Sally Rogers: Intensive Autism Treatment the Denver Way

Interview with Sally Rogers

By Martin Brynskov and Henning Just [1]


Also available in


The psychologist Sally Rogers is one of the world's leading researchers when it comes to autism treatment. Since the beginning of the 1980's, she and her team have been developing a successful treatment program for children with autism called The Denver Model. In some respects it overlaps with the traditional Lovaas UCLA Model but in others it has some significant new features. We met Sally Rogers for a talk about early and intensive behavioral treatment of autism - both as seen from a practical everyday perspective and from a research point of view.

Could you tell us, just briefly, about your background, your interests and your work at the M.I.N.D. Institute?

Sure. I'm a developmental psychologist. I did my Ph.D. with specialization in mental retardation and developmental disabilities, so I have always done this kind of work, and I have always enjoyed this kind of work.

I have also always worked with young children, as well as adults, so I have always had this mixed background of very young children and people who were more mature. So even though I didn't start off with a specific focus on autism, I had always had interest in autism, I had always worked with some children and adults with autism, and with lots of adults and children with mental retardation and other developmental delays.

So, then in 1981 I received a grant to start a preschool program for children with autism, and that's when I got very specialized into autism.

So you both have a theoretical and a practical background.

Yes. I've been developing and using our approach since 1981. Our approach is called 'the Denver model', and we've been publishing on the Denver model and teaching it and trying to refine the method ever since then.


We published four outcome papers in the 80's (Rogers et al., 1986; Rogers et al., 1987; Rogers and Lewis, 1989; Rogers and DiLalla, 1991). They talked about the children's progress in the group model. We currently have two studies going on the rest of the model - one on the 1:1 teaching routines and one on the integrated preschool model. We need to finish those studies, analyze the data, and write them up.

Other Research

In addition to the clinical work, the treatment work, I have several funded studies. We've been looking at different aspects of autism, and I'm particularly interested early developmental processes and how autism takes on its form across the first five years. So we've been studying a very large group of two-year-olds in Denver with autism, Fragile X, Down syndrome, mental retardation and then one-year-olds with typical development. We've been following those children now for the past five years. We're looking at a wide variety of measures.

The Denver Model

Could you briefly describe the elements of the Denver model?

It's a developmental approach which has two dual foci, one on intensive teaching and the other on developing the social-communicative skills that are so affected by autism. We believe that development in early autism is somewhat plastic, and that much of what seems to be the overall handicap in autism is kind of secondary to an initial, probably less pervasive impairment. The initial impaired process, the disruption of social communicative development creates a secondary set of processes, like the exaggerated interest in objects and repetitive patterns, and that if we begin very early to focus on the social communicative processes, we can prevent some of the cascading effects of autism.

So, intensive teaching and intensive focus on social communicative development and skill building are the main emphases of our treatment approach. We understand that social communicative development develops from emotional relatedness, and so, side by side with intensive teaching is this emphasis on affective connection, relationship building and understanding communication as involving an emotional exchange between people.

So these things are done side by side in the Denver model. They alternate back and forth, literally.


How is this the Denver model carried out in practice? In what settings?

It can be done in many ways. We no longer have our center-based program, because we'd started to use more inclusive settings, and I found that just so much more viable. It's such a richer environment for the children that we eventually moved away from our center based teaching. So, all of the children in the Denver model preschool program are in interventions in typical preschools. Generally it's a combination of typical preschool group half a day, intensive teaching half a day.

Is it one-on-one or group based?

Intensive teaching at home is one-on-one. In the inclusive classroom the child is a part of group activities, but his or her teaching is being carried out directly by an adult inside the group activity. But the support of the child is embedded in the group. The kids are learning, they're being taught all the time, but they are inside the group. The teaching is coming from the main teacher or adult who is leading the activity. They're not separate. They don't have a shadow that they are aware of - there is somebody shadowing them. They are in a normal preschool with age-typical peers.

So it is one child with autism?

Hopefully. That's the right way to do it.

Is it common to have an inclusive setting in the USA?

It depends on what part of the country you're in. In the western part inclusive education is very common. Along the eastern sea border it's not so common. There's a history of special schools along the east coast.

