-responsive to touch, pain, or temperature. An attenuated response to pain is often reported, as when the child falls and skins his knee or bumps his head and fails to cry (or, of course, to seek comfort from a parent). Often the child's response to such an injury is to get up and continue with what he was doing, to the amazement of those watching. In contrast to this under-responsiveness, some children display a hypersensitivity to physical contact with other people and become quite agitated when touched.

Abnormalities of Affect

The emotions expressed by individuals with autism are frequently odd. Their emotional responses may be excessive and exaggerated or relatively stable and minimal ("flat"). Some autistic children shift rapidly between hysterical laughter and inconsolable sobbing with no apparent provocation. The affect may be quite mismatched to the situation, as when the child laughs when someone else is hurt or cries when given a birthday present. The children displaying flat affect may seem to be "cruising in neutral" despite the varying conditions in the environment. This flat affect is typically accompanied by little variation in facial expression.

Irrational fears are also frequently noted, and often these are related to the children's demand for sameness. I have known children who have been intensely frightened by balloons, felt, tortillas, ferns, yellow ducks, sesame seed hamburger buns (regular buns were fine), Bill Cosby, the Channel 7 Eyewitness News, and the theme song from the television show Family Ties.

Abnormalities in Intellectual Functioning

Despite impressive feats like memorizing bus schedules, the majority of autistic children are cognitively impaired to some extent, most to a serious degree. In fact, estimates from various studies agree that approximately 75 percent of autistic children are mentally retarded.5 One of the main distinctions between children with autism and children whose primary diagnosis is mental retardation is the fairly distinctive profile exhibited by autistic children on subtests of intellectual ability. While children with mental retardation tend to score at low levels across all areas, children with autism typically show a distinctive pattern: They tend to score poorly on assessments of symbolic thought (such as language) and abstract reasoning, and to score higher on assessments of visual-spatial ability and rote memory. In addition, many of these children display isolated, and usually quite narrow, areas of exceptional skill. In a minority of cases these skills may be at the savant level,6 but it is usually the case that the skill is at a normal or near-normal level for the child's age; it appears exceptional because of the child's low level of ability in other areas. These special abilities most commonly lie in the areas of rote memory, mathematical calculations, mechanical skills, or musical ability.

Not surprisingly, the true savant skills seen in a small percentage of autistic individuals have attracted a great deal of attention. Consider the movie Rain Man, which depicts a man with autism who is adept at performing complicated mathematical calculations in his head, card counting in Las Vegas, and other impressive counting feats. This depiction is not an exaggeration; indeed, the actor Dustin Hoffman based his characterization on a compilation of three known autistic savants. Other such individuals have shown their ability to rapidly complete complicated jigsaw puzzles (picture side down), memorize schedules from TV Guide, or assemble a complex mechanical apparatus.

These special skill areas tend to be narrow and isolated, and completely nonpredictive of the child's overall level of functioning. For example, one 7-year-old musical savant could hear a melody once and subsequently play the melody on any of six musical instruments. He could also instantly play complex harmonies to the melody. However, this same child was not toilet-trained, could not respond to a simple question like, "What's your name?" and could not respond to an instruction such as, "Close the door." Despite his immense musical talent, he was functionally mentally retarded and required constant supervision.

II. Treating Autism

Without knowing what causes autism, we have little hope of finding a cure. But decades of research have resulted in some beneficial treatments. Most professionals in the field would agree that today the treatment of choice is based on the behavioral model of learning. In fact, behavioral treatment is the only treatment that has been empirically demonstrated to be effective for children with autism.7 Fortunately, it is also fairly straightforward to implement. Teachers, paraprofessionals, parents—even siblings as young as 6 years old—hae been successfully trained to use behavioral treatment with autistic youth.

In its original form, behavioral treatment is often referred to as "behavior modification" or "behavior therapy," although now there are many variations that go by different names. All forms of behavioral treatment are derived from the experimental analysis of behavior, which is the science devoted to understanding the laws by which the environment affects behavior. Identification of these laws allows for behavior to be predicted and controlled—and thus changed. The study and application of these laws to socially significant problems is referred to as applied behavior analysis, and we owe much to this field of science for the development of effective interventions for the autistic population.

The original forms of behavioral treatment were based closely on the principles of learning that had been established after many years of work in animal laboratories. Animals such as rats and pigeons were trained to perform easily quantifiable behaviors such as pecking a lighted key or pressing a bar. The animals were rewarded with bits of food, and the schedules of food delivery (schedules of reinforcement) were carefully manipulated to elicit desired response patterns. Some early research with autistic children did involve teaching them to press a lever for bits of food or candy. It makes us cringe today—but it did demonstrate the potential for developing effective behavioral treatments.

