Sensory Integration as a Treatment for Automatically Reinforced Behavior A Thesis Presented by

Sensory integration is a theory that suggests that many of the characteristics of autism are caused by poor integration of sensory experiences. Conceptualized in behavioral terms, sensory integration therapy could serve to abolish automatically reinforced behavior. The purpose of this research was to investigate sensory integration as a treatment for repetitive behavior in individuals with autism. Two teenage boys with autism participated in Experiment 1. Both participants had high levels of automaticallyreinforced motor stereotypy. A typical sensory diet treatment was examined to determine whether this treatment would have an abolishing effect on motor stereotypy. It was determined that the sensory diet had no long-term effects on motor stereotypy for either participant. One participant from Experiment 1 participated in Experiment 2, an extension to determine whether the sensory diet had any immediate or short-term effects on motor stereotypy. Results of Experiment 2, showed that the sensory diet suppressed stereotypy while it was being administered, but had no abolishing effects on stereotypy immediately after removal. Based on the results of these two experiments, Sensory Integration therapy, particularly the sensory diet approach was shown to be ineffective in decreasing motor stereotypy. Sensory Integration 3 SENSORY INTEGRATION Sensory integration is a theory of brain-behavior relationships, which proposes that in order to function properly a person must be able to “integrate” or organize sensory input from the world around him (Ayres, 2005). If an individual cannot properly integrate sensations he can end up with anything from mild problems such as difficulty concentrating to severe developmental problems like autism. According to A. Jean Ayres, the originator of sensory integration theory, people receive millions of sensations from the world around them each day, each of which are sent on to the brain. The brain and central nervous system then have to organize and interpret all of these sensations and make sense of them so that the individual is able to make meaning of the things he experiences in the world, and so that he can make adaptive (purposeful, goal-directed) responses to the things he experiences. Sensory integration theory is based on three main components (Bundy, Lane & Murray, 2002). The first component is that all learning depends on the ability to receive and organize sensations from the environment, and to then be able to use this information to “plan and organize behavior” (p. 5). Second is that individuals who are unable to process sensations accurately may then have difficulty producing appropriate actions based on these sensations. Again, this may interfere with learning as well as behavior. The third component of sensory integration is that meaningful activity can produce enhanced sensation and therefore adaptive responding, which can develop sensory integration in an individual who is lacking appropriate sensory integration. This will then lead to an increase in learning and appropriate behavior. Sensory Integration 4 Sensory integration as a process begins when a child is still in the womb and develops with age through experience and interaction with the world, according to Ayres (2005). Disorders of sensory integration usually begin very early on in development, possibly because of a lack of properly interacting with and experiencing sensory stimulation. Ayres claims normal sensory integration develops because humans have an innate “inner drive” to develop sensory integration and will seek appropriate sensations. People with sensory disorders (such as children with autism) often lack this inner drive, and therefore do not seek appropriate sensations and do not experience appropriate sensory integration. As suggested by Bundy et al. (2002), if normal sensory integration does not occur starting in infancy, normal learning and development cannot take place. According to Bundy et al. (2002), sensory integration theory includes five assumptions involving the neural and behavioral bases of sensory integration. The first assumption is that “the central nervous system is plastic” (p. 10). This means that the brain and CNS can change throughout a person’s life, and that although a person may not be able to properly integrate sensations due to the way their brain and CNS have developed, this does not have to be a permanent impairment. Sensory integration theorists believe that with proper therapy the brain and CNS can be changed and learn to properly integrate sensations. The second assumption held by sensory integration theorists is that “sensory integration develops” (p. 11). This means that behaviors at each developmental stage occur in sequence, and this allows for proper development of more complex behavior. If sensory integration does not occur properly, normal development gets disrupted and certain more complex behaviors cannot occur or will occur out of sequence. The third assumption is that “the brain functions as an integrated whole” (p. Sensory Integration 5 11) and that if lower-level brain structures (sensorimotor experiences) do not develop properly, higher-level brain structures (abstraction, language, reasoning, and learning) will not develop properly either. The fourth assumption of sensory integration theory is that “adaptive interactions are critical to sensory integration” (p. 11). In other