DERIVED TEXTUAL CONTROL IN ACTIVITY SCHEDULES: CONDITIONAL DISCRIMINATION TRAINING VERSUS SUPERIMPOSITION AND FADING Evelyn Clare Sprinkle B.A., California State University, Sacramento, 2007 THESIS Submitted in partial satisfaction of the requirements for the degree of MASTER OF ARTS in PSYCHOLOGY (Applied Behavior Analysis) at CALIFORNIA STATE UNIVERSITY, SACRAMENTO SPRING 2011 DERIVED TEXTUAL CONTROL IN ACTIVITY SCHEDULES: CONDITIONAL DISCRIMINATION TRAINING VERSUS SUPERIMPOSITION AND FADING A Thesis by Evelyn Clare Sprinkle Approved by: __________________________________, Committee Chair Caio F. Miguel, Ph.D __________________________________, Second Reader Becky Penrod , Ph.D _____________________________________, Third Reader Jill Young, Ph.D ____________________________ Date ii Student: Evelyn Clare Sprinkle I certify that this student has met the requirements for format contained in the University format manual, and that this thesis is suitable for shelving in the Library and credit is to be awarded for the thesis. __________________________, Graduate Coordinator___________________ Jianjian Qin, Ph.D Date Department of Psychology iii Abstract of DERIVED TEXTUAL CONTROL IN ACTIVITY SCHEDULES: CONDITIONAL DISCRIMINATION TRAINING VERSUS SUPERIMPOSITION AND FADING by Evelyn Clare Sprinkle Children with autism are frequently taught to follow picture activity schedules to increase independence in task engagement and transitions. As textual behavior allows for greater independence and efficiency in responding, schedules should rapidly become textual in form. Previous research has shown the effectiveness of conditional discrimination training and superimposition and fading procedures in establishing textual control. The current study evaluated the relative effectiveness of these procedures in establishing derived textual control in activity schedules. Participants included three children with autism (ages 3, 8 and 10 years). Following the demonstration of proficiency in picture activity schedules, participants were exposed to superimposition and fading and conditional discrimination training in an alternating treatments fashion. Pretest-posttest measures were used to evaluate emergent relations among stimuli. Results indicated that conditional discrimination training was superior to stimulus fading for one participant. iv The remaining participants showed no clear difference between the procedures related to derived textual control. _______________________, Committee Chair Caio F. Miguel, Ph.D _______________________ Date v ACKNOWLEDGMENTS The author would like to thank her thesis committee, Dr. Caio Miguel, Dr. Becky Penrod, and Dr. Jill Young for their efforts in the development and fulfillment of this project. Specifically, the author would like to thank Dr. Caio Miguel for his ongoing and dedicated commitment to the completion of this study. The insight and information I have received from Dr. Miguel has been invaluable to my professional development. In addition, I would like to express my appreciation for the feedback and support from the Sacramento State Verbal Behavior Research Lab; the encouragement of my peers, family and friends has helped to make this experience unforgettable. As a final note, I would like to thank my research assistants for their assistance in IOA and treatment integrity data collection. vi TABLE OF CONTENTS Page Acknowledgments ……………………………………………………………………… vi List of Tables ……………………………………………………………………………..x List of Figures …………………………………………………………………………....xi Chapter 1. INTRODUCTION ……………………………………………………………….…..1 Textual Activity Schedules …………………………………………………….....2 Establishing Textual Control ………………………………………………..........3 Superimposition and Fading ……………………………………………...4 Conditional Discrimination Training ……………………………………..9 2. METHOD ……………………………………………………………………..........16 Participants ………………………………………………………….……….......16 Settings and Materials ……………………………………………….………......17 Dependent Measures ………………………………………………….……...….18 Experimental Design ………………………………………………….…………19 Interobserver Agreement (IOA)…………………………………………...……..20 Treatment Integrity ……………………………………………………….……..21 Procedures …………………………………………………………....…..……...22 Stimulus Preference Assessment ……………….………………..……...22 Pre-Assessment of Schedule Following Behavior ……………...……….23 vii Training Schedule Following ………………………………………...….24 Textual Control Tests ………………………………………………...….27 Textual Behavior Tests ……………………………………………..…...28 Emergent Relations Tests …………………………………………….....29 Conditional Discrimination Training ……………………...…………….30 Superimposition and Fading …………………………………………….34 3. RESULTS ……………………………………………………………….…….........38 Training……………………………………………………….…………….........38 Schedule Following ………………………………………………...…...38 Training Conditions …………………………………………...………...40 Testing……………………………………………………………………...…….44 Textual Control …………………………………………...……………..44 Textual Behavior/ Reading ………………………...……………………46 Emergent Relations Testing ………………………..……………….…...47 4. DISCUSSION ………………………………………………………..…….….…....50 Limitations ……………………………………………………………...……….55 Future Research ……………………………………………………………...….58 Applied Implications …………………………………………………………….59 Appendix A. Sample Data Sheets …………………………………………...