Association for Behavior Analysis International

Starting with the Olmstead decision of 1999, and accelerated by the nationwide transition to Managed Care, mental health institutions across the country are shuttering. States are under increasing pressure to move individuals out of such institutions and into community-based settings. If an individual engages in severe challenging behavior, executing this transition can be prohibitively difficult and dangerous. In 2014, one New York City agency was awarded a federal grant to integrate advanced care coordination and ABA-based practices into adult residential services. The agency applied this grant to assist the state of New York in the closure of a large mental health institution. Board Certified Behavior Analysts and Registered Behavioral Technicians were tapped to support the work of a newly-assigned interdisciplinary clinical team. In many cases, results were immediate; Challenging behaviors that had persisted for decades under the mental institution's previous auspices were drastically reduced in both frequency and intensity. Over the course of the next 15 months, the individuals in this facility were gradually moved to community-based housing. The discussants will reflect on the strengths and pitfalls of this approach to transition, as well as the overall potential for generalizing these results to other treatment settings.

This symposium contains three presentations regarding improving social functioning for children with autism by teaching them important skills. The first paper involves teaching basic skills of understanding perspective taking. The second presentation is teaching the identification of false-belief tasks, which are commonly used by developmental/cognitive psychologist to test the ability of "theory of mind." The third presentation uses an evaluation form along with video modeling to train parents how to teach their children with autism.

Previous research has reported that the inability of children with autism to discriminate between the reality and non-reality of events may contribute to their difficulties in understanding others perspectives. The purpose of this study is to use behavior analytic approach to teach children with autism to tact mental vs. physical state of verb. A seven year-old child with autism, who had advanced speaker and listener repertoires, participated in this study. A multiple probe across three behaviors design was used. Three target behaviors included (a) discrimination of physical and mental states of events (e.g., Tony takes a train in his hands. Kevin is thinking about the train. Who has the train?) (b) discrimination of reality and imagination (e.g., Mary locked the door. Tim wanted to lock the door. Who indeed lock the door?), and (c) discrimination of reality with and without evidence (e.g., John saw the candy on the table. Helen heard that the candy is on the table and believes the candy is on the table. Who can get the candy for sure?) Multiple exemplar teaching strategy with picture stimuli along with verbal instructions were used during training. Data showed that the rate of correct responses was increased after training. The skills were also generalized to novel scenarios.

Children with autism often have difficulty taking others' perspective--a developmental capacity commonly observed in typically developing children, termed the theory-of-mind (ToM). Teaching the children to tact other's belief is the fundamental skill in ToM. The purpose of this study is to use behavior analytic approach to teach children with autism to identify false-belief tasks. Two children, age 6-7 diagnosed with autism, who had advanced speaker repertoires, participated in this study. A combination of multiple probes across subjects and behaviors was used. Ten scenarios with pictures illustrating the stories were created as instruction materials. The teaching procedure included showing the scenario with drawn pictures, and asking questions regarding the belief of each person. Two target behaviors included:(a) tacting other's belief, (e.g., John is looking for his pencil box. Pencil box may be on the shelf or dining table. John thinks that it should be on the shelf. Where is John going to find his pencil box? Why?) (b) tacting other's false belief (e.g., John is looking for his pencil box that he left on the dinning table. Mary has put John's pencil box on the shelf. Where is John going to find his pencil box? Why?) Both children showed positive results for the acquisition and generalization of the tasks.

The study was conducted in an Early Intensive Behavioral Intervention (EIBI) program of a public children's hospital in the Seoul city. Three mothers whose children were receiving the EIBI service participated in the study. A multiple baseline across participants design was used in the study. During the baseline, a behavioral skill training package was implemented in which lecture, written instruction, modeling, coaching, role-playing, and feedback were provided to the parent during DTT sessions. During the intervention phase, parents were required to watch their performance video scoring their own performance with TPRA forms. Percent of trials which were delivered without errors and rate of delivery of correct trials were measured using TPRA, and overall performance during DTT were evaluated using "Performance Checklist" throughout the study. The results showed that video self-monitoring using TPRA forms were effective in improving parents' performance during DTT with their children with ASD. The improved performance of parents during DTT was maintained during follow-up sessions.