words, individuals must make purposeful, goal directed responses to sensations from the environment in order to advance sensory integration. The final assumption is that “people have an inner drive to develop sensory integration through participation in sensorimotor activities” (p. 12). This inner drive leads people to seek new sensory experiences and integrate new sensations. Sensory Integration and Autism Sensory integration theorists believe that many behaviors characteristic of autism can be explained by poor sensory integration (Ayres, 2005). Supporters of sensory integration therapy believe that sensory integration is the best treatment method for autism because it is the only therapy that changes the conditions of the brain that cause the maladaptive behaviors (Ayres). According to Ayres, children with autism have problems with three aspects of sensory processing that lead to many of their maladaptive behaviors and inability to learn and develop appropriately. The first major problem that most children with autism have is that they are unable to properly “register sensory information” (p. 128). The portion of the brain which decides whether or not to register each piece of incoming sensory information and whether or not to act on that sensory information may not work properly for children with autism, causing them to not pay attention to many sensations they experience. The child with autism’s brain has a difficult time sorting out all of the different sensations it receives, making sensory Sensory Integration 6 integration and, therefore, normal learning and development difficult. Ayres believes that the second problem that most children with autism have is that they are unable to properly modulate sensory input, especially from vestibular and tactile input. Many children will only register very strong sensations or may be extremely defensive to certain sensations such as light touch. These children may also have difficulty growing accustomed to continuous sensations, and may over-register some sensations and underregister others. The third common problem for children with autism is that they are unable to integrate the sensations that they register to “form a clear perception of space and (their) relationship to space” (p. 131). Children with autism are also unable to form a clear perception of their own body because they do not get adequate sensory information from skin, muscles, joints and the vestibular system. This makes it very difficult for these children to interact with their environment or with others when it is difficult for them to feel what their own body is doing. Again, this greatly impacts learning, development and social skills. Ayres (2005) also believes that children with autism may have difficulty registering the meaning or usefulness of objects in their environment, and therefore are unable to appropriately interact with them. Because children with autism have difficulty with sensory integration, the normal path of brain development has been disrupted and complex behaviors such as abstract thinking are sometimes impossible, without first taking a step back and learning to properly integrate sensations. Ayres also maintains that children with autism lack the “inner drive” necessary to achieve proper sensory integration, and therefore need intensive therapy to expose them to different sensations that they would not otherwise seek themselves. Sensory integration therapy for children Sensory Integration 7 with autism seeks to expose children to different sensory experiences and improve sensory processing. Through this therapy, it is believed that the children will learn to better register and modulate sensations, and make more appropriate adaptive responses. While supporters of sensory integration therapy claim that the child’s life can be “changed considerably” they also admit “no therapy can cure autism” (Ayres, 2005, p. 135). Sensory Integration Treatment The goal of sensory integration as a treatment for autism (or other sensory disorders) is to improve the ability to process and integrate sensory information in order to allow children to be more independent and to participate more in daily activities (Schaaf & Miller, 2005). This is accomplished by giving the child access to sensory experiences that they have missed out on, or not had appropriate interactions with, and to allow the child’s brain to learn how to process sensations and integrate sensations (Ayres, 2005). During sensory integration therapy, a child is provided with an enhanced environment, to help to meet his or her specific sensory needs. According to Ayres, sensory integration therapy sessions often involve providing children with equipment or toys that will stimulate them in certain ways, then allowing them to explore and play with the items however they feel comfortable, and allowing the child to guide the therapy sessions. Ayres specifies several guiding principles of sensory integration therapy, including: the sensory aspects of activities are important to learning, as children organize their sensory perception they gain better control over their behavior, simple responses must be learned before more complex ones, therapy should be grounded in play, and Sensory Integration 8 effectiveness of therapy is measured by whether the child is able to successfully respond to challenges that are presented (p. 143). Bundy et al. (2002) stress the importance of incorporating play into sensory integration sessions. Bundy et al. describe three criteria