Discrete Trial Training

Although they focus on more elaborate behavioral repertoires and functional curricula (for example, linguistic, social, and academic target behaviors), more recent adaptations of behavioral techniques continue to adhere to a rather strict learning format in which autistic children learn to discriminate between and respond appropriately to various stimuli (e.g., questions, requests, or commands). Typically, the children are taught through a series of discrete teaching "trials" that include three components. First, an instruction or question is presented. In the early phases of training, this may be accompanied by a prompt to guide the child's response. The second component is the child's response, incorrect or correct, or perhaps lack of response. Third, depending on the child's behavior, a consequence is presented. These consequences take a variety of forms, and their nature is determined by the desired effect. If the teacher wishes to encourage the response because it is correct, a positive reinforcer such as food, praise, or access to a favored toy is presented. To discourage an incorrect response, the teacher may present a "punisher" (for example, saying "no" or frowning) or not present any consequence (that is, ignore it).

Let's look at an example of teaching a child, Carolyn, to say her name. The question "What is your name?" is presented. She either answers correctly with "Carolyn" (or an approximation thereof), provides an incorrect response (such as "geegeegee"), or remains silent. The trainer then provides the appropriate consequence: a piece of cookie and praise for the correct response, "Carolyn"; a "no" for "geegeegee"; or no consequence for silence. These three-part trials are presented in a series of successive blocks, and the child's progress is calculated by determining the percentage of correct responses within a block of 10 or 20 trials. If inadequate progress is found, various things may be altered, including the specifics of the instruction, the addition of prompts, the nature of the consequence, and so on. In addition, the target behavior may be broken down into smaller components to make the task easier.

This type of highly controlled treatment is referred to as "discrete trial training" (DTT). It revolutionized the treatment of children with autism. In fact, one can argue that the development and refinement of DTT provided the first real treatment regimen for teaching simple, and later more complex, skills to autistic children. Its contribution to the treatment of autism, particularly in the early 1960s and 1970s, cannot be overstated. After this treatment had been used for many years, however, some troubling trends emerged as limitations to the generalization of treatment effects became apparent. People noted that the positive effects of the treatment did not always hold up over time as well as one would hope, and the positive responses did not always appear in different settings or around different people. Further, the training itself was often not very efficient in that treatment effects seemed to be specific to the behaviors taught rather than spreading to other behaviors.

Problems with generalizing across settings and people are illustrated with the cases of Kenny and Freddy. Kenny had had many months of intensive DTT in his home; his mother proudly showed me a notebook containing descriptions of specific responses Kenny had learned. I noted that he had learned to say his address and decided to see how he would respond to me. After ensuring that he was paying attention, I asked, "Where do you live?" Kenny stared at me and said nothing. After my sixth or seventh unsuccessful attempt to elicit his address, his mother tapped me on the shoulder and said, "You have to say, ‘What is your address?'" Apparently, this latter phrase was used in Kenny's DTT training, and I was presenting a different question. Obviously, this is a severe problem. If Kenny became lost, a police officer might not ask the question in the specific way Kenny had been taught.

Freddy's problem became clear at a holiday party for families participating in our research program. There was a bowl of punch on the table and Freddy wanted some. We knew Freddy's parents had taught him to appropriately request what he wanted, and thus we were surprised to see him grab a cup, wave it repeatedly over the punch bowl, and vocalize gibberish. He kept doing so until his mother, who had been out of sight, appeared and said, "Freddy, you say it right!" Freddy saw her and said, "Punch, please." It appeared that the presence of his mother was necessary for Freddy to use his speech appropriately; the behavior did not sufficiently generalize to other people in his environment.

While traditional discrete trial training continues to have a strong following, the limitations to its effectiveness have led others to move away from the highly structured nature of DTT to other forms of behavioral treatment that hold promise for addressing these limitations. One strategy is to retain the structured DTT format, but to systematically address each of the areas found to be problematic.8 To enhance stimulus generalization, for example, the child is taught a behavior through the use of multiple stimuli including different task materials, settings, and people. For the child who would only say his address if presented with a specific form of question, this type of problem could be prevented if, during the training, the child were taught to say his address in response to a variety of questions, such as "What is your address?" "Where do you live?" "Where is your house?" and the like.

Today, the best DTT treatment programs incorporate these generalization-enhancing strategies and, as a result, the treatment overall is more effective and less highly structured. However, even though these improved treatment programs are less formally structured, the fact remains that they may still be rather difficult for nonprofessionals (such as parents or peers) to learn and implement, and a good deal of training is required for these nonprofessionals (as well as the professionals) to become effective treatment providers for autistic children.