………...61 Appendix B. Modified RAISD ……………………………………......………………65 Appendix C. Stimulus Sets for Participants …………………………...…...…………67 viii References ………………………..…………………………………………………….68 ix LIST OF TABLES Page 1. Table 1 Interobserver Agreement………………………………………………..21 2. Table 2 Treatment Integrity……………………………………………………...22 3. Table 3 Order of Experimental Conditions…………………………………........20 4. Table 4 Trials to Criterion and Number of Errors……………………………….40 5. Table 5 Stimulus Sets per Participant …………………………………………...67 x LIST OF FIGURES Page 1. Figure 1 Conditional Discrimination Training: Sample Procedure………………...31 2. Figure 2 Superimposition and Fading: Sample Progression……………..…………36 3. Figure 3 Training Results: Schedule Following……………………………………39 4. Figure 4 Training Results: Training Conditions………………….. ……………….43 5. Figure 5 Textual Control Tests……………………………………………………..45 6. Figure 6 Emergent Relations Tests…………………………………………………49 7. Figure A1 Sample data sheet – Activity Schedule Training………………………..61 8. Figure A2 Sample data sheet – Conditional Discrimination Training and Emergent Relations Tests…………………………………………………………..62 9. Figure A3 Sample data sheet – Superimposition and Fading ……………………..63 10. Figure A4 Sample data sheet – Textual Behavior Tests …………………………..64 xi 1 Chapter 1 INTRODUCTION Autistic Disorder affects many children, and is linked to impairment in many areas of functioning. Some of these deficits include difficulty engaging in independent play, beginning or completing tasks (e.g., self help skills), initiating social interaction, and carrying out daily routines (American Psychiatric Association, 1994). One intervention used to teach these skills is activity schedules. Activity schedules have been shown to increase independence in children’s skills related to the sequential completion of activities (McClannahan, MacDuff & Krantz, 2002). McClannahan and Krantz (1999) recommend the use of a three ring binder, comprised of multiple pages with each page containing the representation of a single activity. Each activity in these schedules is represented by a visual cue (McClannahan & Krantz, 2010). This cue can be presented in either picture or word format and through learning, becomes a discriminative stimulus for selection and completion of a particular activity. Activity schedules can take many forms such as linear (e.g., multiple pictures presented on one surface) or binder (e.g., one picture presented per page) formats, and may be comprised of pictures or written words, but each consists of a presented series of stimuli depicting the activities to be completed (e.g., written list of physical fitness exercises). These schedules signal that the learner will complete all activities in the order in which those activities are presented (MacDuff, Krantz & McClannahan, 1993). When presented with an activity schedule, a child is taught to reference each activity by either 2 looking at or touching the pictorial or textual representation of each activity (e.g., touching the picture of a puzzle prior to retrieving and completing the actual puzzle). After referencing the activity, the child is taught to complete the corresponding activity (e.g., picking up each puzzle piece and fitting it into the puzzle). The child is then taught to reference and complete the next activity presented on the schedule. In the event that an activity does not have a built in completion point (e.g., a puzzle is completed when all pieces are in place), the child may play until motivation to do so decreases, or until a supervising adult or preset timer indicates the time to play is up. The child completes all necessary steps for each activity on the schedule, in either the presence or absence of supervision and reinforcement for activity completion. Activity schedules have been used to teach academic skills such as worksheet completion (Bryan & Gast, 2000; Massey & Wheeler, 2000), play skills such as independent play activities (e.g., McClannahan & Krantz, 2010), physical activity skills such as walking a treadmill (Birkan, McClannahan & Krantz, 2007) and the completion of self care or daily living skills including tooth brushing and meal preparation (Anderson, Sherman, Sheldon & McAdam, 1997) in multiple populations, such as children with autism (e.g., Bryan & Gast, 2000), adults with developmental disabilities (Anderson, et al, 1997) and children with developmental delays (Zimbelmann, Paschal, Hawley, Molgaard & St.Romain, 2006). Textual Activity Schedules Activities are initially represented on schedules as pictures, but as children begin to recognize text, written words corresponding to activities can be used to replace those pictures (Miguel, Yang, Finn & Ahearn, 2009). This use of written words to create a 3 schedule is frequently observed in the behavior of typically developed persons, and is a skill that many people use throughout their lives such as when following a daily schedule, event planner or “to do” list (McClannahan & Krantz, 2010). In addition, McClannahan and Krantz (1999, 2010) provide support for the use of textual schedules by asserting that textual representation of activities is more efficient