for a therapy session to be considered “play.” First, any interactions between the child and the environment must be “relatively intrinsically motivated” meaning that the child is engaging in the activity because something about the activity itself is reinforcing, not because of some external reinforcement (p. 230). Second, all of the activities used during therapy must be “relatively internally controlled,” meaning that the child determines who and what they want to play with during the session, and are free to move from activity to activity (p. 231). The final criterion for the session to be considered play is that it must be “free of some of the constraints of objective reality,” meaning that children can engage in pretend or fantasy play and there should not be normal “real life” consequences (p. 232). One common sensory integration treatment plan is known as the “sensory diet.” This protocol was developed by Wilbarger (1991), and involves the “therapeutic use of sensation in the context of daily activities” (Bundy et al., 2002, p. 336). In other words, individuals are provided with certain types of sensory stimulation on a planned and regular basis. Sensory diets are most often prescribed for people who are sensory defensive, and are often used with children with autism (Bundy et al.). With this treatment plan, according to Bundy et al., sensory-based activities are given to the individual at regular intervals throughout his or her day, in order to provide access to certain types of sensory stimulation. Activities, as well as duration of access and length of treatment vary based on each individual’s sensory stimulation needs.. Sensory Integration 9 According to Wilbarger (1991), the rationale behind the sensory diet is that individuals require a “certain quality and quantity of sensory experiences to be skillful, adaptive, and organized in their daily lives” (Bundy et al., 2002, p. 339). Wilbarger believed that repeated sensory input could cause lasting changes in brain functioning; allowing individuals to more properly integrate sensations (Bundy et al.). According to Bundy et al., the sensory diet is especially helpful in achieving sensory integration for those individuals who need appropriate sensory stimulation but do not independently seek this stimulation. Some benefits of the sensory diet, as described by Bundy et al., include improved postural functioning, improved body awareness, improved selfregulation, decreases in tactile defensiveness, reductions in sensory-deprivation, and decreases in self stimulatory behavior. Sensory diets are sometimes used as a sensory integrative technique for children with autism, in an attempt to decrease self-stimulatory behaviors. Bodfish, Symons, Parker and Lewis (2000), found that approximately 75% of children with autism showed high levels of stereotypic behavior and exhibited a variety of different response forms. Bodfish et al. also found that, relative to other children with similar developmental delays, children with autism showed much higher rates of stereotypical behaviors and self-stimulatory behavior. Iarocci and McDonald (2006) also found that between 30100% of people with autism showed some type of sensory sensitivity. In another study, surveys were sent out to caregivers of children with autism (Lam and Aman, 2007),. The results of these surveys showed that 80% of the children had some type of sensory sensitivity. Since children with autism generally display such high rates of stereotypy and self-stimulation, they are often appropriate candidates for a sensory diet approach. Sensory Integration 10 Sensory Integration Controversy Although major proponents of sensory integration theory such as Ayres believe that sensory integration is an effective means of treating many sensory disorders including autism, there has been much debate over the past decade about it’s utility. In a review by Dawson and Watling (2000), four separate studies of the effects of sensory integration on autism were examined. All four studies had positive results, with children showing increased appropriate behaviors (vocalizations, engagement, socialization, etc.) and decreases in inappropriate behaviors such as non-engagement. Dawson and Watling concluded, however, that because minimal research had been conducted on sensory integration, that no conclusions could be drawn about whether it is an effective therapy for autism. In 1992, Temple Grandin, wrote an article about the calming effects of deep pressure for people with autism. In this article she defended sensory integration, and explained from the perspective of a person with autism, how deep pressure and certain forms of touch can help to calm a person with autism and help him to achieve sensory integration. Grandin claimed that in a study she conducted, when 18 children with autism were exposed to a “squeeze machine” that applied deep pressure over their whole body, up to 89% felt calmer. Goldstein (2000) claimed that research evaluating the effects of sensory integration lacked “scientific rigor” and should not be taken to show that sensory integration is an appropriate treatment for autism. According to Goldstein, there is not enough clinically significant research showing that sensory integration is an effective treatment for autism, and therefore it is not appropriate for researchers to advocate its use Sensory Integration 11 as a valuable treatment. Edelson, Rimland and Grandin (2003), responded to Goldstein’s commentary by noting there are