The first presentation reports data on the extent to which a functional reinforcement contingency may facilitate receptive discriminations in children with ASD. The number of trials needed to establish four receptive discriminations was assessed using either a functional reinforcement contingency (e.g., if cookie was the sample stimulus, identifying the cookie produced cookie as a consequence) or an arbitrary reinforcement contingency (e.g., highly preferred stimuli were used as reinforcers, but they had no relation to the stimulus material). The second presentation canters on variables that can predict overall treatment outcome. Given the central role of positive reinforcement in (early intensive behavioral intervention) EIBI, it has been hypothesized that the more reinforcers are available for teaching a specific child, the more that child will benefit from treatment. The second presentation report data on how assessing preferred items that can be used to predict rate of learning in children with ASD receiving EIBI.

Many behavior analytic procedures have proven successful in establishing receptive discriminations in children with autism spectrum disorder (ASD). Most procedures are based on discrete trial teaching, and adding a prompt to the relation between the instruction and the response. Despite applying a number of well-documented effective procedures, some children have difficulties learning receptive discriminations. The purpose of this study was to examine if a functional reinforcement contingency could facilitate receptive discriminations in these children. We compared the number of trials needed to establish four receptive discriminations following well-established procedures under a functional reinforcement contingency and an arbitrary reinforcement contingency in an alternating treatment design. Three out of the six participants showed more rapid acquisition in the functional reinforcement condition. The remaining participants did not establish any discrimination in neither of the conditions. These findings suggest that arranging a functional response-reinforcer contingency should be considered when encountering children that struggle to establish receptive language through more traditional teaching procedures.

Toys, activities and other items that a child express interest in can function as contrived reinforcers during treatment. However, some reinforcers are controlled solely by the stereotyped behavior of the child, and may compete with contrived reinforcement, such when a child produce sensory reinforcement by eye-gazing, rather than complying with a therapist to receive contrived reinforcement. Klintwall and Eikeseth (2012) developed a questionnaire to indirectly assess these types of stimuli, and found that when subtracting the number of stereotyped behaviors from the number of preferred items that potentially could be used as contrived reinforcers (i.e., SMARQ total score); this controlled 50% of the variance in treatment outcome. The present study was designed to replicate and extend the study by Klintwall and Eikeseth (2012), using a prospective design, a new sample, and by assessing preferred items and stereotyped behaviors at intake, rather than later in treatment. Results replicated the findings of Klintwall and Eikeseth (2012) by showing a correlation between SMARQ total score and outcome after one year of EIBI. An interpretation of these results is that for every SMARQ total score, the learn rate in treatment increased by one month per year.

The BACB specifies that supervision provided by BCBAs must be evidence based and include observations of supervisee performance. It is recommended, as outlined in the Supervisor Training Curriculum Outline, that supervision includes competency tests, assessments of skills, direct observation of, and review of written material of supervisees. However, even after going through the required 8-hour supervisor training, supervisors are still unclear on how to make supervision data-based. The current presentation will discuss how to break down the 4th edition Task List in ways that allows the BCBA to continuously assess skill acquisition in supervisees and requires the supervisee to show competency in each Task List item. Combinations of OBM-style approaches can be integrated into what is subsequently created, which should include behavioral skills training, direct observation and assessment of supervisee performance, along with immediate feedback. In doing so, the supervisee knows what is expected of them throughout supervision, the BCBA has a way to monitor skill acquisition of supervisees, and allows both supervisee and BCBA to know that the supervisee has shown competencies in all areas of the Task List.

Practical experience is a vital component of any behavior analytic training program. Practicing skills in supervised settings provides students with the opportunity to further develop their competencies and demonstrate that they have learned the technologies reviewed in their didactic courses. However, ensuring students are receiving quality practical training relevant to their needs while remaining in compliance with the Behavior Analyst Certification Board (BACB) experience standards can be a challenge. During this event our panelists will describe their experiences coordinating practical training systems designed to meet the needs of their students and the standards set by the BACB. Topics reviewed will include, but are not limited to: diversity of training; site development; supervisor management, contact and oversight; assignments and grading; distance supervision; sequenced learning objectives; case management; training of trainers; record review/keeping; and international development. Faculty members directing practical training activities, students accumulating BACB experience hours, and supervisors overseeing their work should benefit from this discussion.