than pictorial representation in terms of selection from and engagement in a variety of actions and activities. As such, the transfer of control from pictures to written words in the context of activity schedules should be targeted when children begin to develop early reading skills (McClannahan & Krantz, 2010), and should be facilitated by the most efficient method available. When control over responding in the presence of schedules moves from pictures to text, both pictures and written words should be equivalent in that they would evoke similar responding (i.e., completing the depicted activity). When a learner starts to follow a written schedule, their behavior is said to be under control of textual stimuli (Skinner, 1957). Establishing Textual Control Learners following picture activity schedules complete a series of activities when presented with pictures representing those activities. Their responses, including selecting and completing each activity are occasioned by the corresponding picture. The selection of a particular activity is under control of the picture corresponding to that activity. Similarly, when learners follow a written schedule, their responses are under control of the corresponding written words. Several studies have examined ways to transfer control over responding from pictures to textual stimuli (i.e., written words) in relation to various 4 tasks, including mands (Rosales & Refeldt, 2007), vocal and non-vocal responding (Birkan et al, 2007), and independent play activities (Miguel et al., 2009). Within the context of activity schedules, two teaching methods including superimposition and fading and conditional discrimination training have been indicated by research as effective in establishing textual control. Superimposition and Fading One way to transfer the control from pictorial to textual stimuli is through the use of superimposition and fading (e.g., McClannahan & Krantz, 1999). Fading involves the transfer of control from one stimulus to another via gradual stimulus change (Catania, 2007). Through this teaching procedure, pictorial and textual stimuli are presented together, and control over responding is transferred through gradual stimulus change to only textual stimuli. For example, pictures are initially presented with superimposed text, and those pictures are then systematically reduced so that stimuli are comprised of smaller pictures and text, and eventually consist solely of text. While learners respond to textual stimuli following exposure to this procedure, it remains unclear whether or not the procedure produces comprehension of the written word (Sidman, 1994), or if instead, the written word merely functions as another trained visual stimulus which occasions responding. When a student comprehends a written word, the written word and a picture corresponding to that word should be interchangeable, or substituted for one another. Learners who demonstrate comprehension, or equivalence between stimuli, match pictures and printed words accurately. They may be said to “understand” written words (Sidman, 1971). If comprehension is not produced by the teaching procedure, and the 5 written word is simply another visual stimulus which occasions responding, there is limited value in transferring control over responding to the written word, as control could simply have been transferred to an arbitrary stimulus. In order to determine the relative efficacy of a given procedure used to establish textual control, researchers must consider additional skills which may emerge following the use of a particular teaching method. When a skill is not directly taught in a given procedure, yet that skill is present immediately following the procedure as determined by testing, that skill is derived and is a product of the teaching procedure used. For example, testing whether learners match printed words to pictures could show evidence that the teaching procedure produced comprehension of written words. When devoting resources to teaching learners with special needs, the production of derived skills benefits practitioners as well as the learners themselves. The tendency of a procedure to produce derived skills should be taken into account during teaching. While studies using conditional discrimination training to establish textual control examine the emergence of derived skills, studies using superimposition and fading have not. In a recent study evaluating the transfer of control from pictures to written words in activity schedules, Birkan et al (2007) used superimposition and fading to establish textual control over the vocal and motor responding of a six year old autistic boy. The targeted responses during this study were sight word reading, or the emission of textual behavior (Skinner, 1957), while completing each activity in a schedule. During training, printed words of familiar activities were superimposed over pictures corresponding to those activities. The pictures were then systematically faded by decreasing their size over 6 a progression of steps, each of which removed 1 cm. strips from the top and bottom of each picture, for three sets of activity cards including both pictures and written words. The