plenty of data supporting sensory integration as an effective treatment, and research findings do warrant the use of sensory integration as a treatment for autism. In defense of Grandin’s deep pressure research, Edelson et al., claim that it is acceptable to use a therapy that has not yet been proven effective, as long as there is no danger to the individuals receiving the treatment or when there are no alternative treatments available and when there is no evidence that the treatment is ineffective. Goldstein wrote a second commentary in 2003 in response to Edelson et al., again arguing that the data supporting sensory integration was weak, and that it is unethical to provide services without some reasonable evidence that they are effective. Goldstein also discussed how the cost of using treatments that are not scientifically proven effective greatly outweighs the benefits, and how it is better to just use proven treatments. Griffer (1999) also contributed to the debate over whether sensory integration is an effective treatment with a review of research using sensory integration as a treatment for children with “language-learning disorders.” Griffer, like Goldstein, concluded that there was a lack of empirical evidence supporting sensory integration as a treatment. In the research reviewed by Griffer, there were mixed reviews of sensory integration, with sensory integration being most effective when applied to some type of motor or reflexive measure, and least effective when applied to language. Treating Automatically Reinforced Behaviors with Sensory Integration Sensory integration therapy and sensory diets, despite the controversy that surrounds them, are sometimes used in an attempt to control automatically-reinforced Sensory Integration 12 problem behaviors, such as stereotypy, self-injury or other self-stimulatory behaviors, often with children with autism (Bundy et al., 2002). According to Vaughan and Michael (1982), automatic reinforcement can be defined as any behavior that does not require mediation by another person. Vollmer (1994) specifies that any behavior that is automatically reinforced should meet the following criteria: the behavior persists when the person is alone, contingent access to the automatically-reinforced behavior should function as a reinforcer, blocking the automatically-reinforced behavior will result in response suppression, and social contingencies are not necessary to maintain the behavior. Sensory integration interventions may be effective for treating automatically reinforced behaviors because they may alter the sensory consequences of those behaviors by providing alternate forms of stimulation. Few empirical studies evaluating the effects of sensory integration have been conducted. Of those that have, as previously mentioned, many lack scientific rigor and do not have sufficient data to support sensory integration as a valid treatment (Goldstein, 2003). Ayres’ book about sensory integration theory is based on personal anecdotes involving her experiences with clients. No experimental data are provided, and she mentions few empirical studies or measurable results of therapy. Despite the lack of research evaluating sensory integration, it has become a popular therapy option for children with autism. In the following section the few empirical studies evaluating sensory integration are reviewed. Mason and Iwata (1990) studied the effects of sensory integration therapy on the self-injurious behavior of three children with profound disabilities. Results of a functional analysis showed that one participant’s self-injury was maintained by attention, Sensory Integration 13 one by escape and one by automatic reinforcement. After a stable baseline was established, sensory integration techniques were used for each participant as treatment for their self-injury. During sensory integration therapy, participants were provided with continuous access to items that provided them with auditory, visual, kinesthetic, tactile, and vestibular stimulation for 15 min. During treatment a therapist was present, but did not provide attention (the therapist was only there for safety reasons). Following the sensory integration therapy, the participants were each exposed to a behavioral intervention for their self-injury. For the two participants who had socially maintained self-injury, extinction was used as the behavioral intervention, and for the participant whose behavior was automatically-reinforced, a treatment package involving competing items and DRO was used. Results of this study showed that sensory integration was ineffective in reducing levels of SIB for the participant with attention maintained behavior, unless attention was non-contingently provided as part of the sensory integration treatment (in this case the behavior decreased slightly). For the participant with escape-maintained SIB, sensory integration slightly decreased SIB relative to baseline. For the participant with automatically reinforced SIB, sensory integration actually increased the behavior relative to baseline. In the second phase, when behavioral techniques were applied to the SIB of each participant, SIB was reduced to the lowest levels (zero or near zero levels for all participants). These results showed that sensory integration was not an effective treatment for SIB, especially when it was maintained by automatic reinforcement. In all three cases, contingency-dependent, behavioral treatment strategies were more effective than the sensory integration