Effective supervision is vital when delivering quality applied behavior analysis services. However, understanding and adhering to the BACB experience standards in addition to providing effective supervision can be a challenging endeavor for behavioral practitioners at all levels. This panel will discuss two critical levels in which the provision of quality supervision is required, line therapists and supervisors. Important questions will be reviewed, such as how to select the right therapists for an ABA organization and how to ensure that supervisors are providing quality supervision. Moreover, the panel will discuss the utility of organizational behavior management and the role of a leadership team in building and implementing performance management systems that can significantly impact the efficacy of supervision, which in turn, can impact the quality of service. Each member of the panel will present systems used at each respective level, discuss findings, and contribute their own unique experiences as they pertain to each of these key areas.

Supervision is a critical component of the development of future practitioners of applied behavior analytic services, and required for all individuals seeking credentialing in ABA. All future professionals seek excellence in supervision, however, despite increasingly structured guidelines provided by certification and licensure organizations, variation exists in the quality and rigor of supervision experiences. Indeed, not all supervision experiences are created equal. For those developing or supporting supervision for the future behavior analyst, minimal guidance exists thus far in behavioral literature related to effective supervision management. Related disciplines have much to offer in the development of quality supervision models, and behaviorally-based literature related to training and instruction, specifically, literature on competency-based training and behavioral skills training models, can support and strengthen the development of quality supervision experiences. Finally, invested constituents have beneficial feedback to add to the supervision process. This symposium targets the developers of supervision in ABA, offering models and guidance for creating quality supervision experiences.

Bridging the gap between theory and applied practice can pose a challenge for many graduate Applied Behavior Analysis students. Employers of these scholars face the unique challenge of accommodating rigorous supervision requirements, changing course schedules and other graduate work commitments that often complicate the landscape. This presentation will discuss several strategies that employers can utilize to support graduate ABA students as they progress through the supervision requirements associated with eventual Board Certification. Specific examples of employer driven initiatives to improve quality of services provided, employee retention, and outcomes for graduate students will be provided. Methods to incentivize employees considering graduate coursework to create a larger more diverse pool of productive clinicians will be reviewed. Finally, opportunities for collaboration with local college programs and course sequences will be discussed with the ultimate priority being the quality of the overall experience for both the student and recipients of the behavior analytic services they provide.

Modeling is a component of Behavioral Skill Training, and has been shown to be effective in evoking desired behavior. Modeling, both in live and video formats, has been incorporated into many models of training successfully. Evidence suggests, however, that modeling is often omitted from supervision experiences, particularly when distance or remote supervision is utilized. This presentation reviews strategies for successfully incorporating modeling into supervision sessions for applied behavior analytic graduate students. Characteristics of the competent model are reviewed, and technical aspects of designing effective models are presented. A model for training supervisors to utilize modeling in supervision is presented.

While there is a strong consensus that applied behavior analysis (ABA) is an effective treatment for autism spectrum disorder (ASD), evidence also indicates variation in individual response to treatment. Several factors have been suggested to have an effect on ABA treatment outcomes. Some factors are specific to the child at the start of treatment (e.g., age, IQ, symptom severity, and skill level), while other factors are treatment specific (e.g., treatment intensity and treatment duration). The present studies evaluate the effects of treatment specific factors on outcomes in large and geographically diverse samples of children with ASD receiving ABA services in community-based settings. These studies investigate the relationship between treatment intensity and skill acquisition, the effects of treatment hours on outcomes across all areas of a comprehensive treatment program, and the impact of features of supervision (i.e., supervision intensity, supervisor credentials, years of experience, and caseload) on skill acquisition. The findings of these studies have significant implications on treatment delivery practices and the optimization of treatment response.

Ample research has revealed that high intensity applied behavior analysis (ABA) treatment (i.e., 30-40 hours per week) significantly improves outcomes of children with autism spectrum disorder (ASD). However, relatively few studies have directly compared higher with lower intensity treatment or investigated these effects across all domains. Two studies were conducted with groups of children receiving behavioral intervention in community-based settings. The first study evaluated the relationship between treatment intensity and learning. A regression analysis was conducted with 810 children between 1.5 and 12 years of age. Results indicated a strong linear relationship between treatment intensity and skill acquisition, where a greater number of treatment hours consistently predicted greater progress over time. The second study examined the relationship between treatment intensity and outcomes within eight treatment domains. A multiple regression analysis was conducted with 599 children. While positive effects were observed across all treatment domains, the greatest effects based on treatment intensity were seen for language, play, and academic skills, and the weakest effects seen for adaptive skills, executive function, and cognition. Treatment duration showed a relatively week impact on outcomes. These findings support existing evidence of the benefits of high intensity ABA treatment programs for children with ASD.