Family in Focus

So that's kind of the underpinnings of the Denver model: the dual approach, very family based. Families choose the objectives, they decide what's important. We follow the family's lead. The family's part of the team, every meeting. So it's very family-focused. And it's very individualized. But we do have a curriculum. There's a very specific teaching approach, we have a treatment fidelity measure so we can measure the quality of it.


What is your opinion about Ole Ivar Lovaas' study and the replication sites?

I think that Lovaas's group have provided the best studies of the method, and we have now two replications of his work (Lovaas, 1987; Smith et al., 2000; Eikeseth et al., 2002). And I think that all the Lovaas studies consistently show that children who received the treatment did better than the comparison groups. The degree of improvement varies from study to study, and only one of those has randomized assignment to groups. And it hasn't been compared to any other specific method, so we don't have comparative data, but I feel that it's established that children who receive that treatment do better than those who receive minimal treatment.

However, those treatments were developed in the 60's, and there's been quite a bit of modernization of Applied Behavior Analysis since Dr Lovaas developed that technique, and my guess is that the more modern approaches to ABA may be even more effective than the mass trial approach that he used, but it hasn't been studied in the same way. It's been studied in small studies.

Comparing Treatments

What is your comment on the Norwegian study (Eikeseth et al., 2002) comparing ABA with an eclectic approach?

It's interesting. The score differences were non-significant, but the change scores were significant.

What do you make of that?

I think it's real. I think the reason that the standard score differences are non-significant is probably because the groups were so small. It's got a lot of variability. There were 8 [ABA] and 11 [eclectic], really tiny groups. But I would expect that the differences are there. What is interesting in that study is that the groups are equated for number of hours and ratio, and that is the first time that has been done. But it is still not comparing one defined treatment - with fidelity and curriculum - with another defined treatment. We still don't have that. And also, the Eikeseth group also had parent training, and there wasn't parent training in the control group. As I said, I believe the data that show that Lovaas's way of teaching children is effective in accelerating improvement. What we need now are studies that compare other well-developed and well-defined treatment approaches to Lovaas's approach. We need to understand which children respond better to which treatments, or approaches, and why. When we start to understand that, then we will understand much more about individualizing treatments to get the best outcomes for all children.

I don't think we will ever be able to say that one treatment flatly is better than another for every child with autism. You know, it is always a question of, what are the most powerful treatments and what do they do the best and for which children. And we're a long way from knowing that.


You talk about 'well-defined programs'. But one could say that there is a tendency to join many different techniques in an eclectic way that doesn't necessarily work.

I agree with you on that. Just saying that a program is eclectic, using any teaching technique with the children that one has learned how to do - I don't think there is any reason to think that is particularly useful. And I think it does result in better treatment if there's a high level of expertise with a certain method. But I do think that for the benefit of the children it is probably the best to have a couple of people with different method expertise together. Because you can see the hole in another person's method, and you may have ways to fill them in.

There is a very skilled Lovaas therapist in Denver, and I have always enjoyed being on a team with her and her patients. And there could come up times in which - this child that wasn't moving in this one particular area, and I had treated that before and I had an idea, and so I could bring that to it. It was not a technique that she had done before. And vice versa - she had experiences with interventions that I had not used before, and I learned those from her. Two heads are better than one.

So I think, you know, if there is only one method developed to expertise, you are probably not individualizing as much as you can for every child.

But I agree with you. Just having a little bit of knowledge of lots of techniques is not enough.

Working together in practice

Then how do you bridge the gap between methods in practice?

In Denver I had different people on my team who had gotten very different kinds of training. My occupational therapist has been to all kinds of ABA training, and she's amazingly well trained in ABA - I mean, she really has mastered that. She is the best person I have for maladaptive behavior. My speech and language person, she knows the communication work, and she has recently been trained in the PROMPT method, [2] which is a method to treat oral-motor dyspraxia. So that's her area of expertise. I have expertise in imitation and symbolic play training. So each of us has mastered a different set of methods. Another woman is better with the older children and the high functioning children, which I'm not so good at. And she's really good at the pragmatics and social communication and social groups for very high functioning children. So each person on my team, we all do the Denver method, and we all deliver it similarly to young children with autism. But each person on the team has an area of expertise that has been developed and that we don't have overlap in. So, when we use each other we can really add something. All of us know each child. One person is the main therapist, but the other disciplines see that child every now and then and they are there, so when one of us is having a problem with some aspect of a child's treatment, we can pull the others in for help and new ideas. That's what really matters.