techniques. Sensory Integration 14 Hoehn and Baumeister (1994) examined meta-analyses of sensory integration research. In eight studies involving children and adults with developmental disabilities, sensory integration appeared to have an overall positive effect on behavior, being most effective for treating motor/reflexive responses, and least effective for treating language problems. However, Hoehn and Baumeister noted that many of these studies had empirical flaws or lacked scientific rigor, preventing their use to support sensory integration therapy. A second meta-analysis was conducted which included seven studies of children with learning disabilities who were exposed to sensory integration therapies. Based on the results of these studies, Hoehn and Baumeister concluded that sensory integration therapy lacked validity and utility as a therapy, and should not be used as a therapy until there is clear empirical support for it. Wells and Smith (1983) studied the use of sensory integration to reduce SIB with four profoundly mentally retarded participants. SIB in this study was assumed to be automatically maintained, although no experimental analysis of maintaining variables was conducted. Sensory integration treatment consisted of slow repetitive, vestibular stimulation (rocking in a hammock or chair) and firm, deep tactile stimulation (vibrator or massage), rolling on a therapy ball, and rolling a large bolster over legs, back and shoulders. Sensory integration sessions lasted 30 min each weekday. Results showed immediate and large decreases in SIB for all participants, and an increase in SIB when baseline was reinstated for one participant. Although these results seem promising, this study, like many others lacked scientific rigor. For example, there were no reliability measures, and there was only a reversal in treatment for one participant. Sensory Integration 15 Function-based Treatment for Automatically-Reinforced Behavior Although there has been little experimental research on treating automaticallyreinforced behavior with sensory integration, there has been much research on other empirically supported function-based behavioral treatments for these behaviors. Vollmer (1994) suggested several methods for treating automatically-reinforced behaviors including: punishment, extinction, differential reinforcement, and altering motivating operations. As with all behavioral treatments, it is important to first identify the maintaining variables of problem behavior (Iwata et al., 1994). If it is determined that the behavior is maintained by automatic reinforcement, punishment, differential reinforcement, and manipulation of motivating operations have been shown to be effective interventions. Punishment-based Procedures Punishment-based procedures for automatically maintained behavior involve delivering an aversive consequence contingent on the target behavior, with the goal of decreasing the target behavior (Vollmer, 1994). In a study conducted by Foxx and Azrin (1973), overcorrection (practicing the correct form of behavior for a specific period of time contingent on the target inappropriate behavior) was compared with other potential punishers (slap contingent on target behavior, differential reinforcement of other behavior, distasteful solution painted on the hand of a hand-mouther, and free access to reinforcement) for various behaviors of children with mental retardation and autism. Results showed that overcorrection eliminated behaviors in all four participants, and was more successful than all of the other alternative procedures. The experimenters Sensory Integration 16 concluded that overcorrection appeared to be a rapid, effective, enduring treatment method for eliminating problem behavior. Ricketts, Goza and Matese (1992) also used punishment to decrease automatically maintained behavior. In their case study, they used a combination of Naltrexone (an endogenous opiate used to decrease sensation from automatically reinforced behaviors) and contingent electric shock (Self Injurious Behavior Inhibition System) to treat the severe self-injury of a 28-year-old man with profound mental retardation. A single electric shock was delivered contingent on each instance of self-injury during treatment. According to the experimenters, SIBIS was clearly superior in treatment of SIB to no treatment or Naltrexone alone in the beginning of the case study, however as SIBIS and Naltrexone were combined, the effectiveness of SIBIS became dose-dependent, with more and more Naltrexone being needed to control SIB and SIBIS having less of an effect (possibly because Naltrexone weakened the painful effects of the shock). Ahearn, Clark, MacDonald and Chung (2007) also examined a treatment strategy that could be considered punishment for treating automatically-reinforced behavior. They used a response interruption and redirection (RIRD) procedure to decrease automatically-reinforced vocal stereotypy in four children with autism. The procedure consisted of issuing a series of vocal demands contingent on any vocal stereotypy, and continuing these vocal demands until the child successfully complied with three of them while not exhibiting vocal stereotypy. Relative to baseline, the RIRD procedure produced substantially lower levels of vocal stereotypy, and higher levels of appropriate communication for all three participants. Although the RIRD procedure may function as punishment for automatically-reinforced behavior (the vocal demands can be aversive