Applied behavior analysis (ABA) is a well-established treatment for autism spectrum disorder (ASD). While ample research has shown the benefits of high treatment intensity, very little research has investigated the role of supervision intensity or other elements of supervision in treatment outcomes. The present study examined the relationship between ABA treatment response and supervision intensity, supervisor credentials, years of experience, and caseload in a large and geographically diverse sample of children receiving ABA services in community-based settings. The present analysis included 663 children with ASD. A multiple linear regression analysis was performed to evaluate the impact of supervision and treatment intensity on learning outcomes. When analyzed together, supervision and treatment intensity accounted for slightly more of the observed variance than therapy hours alone. Additional regression analyses were conducted to evaluate the effect of supervisor credentials, years of experience, and caseload. Supervisor credentials were found to have a significant impact on treatment outcome. Supervisor years of experience and caseload were unexpectedly not found to have a meaningful relationship to skill acquisition. These findings provide guidance for best practice recommendations.

In this symposium, we extend the use of the term makerspace to environments that allow for the contingency-shaping of behavior-analytic interventions and verbal repertoires. A makerspace is an informal establishments where makers gather to create science, technology, engineering, and mathematics (STEM) projects of their own devising. Concomitantly, an increasing number of educational researchers are dedicating substantive time to examining the affordances of makerspaces for supporting interest, engagement, and participation in STEM learning. A defining feature of a makerspace is the informal approach in which instructions are minimized to allow for maximal contingency-shaping. Extended to the science of behavior, we propose the term "shaperspace" to describe an environment in which ABA students are challenged with behavioral excesses and deficits, and learn to employ behavior-analytic interventions primarily through successive approximations. Additionally, shaperspaces offer a verbal behavior community to differentially reinforce tractable, pragmatic descriptions over explanatory fictions. The current state of shaperspace and project-based learning community research should not obfuscate the potential benefits of such environments for facilitating the acquisition of a behavior-analytic repertoire. Nor should the current state of shaperspace research, presented here, impede rigorous behavioral research into the educational affordances of such environments.

A significant obstacle towards developing the behavior-analytic perspective is penchants for using tautological "explanatory fictions" or "mentalisms,"which obfuscate the behavioral conceptualization necessary to effectively address clinical concerns. At a university-based center serving as a field-experience site, pre-service behavior analyst participating in verbal operant training with children with Autism Spectrum Disorders were trained to write a technical description of their interactions with their client subsequent to composing a daily therapy notes. The daily therapy notes were prepared to review with the children's caregivers, and so were specifically composed in common parlance. For the technical description, however, the pre-service behavior analysts were asked to precisely describe how their manipulation of the environment supported their clients' performance and contributed to reducing challenging behavior using the concepts and principles of applied behavior analysis. Specifically, the technical description asked:(a) Which of the childs behavior are you strengthening/weakening? (b) Under what circumstances are the childs behavior excesses/deficits present? And (c) How do you manipulate the childs environment? How does your behavior affect childs responding?The supervising BCBA reviewed this documentation tallying behavior-analytic terms for cumulative examination. This paper presents the results of a multiple-baseline across participants design employed to evaluate the contingent effects of written feedback on supervisees use of behavior-analytic terminology.

The purpose of the Behavior Analyst Certification Board (BACB) supervision experience is to improve behavior analytic, professional, and ethical repertoires of the supervisee, and monitor the performance of supervisees in the field. The independent fieldwork supervisory requirements can be a challenge for supervisees to adhere to if the individual providing supervision is not well versed in the regulations, and up-to-date on the changing requirements. The supervision experience should be carefully programmed, with competency-based assessments conducted of the supervisee's skills. There should be a clear course of study, supervisee's behavior should be operationally defined with objective and measurable goals to determine the application of their skills, and supervisees should receive prudent guidance to enhance their professional development. Supervisees should clearly demonstrate mastery of the competencies outlined in the BACB Task list. This will increase the quality of their experience, uphold the values of the field, and ensure practice requirements are of the highest fidelity and rigor. Panelists will discuss competency-based training standards, the use of a professional portfolio todocument the supervisory experience, and the use of video modeling. Challenges encountered while supervising individuals in community settings will be discussed, as well as solutions to address these issues.