participant progressed through fading steps following two consecutive sessions with 100% correct vocal responses (textual behavior) in the presence of the stimuli assigned to a given set. For example, after two consecutive 100% performances on step 1, the participant would progress to step 2. This procedure was implemented until the participant was responding to the printed word alone. Correct responses consisted of the participant producing a vocal response that corresponded to the superimposed text prior to completing the activity depicted on the schedule. The participant responded correctly in the presence of the printed words for 93% of activities following the superimposition and fading procedure. A total of 30 sessions was required to establish textual behavior and textual control over schedule following behavior. Additionally, the authors reported that skills were maintained at a 44-day follow up, and responding generalized to textual stimuli of different font sizes and colors. Stimulus equivalence (i.e., comprehension of stimuli and substitutability for one another) among written word and pictorial stimuli was not tested in the study. In other words, experimenters did not test the participant on matching of written word to picture or vice versa. The results of Birkan et al (2007) are limited by the design and the potential mediation of the fading procedure by the vocal response. The effects of the stimulus fading (superimposition and fading) procedure cannot be distinguished from the effects of requiring the participant to produce the vocal response corresponding to each item in the schedule. When the participant was required to attend to the presented stimulus, and 7 to simultaneously produce a vocal response, the auditory stimulus produced by the response could have served as a discriminative stimulus for listener behavior of identifying or reorienting to that stimulus (Horne & Lowe, 1996), or as a discriminative stimulus for completing the designated activity. During the study, the participant learned to produce a vocal response in the presence of both the picture and the printed word, and following the production of that response also learned to engage in the activity indicated by the picture, printed word and response produced auditory stimulus. As such, this procedure could have facilitated learning by simultaneously teaching speaker behavior when the participant was required to emit a vocal response, and listener behavior when the participant would orient to the text/picture when hearing its name, as well as listener behavior in the form of activity completion. If the speaker behavior (i.e., reading) acted as a discriminative stimulus for activity completion, the activity completion may never have come under sole control of the written word. Instead, the listener response of activity completion could have been under control of the auditory product of the participant’s vocal response. Several other studies showed support for the efficacy of superimposition and fading while teaching social skills and community skills (e.g., Corey and Shamow, 1972; Krantz & McClannahan, 1998; Mueller, Palkovic & Maynard, 2007). For example, Corey and Shamow taught sight word reading to non-reading, typically developed children, ages 4 to 5.5 years using a fading procedure. They initially paired pictures with written words, and systematically faded the pictures in a progressive fashion by darkening the pictures over 6 steps until only the word was visible. Following training, children accurately read 8 written words aloud. Results indicated that control over vocal responding was transferred from one visual stimulus to another (e.g., from pictures to printed words). As with the experiment conducted by Birkan et al (2007), participants were not required to demonstrate matching of written words to pictures or vice versa, and as a result, this study provides no evidence that participants comprehended written words. In another study using stimulus fading procedures with activity schedules, Krantz and McClannahan (1998) first taught three children ages 4 to 5 years with autism textual behavior by requiring them to engage in textual behavior when presented with two cards containing written words. These words were comprised of mands for social interaction (i.e., “Look” and “Watch me”). Participants’ vocal responses in the presence of textual stimuli were reinforced. Participants were then trained to complete picture activity schedules containing 16 independent play activities. Following activity schedule training, participants were required to complete picture schedules in which the previously learned textual stimuli (e.g., “Watch me”, “Look”) had been imbedded. For example, participants would be presented with a picture activity schedule with cards containing written words placed on pages in addition to the picture of the activity to be completed. Textual stimuli contained words that when produced by participants, were mands for social interaction (e.g., “Watch me”). These written words (e.g., “Look”) were systematically faded as participants demonstrated success by removing portions of the written word cards. Over the course of the