Even with a Floortime approach, some children with severe difficulties in relating and communicating continue to show very fragmented behavior. Because their ability to sequence behavior is severely impaired, these children have difficulty problem solving, being purposeful, and imitating simple and complicated patterns. They are unable to move spontaneously from a state of engagement and intimacy, accomplished through improved relationships, to more complicated nonverbal communication and eventually symbolic communication, which require more complex sequences of actions, play, and language. In clinical work with a number of these children, we found it useful to combine behavioral and Floortime methods in the context of a broader relationship and family-support approach.

This model is one that might be abstracted from teaching a skill such as tennis. A certain amount of structure and practice (a person hitting the ball a certain number of times with the backhand) is combined with matches and games (hitting the ball while moving). The more structured behavioral approach may be similar to practicing the backhand while standing still, whereas the relationship-based, dynamic-interaction approach might be similar to practicing under game conditions with more spontaneous action patterns. Some children appear to require both approaches to make progress.

Behavioral techniques are quite helpful in enabling children to sequence behavior and learn to imitate and use some words, but when the techniques are not integrated into a broader floor-time, relationship and family-support model, these behaviors tend to remain dependent on external prompts. For example, only questions asked in a certain way may elicit responses; "What do you want for dinner?" may produce an answer, whereas "Are you hungry?" may not. Skills that are initially learned in a behavioral manner can be brought into a dynamic-relationship context through floor-time activities such as pretend play and spontaneous interactions. For example, the child may learn to imitate the word cup and to use it to label a picture. Later, as cups are used in pretend play, Dad's doll may ask the child for the cup, and the child may respond, "No, my cup!" The word and concept are becoming the child's own, ready for spontaneous use. As behavior is organized under internal emotional cues rather than external prompts, the child's spontaneity and flexibility increase. We often recommend 30 to 45 minutes of structured work followed by 30 to 45 minutes of floor time that incorporates whatever was just practiced. This pattern is repeated many times each day. We also recommend that the goals of the integrated intervention be based on the key developmental foundations that need to be in place for healthy development, rather than on compliant behaviors or isolated academic skills. These goals often are joyful engagement and attention, a continuous flow of gestures (many circles of communication in a row), spontaneous and creative use of more complex, purposeful gestures, imitation, symbols (words and imaginative play), and nonverbal problem solving.

We have also observed in our clinical work that the capacity for abstract thinking develops when a child becomes able to connect different ideas and behaviors to underlying emotions. Behavior and thoughts are tied to desires and wishes, which carry from one situation to the next. Higher-level abstractions often become possible because of the connection of ideas to multiple contexts and, even more important, to multiple aspects of one emotion (the many variations of fear or love, for example). Children ordinarily learn when to wave, hug, argue, or use any other behavior, thought, or conceptual tool by connecting these behaviors to internal emotional prompts.

The hope that behavior controlled by external prompts will generalize is a hope that it will come under internal emotional control, because children take their emotions with them from place to place and use these emotions to determine what to do and say in different situations. Our experience is that children's behavior comes under control only through practice under game conditions. Consequently, many children with severe difficulties in relating and communicating who are in behavioral programs only, and whose parents are not working intuitively with emotions and interactions, are likely to be less flexible and abstract than they could be. We found that several children who had been in intense behavioral programs and could sometimes master rote academic skills and do well on IQ tests lacked the ability for spontaneous, creative interchanges with adults or peers and could not generalize or engage in abstract thinking (they might match words to pictures, for example, but be unable to explain why they wanted to go outside or debate the merits of staying up later). When we initiated a dynamic, problem-solving approach, these children began to acquire abstract-thinking skills.

Because such a range of outcomes is possible for children with severe disorders of relating and communicating, it is especially important to look at each child's unique individual characteristics and developmental capacities when planning intervention. It is also important to consider which type of program will offer which types of children and families the most opportunities for substantial gains. A child without severe processing difficulties may be able to generalize from rote-learning experiences to construct abstractions. A core difficulty of children with autistic spectrum problems (as well as of many other children with different types of special needs) may be their inability to make generalizations and construct patterns of abstract thinking. It is critical to involve children in dynamic, emotional-based, problem-solving interactions that are likely to foster abstract thinking and the ability to generalize. An integrated developmental model can help determine the most appropriate interventions.