Naturalistic Treatment Strategies

Another more recent approach to treatment uses naturalistic strategies that allow the child with autism to learn behaviors in their usual context under more typical conditions. The idea behind the development of these naturalistic strategies stems from the view that the reported lack of generalization and maintenance of treatment may be directly related to the specificity and artificiality that characterize the traditional DTT approach.

A brief description of one naturalistic strategy, Pivotal Response Training (PRT), will illustrate how naturalistic strategies work and how they differ from DTT. One difference is the nature of the training stimuli used. Let's say we want to teach the child the concept of color. We could teach "yellow" by using highly structured teaching involving yellow cards, yellow blocks, and so on. Indeed, this is probably how it would be done using DTT techniques. However, the color concept can also be taught in a format that involves the natural contexts in which colors are found. Thus, one may walk through a park and teach that a car is yellow, a bench is yellow, a leaf may be yellow, and so on. Because the yellow stimuli are observed in naturally occurring circumstances, generalization of the concept is more likely to occur without specific training for generalization.

Another difference between DTT and PRT is the nature of the consequences used in training. Autistic children are notorious for being difficult to motivate. DTT overcomes this problem by identifying consequences that are motivating to the children, including positive reinforcers such as food, drinks, and access to favored toys. However, since the "real world" typically does not provide such consequences, it is not surprising that behavioral gains are not generalized or maintained well with this sort of training. No one is walking around giving us pieces of candy or sips of a soft drink for being social or talking as we go about our daily lives. Rather, our social skills and speech are maintained by the natural consequences directly associated with our behavior. If I want a book at the library and say to the librarian "I want this book," it is not because I expect, or want, him to say "Good talking" and give me a piece of candy. I want the book, and it is the delivery of the book that maintains my speech in this context.

Similarly, PRT employs consequences that are directly related to the child's behavior so that these natural consequences will maintain the behavior and assist in its generalization. For instance, if I am teaching the child using a toy car she has chosen, I know that the car is a direct reinforcer for her speech. If she says she wants to "roll" the car, then the consequence (reinforcer) for saying "roll" is being allowed to roll the car. Rolling the car when she says "roll" is a much more natural consequence than a piece of candy and saying "Good talking." This direct reinforcement is how typical children learn to use language in their environment.

Another approach used in PRT to increase motivation to learn is to give the child a choice about the nature of the teaching interaction. It is common in DTT procedures for the therapist to decide what skill will be addressed, what training stimuli will be used, and what the available consequences will be. In naturalistic strategies such as PRT, the child is allowed to make these choices. If we present a variety of toys, edibles, and similar things to the child and ask her to choose one, we can be fairly confident that the child is interested in (that is, is motivated for) the chosen item. This means that the child is likely to be willing, perhaps even eager, to work for that item and that the item is a powerful reinforcer.

Another difference between DTT and naturalistic strategies such as PRT is the nature of the response required for a reward. Typically in DTT, a specified response is designated as correct, and only that response, or responses at least as good as a previous response, are reinforced. This can lead to frustration on the part of the child because, unless the training is conducted very carefully, the child will make many errors. Under such conditions it is not unlikely that the child will give up and stop responding, or perhaps have a tantrum. PRT avoids this problem by providing reinforcement not just for correct responses or even near-correct responses, but for any reasonable attempt to respond. This means that the child obtains access to the reinforcer for trying, not just for correct responding. If trying is reinforced, we will get more trying, and with more trying there are more opportunities for teaching.

It is important to emphasize how the teaching environment is structured in naturalistic strategies. The setting is carefully arranged to encourage and promote language, play, and other activities that may be naturally rewarding to the child. Yet, the learning challenges allow for instruction. For example, highly desirable toys are placed out of reach so the child has to use language to request them, brightly colored toys are made available if teaching colors is a goal, and lids are placed tightly on containers of preferred toys so the child learns to request help. It is also important to point out that while naturalistic strategies do not involve the "drill" type of repetitive trials that one finds in DTT, many trials may be presented in a short period of time, but in a more naturalistic manner. For example, turn-taking between the child and therapist may consist of short turns with lots of give and take, or the child may have to request a cookie several times because only part of the cookie is given each time.

One last difference between naturalistic strategies and DTT is that naturalistic strategies appear to be more enjoyable for the children. Researchers have found that autistic children who were being taught through naturalistic strategies were rated as having more positive affect than children being taught through DTT.9 Further, when the number of escape and avoidance behaviors such as crying and attempts to leave the teaching situation were recorded, there were significantly fewer such behaviors during the naturalistic teaching.