study, participants were taught to read written words which consisted of mands for social attention. These written stimuli were imbedded in independent play schedules so that participants were prompted via text to 9 mand for social attention during completion of schedules. Textual prompts were faded so that participants continued to mand for social attention when completing play schedules. Following the fading procedure, all participants continued to demonstrate social interactions for activities taught using the prompt. Conditional Discrimination Training A recently evolved method for establishing textual control and reading comprehension is the use of conditional discrimination training (e.g., Miguel et al, 2009; Sidman, 1994). This method teaches participants to select pictures and written words in the presence of the spoken name corresponding to the target stimulus (e.g., picture of written word), and often results in the emergence of matching picture to words and words to pictures, as well as textual behavior (i.e., sight reading). If learners are taught to follow picture activity schedules, and are then taught that pictures and written words are equivalent or substitutable for each other, they should then follow written schedules. The advantage of this procedure is that, in addition to control by the text, it produces untrained or derived stimulus relations between pictures and printed words, which are identified via pre and posttest measures. Conditional discrimination training is based on the stimulus equivalence literature (e.g., Sidman 1994; Lynch & Cuvo, 1995; Vause, Martin, Yu, Marion & Sakko, 2005; Toussaint & Tiger, 2010; Walker, Rehfeldt & Ninness, 2010). Conditional discrimination training includes the training of two relations, and results in the emergence of four derived (i.e., untrained) relations among stimuli (e.g., Sidman, 1994), thus maximizing learning. When all stimuli (i.e., spoken word, pictures and printed stimuli) are interchangeable in a certain context, such as within a 10 matching procedure, they are considered members of an equivalence class (Sidman, 1994). When these same stimuli exert the same control over behavior however, they are said to be functionally equivalent (Catania, 2007). It has been suggested that when stimuli are members of an equivalence class, establishing control over a response by one member of the class will lead to control over the same response by other members of the equivalence class in the absence of direct training (Goldiamond, 1962). In order to examine the transfer of function of the members of established equivalence classes, specific tests of control over responding by the various class members would need to occur (Sidman, Wynne, Maquire & Barnes, 1989). Previous research has examined the effects of training simple vocal responding (Wirth & Chase, 2002) in the presence of one member of a previously established stimulus class and has observed transfer of function to other members of that class. Wirth and Chase first trained college students to make nonsense vocalizations in the presence of arbitrary stimuli. Stimulus classes between previously taught and additional arbitrary stimuli were then established via MTS procedures. After testing showed the establishment of equivalence classes among stimuli, newly taught members of these classes were then tested to determine if control over nonsense vocalizations transferred to other members of the established classes. Transfer of functional control was demonstrated. McVeigh and Keenan (2009) examined the transfer of stimulus control over a more complex non-vocal response consisting of detailed drawing in the presence of various members of established equivalence classes. Results indicated that some 11 control was transferred, and the authors suggested that complexity of the tasks may have effects on the extent of derived stimulus control. Rosales and Rehfeldt (2007) used conditional discrimination training to establish derived textual mands by two 34 and 58 year old developmentally disabled adults. A mand is a verbal operant that is evoked by a motivating operation and is followed by a particular reinforcing stimulus (Michael, 2004). A derived mand is a mand that is not directly taught to a learner. So, derived textual mands are mands made in the form of written words, which had not been taught in that form. During this study, participants were initially taught to request for items using the Picture Exchange Communication System (PECS; Bondy & Frost, 1994), which is a system used to support non-vocal learners in acquiring verbal behavior. Through a specific sequence of teaching conditions, students learn a variety of verbal operants, including mands (i.e., requests). Participants were then taught to engage in a chained task, consisting of completing the steps to either play a CD on a portable CD player, or to make a pitcher of Kool-Aid®. Participants were then instructed to complete the task, but items required to complete the task were removed. Access to these items was granted contingent on participants exchanging a PECS icon corresponding to the item. During conditional discrimination training, all participants were