Over the past decades, behavioral techniques have become a staple of "mainstream" psychotherapy. Behavioral and cognitive-behavioral therapies are now the treatment of choice in major healthcare systems that support evidence-based practice. To facilitate the implementation of behavioral technologies on large scale, behavior-analytic interventions have been translated into to treatment packages and manuals accessible to mental health practitioners with varied backgrounds and training. However, this good news about the uptake of behavioral approaches is accompanied by significant limitations in the efficacy and reach of manualized behavior therapy. Behavioral treatment packages demonstrate superior efficacy to non-behavioral control therapies, but yield relatively small effects compared to the early treatments developed by pioneers of applied behavior analysis. For many problems, manualized behavioral treatments fail with a majority of patients. Where has the power of our interventions gone? We argue that packaged behavioral therapies retain our techniques, while underemphasizing (or omitting) the function-based approach that is hallmark of contemporary applied behavior analysis. We discuss this issue as it relates to the implementation of behavioral treatments for several common behavioral/psychiatric problems. Factors contributing to this phenomenon, relevant clinical trials data, and potential remedies are discussed.

Disruptive behavior problems are among the most prevalent child health problems in the U.S. In the 1960s, behavior analysts began to develop powerful behavioral parent training (BPT) interventions, grounded in principles of learning, that led to behavioral normalization in a majority of treated children. These interventions have gained mainstream popularity, with major physician-led bodies now recommending them as a first-line treatment for disruptive behavior in typically developing children. To increase BPT's reach, various treatment packages and manuals aimed at non-behavior-analytic providers have been developed and disseminated. In clinical trials, these treatments demonstrate superior efficacy to waitlist or non-behavioral controls. However, their effectiveness is often suboptimal, with only a minority of children demonstrating a clinically significant response in some studies. Reasons for this variability in child outcomes are discussed from a function-analytic perspective. It is suggested that overreliance on group teaching formats, insufficient function-based individualization, and suboptimal programming for parent behavior change may account for many "treatment failures" observed in applied practice. Strategies and tactics for addressing these issues and strengthening the public health impact of BPT are discussed.

The ideographic tailoring of treatment to a clients specific needs is a central feature of clinical behavioral analysis (CBA). However, contemporary CBA-based therapies are disseminated as packages of techniques as well as conceptual frameworks for conducting treatment with particular populations of clients. In the absence of adequate training in behavior analysis, therapists on the receiving end of dissemination and implementation efforts can only relate to these therapies as collections of techniques. One way to anchor CBA-based therapies to a behavior analytic conceptual frameworks is to have measures of the functional dimensions of clinical presentations guide treatment selection and progress. Unfortunately, we have yet to develop a bank of such measures in CBA. Molar functional relations and metrics from behavioral economics will be presented as measurement opportunities that can help fulfill this need. Second, CBA-based treatment packages need to be dismantled into multiple modules that each address particular functional concerns. This would serve the multiple exemplar learning needs of both therapists and clients and pave the way for ideographic treatment tailoring grounded in CBA-based measurement. IF CBA-based therapies are to be functional as well as popular, we will have to lead the way.

Pioneers such as Lindsley (1964) and Goldiamond (1974) introduced behavior analytic approaches to living well with cognitive difficulties more than half a century ago. Since then, interventions rooted in behavior analysis and targeting individuals who acquired problems remembering, thinking, reasoning, or problem-solving later in life, have been packaged and widely disseminated within the mainstream healthcare landscape (see Projects REACH I and II, for example). Questions have been raised regarding the utility and clinical significance of many of these intervention packages (e.g., Schulz, 2002). We will argue that an understanding of neurocognitive loss and its associated behavioral and emotional changes from a functional perspective is countercultural and thus difficult to acquire without individualized instruction. In effect, most formal and informal caregivers are not able to gauge the deficits and the strengths of the person for whom they care, and packaged interventions do not help caregivers interpret a person’s narrowing skill set from a functional perspective based on behavioral principles. Unsupportive and often coercive environments are inadvertently propagated, even when caregivers receive services in the form of treatment packages. Best practices will be suggested.