*  *  *

As I wrap up this description of behavioral treatments, readers may be tempted to conclude that naturalistic strategies are always the best. Let me caution against any firm conclusions: No single treatment can claim to be substantially effective for all children with autism. In fact, no single form of treatment can claim to be very effective (defined as truly substantial clinical improvement) for more than 50 to 70 percent of children with autism. The fact that there is such variability in treatment outcomes tells us that other factors aside from the choice of treatment are important in determining treatment effectiveness. All of the behavioral interventions described here are effective. So, the question should be which behavioral treatment is best for which child, under which conditions, and at which point in time?

While my fellow researchers and I can't name one best treatment, we can say that the best time to begin treatment is early. Autism is a progressive disorder in that early deficits in social responsiveness and communication have a seriously negative effect on the development of subsequent important behaviors. A child who is not socially engaged and does not learn to communicate effectively is at a definite disadvantage in learning more complex and subtle social and communicative skills. Moreover, ritualistic behaviors that autistic children spend a lot of time performing not only interfere with learning appropriate behavior, but also serve to stigmatize the child and thus affect how the child's social world responds to her. If we can begin treatment when the child is just 2–3 years old, we may be able to avoid many of the behavioral and communication problems that we typically see as the child gets older.

Laura Schreibman is distinguished professor of psychology and director of the Autism Research Program at the University of California at San Diego. This article is adapted and reprinted with permission from The Science and Fiction of Autism, by Laura Schreibman, Harvard University Press, © 2005.

*There is ongoing debate among researchers regarding the validity of a diagnostic distinction between Asperger's Disorder and high-functioning autism, both of which are distinguished from "regular" autism by the development of appropriate language and the lack of cognitive impairment. For parents and educators, the distinction is not very important: The two diagnoses are similar enough that they necessitate the same kinds of interventions. (back to article)

The highly structured, repetitive-practice nature of DTT goes well beyond the three major components I described. There are important rules regarding the specifics of stimulus presentation, timing, successive approximations of correct responses, and both the nature and schedule of consequences. Adherence to these rules is very important for delivering the treatment effectively. (back to article)


1. Dales, L., Hammer, S.J., and Smith, N.J. (2001). "Time trends in Autism and in MMR Immunization Coverage in California," Journal of the American Medical Association, 285, 1183–1185. See Fombonne, E. and Cook, Jr., E.H. (2003). "MMR and Autistic Enterocolitis: Consistent Epidemiological Failure to Find an Association," Molecular Psychiatry, 8, 113–134.

2. Kanner, L. (1943). "Autistic Disturbances of Affective Contact," Nervous Child, 2, 217–250. Reprinted in Leo Kanner (1973), Childhood Psychosis: Initial Studies and New Insights, Washington, D.C.:V. H. Winston and Sons.

3. Schreibman, L. (1988). Autism. Newbury Park:Sage Publications.

4. See, for example, Carr, E.G., Schreibman, L., and Lovaas, O.I. (1975). "Control of Echolalic Speech in Psychotic Children," Journal of Abnormal Child Psychology, 3, 331–351; Schreibman, L. and Carr, E.G. (1978). "Elimination of Echolalic Responding to Questions Through the Training of a Generalized Verbal Response," Journal of Applied Behavior Analysis, 11, 453–463; Prizant, B.M. and Duchan, J.F. (1981). "The Function of Immediate Echolalia in Autistic Children," Journal of Speech and Hearing Disorders, 46, 241–249.

5. National Research Council (2001). Educating Children With Autism, Catherine Lord and James P. McGee (Eds.), Committee on Educational Interventions for Children with Autism, Division of Behavioral and Social Sciences and Education, Washington, D.C.: National Academy Press.

6. Treffert, D.A. (1988). "The Idiot Savant: A Review of the Syndrome," American Journal of Psychiatry, 145, 563–572.

7. National Research Council (2001). Educating Children with Autism, Committee on Educational Interventions for Children with Autism, Catherine Lord and James P. McGee (Eds.), Division of Behavioral and Social Sciences and Education, Washington, D.C.:National Academy Press.

8. See Stokes, T.F. and Baer, D.M. (1977). "An Implicit Technology of Generalization," Journal of Behavior Analysis, 10, 349–367.

9. Koegel, R.L., Koegel, L.K., and Surratt, A. (1992). "Language Intervention and Disruptive Behavior in Preschool Children with Autism," Journal of Autism and Developmental Disorders, 22, 141–153; Koegel, R.L., O'Dell, M., and Dunlap, G. (1988). "Producing Speech Use in Nonverbal Autistic Children by Reinforcing Attempts," Journal of Autism and Developmental Disorders, 18, 525–538.

Related Articles

Alone in the World
For Autistic Children, Relating to Others Is Life's Greatest Challenge
By Laura Schreibman

Is Autism on the Rise?

Educating Autistic Children
By Aubyn Stahmer and Laura Schreibman

American Educator, Summer 2006