taught to conditionally relate the dictated name, picture (PECS icon) and printed name of various items necessary for completion of the previously taught chained task. Participants were first taught to select the picture (B) in the presence of the dictated name (A) and were then taught to select the printed word (C) in the presence of the dictated name (A). Following this training, participants 12 demonstrated matching of pictures to corresponding written words (BC) and vice versa (CB). Additionally, participants demonstrated derived mands for items using written words. In other words, participants requested for items using both PECS icons and printed words. Equivalence between stimuli was measured by the correspondence between textual mands and required items, as well as during tests of matching printed words and pictures. This type of posttest was a measure of reading comprehension in that, if participants could to match printed words to pictures or substitute one for the other (e.g., mand via written word or picture), they comprehended the meaning of the written words (Sidman, 1994). While participants had previously requested for necessary items using only pictures, following conditional discrimination training, they requested for these items using text as well, and demonstrated the effectiveness of this method for teaching comprehension in adults with disabilities. In a related study, Rehfeldt and Root (2005) used conditional discrimination training to teach three adults with severe mental retardation to engage in derived mands for preferred items. Researchers first tested the participants’ ability to select pictures and corresponding written words when presented with the dictated name of a stimulus. Textual behavior (i.e. reading a printed word), tacting pictures, matching pictures to text and text to pictures, and derived requesting were also examined during pre and posttests. Tests of derived requesting included whether a participant would request for a preferred item using a textual label corresponding to that item. Following pretesting, researchers taught participants to request for preferred items using the Picture Exchange Communication System (Bondy & Frost, 1994). Following PECS training, participants 13 were taught to conditionally discriminate both pictures and printed names of preferred items (i.e., were taught to select pictures or written words in the presence of the dictated name of those items). All relations tested during pretesting were again tested following training as a posttest measure. Results indicated that all participants made derived textual mands for at least two of three items. In the only study to use conditional discrimination procedures to teach children with autism to respond to written words in activity schedules, Miguel et al. (2009) examined the emergence of derived stimulus control and stimulus relations. Participants were also tested on the emergence of textual behavior (i.e., reading aloud). Participants that had not previously responded correctly to textual stimuli or engaged in textual behavior learned to do so following conditional discrimination training. During this study, two 6 year old children with autism were taught to complete a picture activity schedule. Derived textual control (e.g., responding to written activity schedules) was evaluated via pre and posttests measures in a multiple baseline design across picture sets. Pre and posttests of textual activity schedule following consisted of experimenters presenting and activity schedule comprised solely of printed words and instructing participants to complete the schedule. Textual control was demonstrated by completion of the activity currently depicted on the schedule as a written word. During pretests of textual control, children did not respond in the presence of textual stimuli. Pre and posttest measures of stimulus equivalence examined correct matching of printed words (C) to pictures (B) and vice versa, as well as textual behavior (i.e., sight word reading). Equivalence relations pre and posttests were conducted via a visual-visual MTS format 14 for measures of comprehension (BC and CB). Additionally, the participants’ oral reading responses were tested by asking the participants to read aloud individually presented printed words. During pretests of equivalence, participants did not reliably match pictures to printed words, printed words to pictures, or accurately read aloud printed words. Conditional discrimination training procedures were then implemented. During this training participants were taught to select first pictures and then written words in the presence of their dictated names. For example, the experimenter would dictate a name and the participant would select the corresponding stimulus (e.g., either picture or written word) from an array. Participants were taught to select pictures and printed words in the presence of dictated names using a progressive prompt delay across two stimulus sets of three stimuli each. Following conditional discrimination training, posttests of equivalence were conducted and were identical to pretests. During posttests, participants selected printed words in the presence of their