Sensory Integration 17 and are applied contingent on the behavior), another component of this procedure involves differential reinforcement of mutually exclusive response topographies (i.e. vocal stereotypy and appropriate vocalizations can not occur at the same time). Extinction-based Procedures Extinction procedures can sometimes be used to decrease automatically reinforced behaviors, but these procedures are often very difficult to implement. In order for a procedure to be technically considered extinction, the behavior must be able to occur, without its typical reinforcing consequences occurring. Because the reinforcing consequences of the automatically-reinforced behavior are often unknown, it is difficult to conduct true extinction. Rincover, Cook, Peoples, and Packard (1979) began an important line of research in which they studied sensory extinction for automatically reinforced behaviors. In this study, four children with autism who exhibited high rates of self-stimulatory behavior participated. First, each child’s topography of problem behavior was observed to develop a hypothesis about the potential sensory consequences that could be maintaining the behavior (auditory, visual, proprioceptive, etc.). Subsequently, a sensory extinction procedure was created based on this hypothesis. For example, with one participant who spun a plate on a table, it was hypothesized that the auditory consequences of plate spinning were maintaining the behavior. So the sensory extinction procedure involved covering the table with carpeting, so that the participant could still engage in the behavior, but would no longer receive the auditory feedback. It was found that with all participants, when the sensory extinction procedure was conducted, the behavior decreased to low levels. The authors noted that the mechanism responsible for the reductions observed was sensory extinction. Ellingson, et al. (2000), Sensory Integration 18 also used a sensory extinction procedure to decrease automatically maintained thumb sucking in two typical children. In this study, they applied a glove to the child’s hand, reducing the sensory consequences of thumb sucking. Thumb sucking was reduced to zero levels for one participant, and reduced moderately for the other participant. Lerman and Iwata (1996) also used sensory extinction (termed response blocking) to reduce hand mouthing in a man with profound mental retardation. Based on an analysis of response patterns across schedule changes in blocking, the authors determined that response blocking actually functioned as punishment. Differential Reinforcement-based Procedures Differential reinforcement procedures can also be used to decrease automaticallyreinforced problem behaviors (Vollmer, 1994). According to Vollmer, when differential reinforcement procedures are used, reinforcement is delivered contingent on appropriate behavior and (sometimes) withheld contingent on the target behavior. One limitation with using this procedure with automatically reinforced behavior is that it is often difficult or sometimes impossible to withhold reinforcement contingent on the target behavior. Taylor, Hoch and Weissman (2005) conducted a case study in which they compared the effects of differential reinforcement of other behavior (DRO) to fixed-time reinforcement for automatically-reinforced vocal stereotypy of a 6-year-old girl with autism. Taylor et al. conducted both an antecedent analysis and a concurrent-operants analysis to identify leisure items that would compete with the participant’s vocal stereotypy, and to ensure that the participant would engage with these toys. After preferred leisure items were found that competed with the participant’s vocal stereotypy, two types of treatments were compared using an alternating treatments design. The first treatment was fixed-time Sensory Integration 19 reinforcement, in which the participant was given access to items once per min for 30-s. The other treatment consisted of a 1-min resetting DRO interval, in which the participant received praise and access to leisure items for 30-s if she did not display vocal stereotypy. If she did engage in vocal stereotypy during the interval, it was reset. Results showed that the DRO was much more effective than FT reinforcement in reducing vocal stereotypy, and the DRO interval was successfully increased to 5 min. Although DRO has been found effective for reducing problem behavior, a limitation of this intervention is that it does not involve delivery of consequences for appropriate alternative behavior. Also, DRO interventions can be very time and labor intensive to administer and require some level of training to implement accurately. One way to reduce the time and resource intensity associated with DRO is to use momentary-time DRO instead of whole-interval DRO. In a momentary-time DRO, reinforcement only depends on whether the behavior is occurring at the moment at the end of the interval. In whole-interval DRO, reinforcement depends on whether the problem behavior occurs during any portion of the interval. Barton, Brulle and Repp (1986) studied whether momentary-time DRO was as effective as whole-interval DRO for reducing the problem behavior of nine children with mentally retardation. The experimenters found that momentary DROs could be used to maintain behavior changes, and were just as effective as whole-interval DROs. The authors also concluded, however, that neither whole-interval DRO nor momentary-time DRO completely eliminated the