Team meetings

Is that a supervising team, or are you actually participating in the treatment meeting with the child present?

In the regular meetings of the treatment team, we are in the treatment meetings with the child present. Each person in the meeting does some treatment, parents, assistants, therapists. If the child goes to school, then the team meetings include all the teaching staff, as well as the supervising person.

How often do you have team meetings?

It depends on where the child is in the program. In the beginning you want to meet every week, and then, if the team is very mature and the child is moving along well, we taper meetings off to every couple of weeks. It depends.

Social interaction

You have been focusing much of your research on social interaction?

Well, imitation, I've studied a lot on imitation (e.g. Rogers et al., 1996). We are also beginning to compare development in Fragile X syndrome and autism. We have published papers on how well the CHAT (Checklist for Autism in Toddlers) discriminates autism (e.g. Scambler et al., 2001). I'm part of a group that's done some neuroimaging studies of autism, looking at responses to auditory stimuli, looking at brain structural differences. So the research in my group covers several areas.

In Denmark, when we discuss treatment methods, one of the things that we always end up discussing is how to develop social skills. Is it possible to teach social skills and social interaction?

Absolutely. We have a whole set of skills for teaching that. We start with sensory-social routines, which are infant games, to develop nonverbal communication, and children's understanding that their behavior affects other people. Imitation training is a huge part of social interaction. Language is a huge part of social interaction. We teach children that bodies communicate and foster their use of nonverbal communicative gestures. That's a huge part of social interaction. We teach peer play, teach the children to imitate other children. We teach them to play typical games, to carry out pretend play. For young children, those are all important parts of social interaction. We teach the children social dialogs for various situations. We teach these with puppets, with dolls, with each other. Dialogs for birthday parties, dialogs for all kinds of play scripts, so they know what to say when they are playing.

Verbal Communication

Is all this verbal? Can you teach autistic children both speech and social interaction?

Yes. We believe that a large proportion of children with autism are capable of learning to use speech communicatively, if they receive appropriate treatment, at high enough levels of intensity, early enough. That is one of the things I have learned in the last ten years, that most children with autism can be verbal if you do the intervention correctly.

What is the 'correct' intervention to teach verbal language?

Well - There are several groups, using different methods, that are reporting very high levels of speech acquisition. Gail McGee's work uses applied behavior analysis within a naturalistic communicative framework (see e.g. McGee et al., 1999). Lovaas has demonstrated the use of massed trial teaching and a particularly curriculum for teaching speech and language. Both these groups have demonstrated a very high rate of verbal acquisition. In the studies that our group published in the 80's, 75% of the children spoke by the age of five, and now it's probably closer to 90%.

What are the crucial factors for learning to speak?

Different people go at it differently. You know, Gail McGee would say: "shaping." Developing intentional vocalization in intrinsically motivated exchanges, and then using shaping techniques to develop the child's vocalizations into speech.

We want to know what you do .

We think of it as four pieces of the training: 1) imitation training, 2) development of nonverbal communicative gestures, 3) receptive language understanding, and 4) development of understanding symbols. So, in our curriculum, these are four different tracks. Several of them map onto the Lovaas curriculum: Symbols would be like the matching program, receptive language like the receptive language program, imitation like he teaches it. But we teach nonverbal communication in a way that doesn't happen in the Lovaas curriculum. That is the sensory-social routines, turn-taking dialog, body language, affective display. Alternating intensive instruction with these socially affective exchanges - I think that is what we have published before anybody did. Stanley Greenspan talks about it now a lot in a way that is very similar to what we published fifteen years ago . [3]

Is that one-to-one in the classroom?

As I said earlier, children receive their teaching directly from an adult, so the interaction is one to one. The child may be in a small group of children, in which case the teacher would provide teaching directly to the child within the activity of the group.

Do you begin in a more confined setting and then you generalize it?

It depends on the setting where the child already is when we meet him or her. If the children are in a class and that is where we are starting, then we begin to create simple repeated teaching routines in the classroom setting that are repeated frequently and fit inside the ongoing activity. It's more the quality of the interaction with the adult that varies in terms of how directive, structured, and individualized it is, rather than the physical setting, since there are fewer choices about the physical setting.