Dixon et al. (2015) ranked graduate programs in behavior analysis on the basis of their faculty research productivity. Although controversial, this paper prompted a number of responses from researchers and practitioners in ABA on the important of research in graduate ABA training and how to appropriately rank graduate programs according to the productivity of their faculty. The purpose of this panel is to continue that discussion in an open forum. The panel includes three members with experience in both research and practice in ABA. The panel will discuss the importance of research in graduate ABA training, the importance of formal ranking systems for graduate programs in ABA, whether programs should be ranked on additional factors, such as the research productivity of its students and / or graduates, and the relationship between research productivity and clinician competency. The panel will be chaired by a fourth participant with an interest in this topic.

If one looks for data on the presence of Applied Behavior Analysis-based (ABA-based) autism services in Latin America, nothing systematic will be found, besides a few clinic names and professionals. In another example, if you search for Board Certified Behavior Analysts you will find two in Brazil, two in Mexico, and one in Peru; the numbers in other countries are not very different. People trying to implement and disseminate ABA-based autism services in Latin American countries have been finding different obstacles such as (a) limited resources in their languages; (b) few well prepared professionals; (c) few students and professionals willing to be trained in ABA, (d) lack of openness from schools, parents, and health professionals to implement ABA procedures, (e) lack of awareness and many misconceptions regarding ABA, autism and other developmental disorders. Despite the obstacles, solutions are being designed and implemented, including lectures and workshops to decrease prejudice and misconceptions about ABA; training parents, students, and professionals; creating educational resources in peoples languages; and media dissemination (e.g., radio and TV programs). This panel will discuss different obstacles and solutions related to the lack of well-trained applied behavior analysts who provide autism services in Latin America.

Discriminable shifts between rich and lean schedules of reinforcement can produce maladaptive disruptions in behavior. The disruptions have been studied in rats, pigeons, monkeys, and humans using a variety of procedures to arrange transitions between rich and lean schedules. The presentations in this symposium represent some basic and applied efforts to investigate this phenomenon and to expand findings to novel areas of the field. The first presentation investigated effects of the rich-lean transition on water consumption in non-thirsty rats; the second, effects of warning pigeons of an upcoming transition to a lean schedule of reinforcement; the third, pausing and escape in children with autism in the presence of stimuli associated with the rich-lean transition; and the fourth, the use of aversive features of the rich-lean transition to correct errors when teaching skills to children with autism. Our aim is to promote the dialogue between basic researchers, applied researchers, and practitioners interested behavior during transitions between rich and lean schedules.

Differential pausing in signaled transitions from more favorable to less favorable conditions has been demonstrated with humans and animals in the experimental analysis of behavior. Analysis of the variables responsible for pausing could be useful in understanding problem behavior in transitions between activities for children with autism. We conducted two experiments to extend previous research on pausing and escape during transitions between relatively rich and lean schedules of reinforcement. Individuals with autism spectrum disorders served as participants. The purpose of the first experiment was to replicate previous research on pausing in a two-component multiple schedule with a richer and a leaner schedule of reinforcement. The purpose of the second experiment was to examine whether escape responses would be emitted under the same conditions as pausing. In addition, we examined whether the participants would emit an escape response that removed the schedule or the stimuli associated with the lean schedule. For 2 participants the longest median pause duration was in the LL transition type and for 2 participants the longest median pause duration was in the RL transition type. Mixed results were found when pausing duration data was compared to frequency of escape data.

Research on error correction procedures often include the manipulation of different prompting strategies (e.g., Carroll, Joachim, St. Peter, & Robinson, 2015) or reinforcement schedules (e.g., Hausman, Ingvarsson, & Kahng, 2014), both of which can improve independent responding and acquisition during discrete trial training. We extended error correction research with different schedules of reinforcement by incorporating rich-to-lean transitions following incorrect responses with three boys diagnosed with autism. In the rich-to-rich condition, there was no differential reinforcement and the more-preferred edible was presented regardless of correct responding. During the rich-to-lean condition, errors resulted in the participant receiving less-preferred edibles for the next three correct responses. In the final comparison, the rich-to-no reinforcement condition, errors resulted in no reinforcement for a single trial. The latter two conditions were the most efficient and effective procedures for improving accuracy for two of the three participants. This finding suggests that the aversive properties of rich-to-lean transitions might function to correct errors in the context of differential reinforcement.