One Set of Objectives for All People, All Settings

The way it is now, most parents who have young children also want the children being taught at home, and so we teach the families right away, in the very beginning, how to start this kind of teaching. So it is usually happening both at home and in the school. It is very parallel. There is one set of objectives for all people and all settings. We write a detailed set of objectives that is taught at school, that is taught at home. They go to occupational therapy, they go to speech, wherever. One set of objectives that everybody is teaching.

Do you prepare a teaching program two weeks ahead at the team meetings?

Yes, there is just this one program with the objectives and programs and the data-sheets. And we do it, we teach the parents to do it, we teach the school - whoever is working with the child, this is the curriculum. Then we meet every two weeks, and the people doing it make sure it is consistent, look at the data, move it along, similar to the way Lovaas manages his programs.


How important is it to have a data-based treatment?

I think data is quite important for making decisions. I don't know if you need trial-by-trial data, and you don't need to drown in data. But data is very important. One of the things you see in less sophisticated programs is that too many new things are added and kids don't really master something. And I think it is important to have a criterion for therapist to be really sure that this child has learned this skill before you go on. And so a good data system keeps you honest, I think. We don't want to go too fast and have the child have a very shaky foundation of skills. I want to know that the child knows a skill, but I don't need hundred pencil marks to know it, but I want to be sure, I want to see some kind of data, first-trial data that objectively demonstrate the child's level of mastery.

Do you all present actual training with the child at your team meetings?

Yes, exactly. I want to see how the teaching is going. I want to see the skills that have been mastered, and the skills that are not progressing. So the team can decide when its time to add a new activity, when its time to change a teaching approach.


[1] We talked to Sally Rogers for half an hour, just before she gave a talk at the International Autism Conference in Skive, Denmark, November 9, 2002.

Sally Rogers is a Ph.D. in developmental psychology and professor of psychiatry and behavior science at the M.I.N.D. Institute at University of California Davis Medical Center (

Martin Brynskov is co-editor of and David's father. Henning Just is a journalist and has a daughter with autism.

[2] See

[3] See


Eikeseth, S., T. Smith, E. Jahr and S. Eldevik (2002). "Intensive behavioral treatment at school for 4- to 7-year-old children with autism: A 1-year comparison controlled study", Behavioral Modification 26, 1.

Lovaas, O.I. (1987). "Behavioral treatment and normal educational and intellectual functioning in young autistic children", Journal of Consulting and Clinical Psychology 55, pp. 3-9.

McGee, G.G., M.J. Morrier and T. Daly (1999). "An incidental teaching approach to early intervention for toddlers with autism", Journal of the Association for Persons with Severe Handicaps 24, 3, pp. 133-146.

Rogers, S.J., L. Bennetto, R. MacEvoy and B.F. Pennington (1996). "Imitation and pantomime in high-functioning adolescents with autism spectrum disorders", Child Development 67, 2060-2073.

Rogers, S.J. and D. DiLalla (1991). "A comparative study of a developmentally based preschool curriculum on young children with autism and young children with other disorders of behavior or development", Topics in Early Childhood Special Education 11, pp. 29-48.

Rogers, S.J., J. Herbison, H. Lewis, J. Pantone and K. Reis (1986). "An approach for enhancing the symbolic, communicative, and interpersonal functioning of young children with autism and severe emotional handicaps", Journal of the Division of Early Childhood 10, pp. 135-148.

Rogers, S.J. and H. Lewis (1989). "An effective day treatment model for young children with pervasive developmental disorders", Journal of the American Academy of Child and Adolescent Psychiatry 28, 207-214.

Rogers, S.J., H.C. Lewis and K. Reis (1987). "An effective procedure for training early special education teams to implement a model program", Journal of the Division of Early Childhood 11, pp. 180-188.

Scambler, D., S.J. Rogers and E.A. Wehner (2001). "Can the Checklist for Autism in Toddlers Differentiate Young Children With Autism From Those With Developmental Delays?", Journal of the American Academy of Child and Adolescent Psychiatry 40, pp. 1457-1463.

Smith, T., A.D. Groen and J.W. Wynn (2000). "Randomized trial of intensive early intervention for children with pervasive developmental disorder", American Journal on Mental Retardation 105, 4, pp. 269-285.