respective pictures and vice versa and accurately read printed words aloud. The results demonstrated the efficacy of conditional discrimination training at establishing derived stimulus relations and textual behavior. Participants also completed written activity schedules (i.e., textual as opposed to picture schedules) accurately following conditional discrimination training, showing evidence that this procedure was effective at establishing derived textual control over schedule following. While both conditional discrimination training and superimposition and fading procedures have been shown to be effective in establishing textual control over behavior in activity schedules, these procedures have not been directly compared. Anecdotally, 15 practitioners report use of superimposition and fading procedures for this purpose. In addition, measures of emergent relations (i.e., matching pictures and printed words and sight reading) for superimposition and fading procedures have not been previously reported. As such, it would be important to determine if the stimulus fading procedure resulted in equivalence among stimuli. The primary purpose of the present study is to evaluate the efficacy of the stimulus superimposition and fading procedure as compared to that of conditional discrimination training in establishing textual control over behavior in the context of activity schedules. A secondary purpose of the present study is to evaluate whether both procedures produce emergent relations among pictorial and written stimuli, and produce emergent textual behavior. 16 Chapter 2 METHOD Participants Three children diagnosed with autism participated in the study. Larry was a 10 year, 11 month old boy who participated in home school, and had had extensive previous experience with 1:1 teaching using the methods of applied behavior analysis. Laura was an 8 year, 11 month old girl whom attended a public school district special education classroom for children with autism and other diagnoses. Laura also had extensive experience with 1:1 behaviorally based teaching methods. Virginia was a 3 year, 4 month old girl, who was enrolled in a mainstream preschool class. Virginia had some experience (e.g., less than one year) in 1:1 behaviorally based teaching. All participants communicated using a minimum of three to five word sentences and demonstrated listener skills such as selecting items from an array. Participants also demonstrated speaker skills in that they were able to produce tacts of items or pictures. Additionally, all demonstrated simple intraverbals such as answers to simple questions. All participants had a limited sight word vocabulary. Larry had previous experience with, and could independently complete picture activity schedules, while Laura and Virginia were taught to follow schedules over the course of the study. All participants were able to sit for up to five minutes in structured teaching settings. 17 Setting and Materials Sessions were conducted in a designated area within each participants’ home. Each area included open space for engaging with leisure/independent play items, as well as a table with two chairs. A video camera or additional researchers were present for all sessions for data collection purposes. Materials included a variety of preferred items (e.g., edibles) and activities (e.g., games, puzzles). Activity schedules for all participants consisted of vertical, linear presentation of three activities. One set of three cards corresponding to activities was used for conditional discrimination training (CDT), and one set of three cards corresponding to activities was used for superimposition and fading (SF) procedures. Items identified as preferred by each participant’s primary caregiver were counterbalanced across sets for length of words/total number of letters in words within each set. Twelve laminated cards (3 x 5 in.) including six photographs of the activities for each participant, and six cards printed with corresponding names were presented to each participant during the study (Please see Appendix C). These twelve cards were used during pre and posttest measures of textual control, textual behavior and stimulus equivalence, and were also used during conditional discrimination training. Cards used during the superimposition and fading procedure also contained pictures and corresponding words of items and activities identified during the preference assessment, but a total of 36 additional cards were used during the stimulus fading procedure. Additional cards used in the superimposition and fading procedure included both the printed name and systematically smaller portions of the pictures used for this procedure. 18 All cards measured 3 x 5 inches, and all text was printed in black in 72 point Arial font. All pictures were printed in color on white backgrounds. Please refer to Appendix C for pre and posttests cards/pictures for Sets SF and CDT. Additional materials were presented to Larry during the stimulus fading procedure. Larry was exposed to nine additional cards, including text, colored text, pictures and various combinations of text and pictorial stimuli as described below. Dependent Measures The primary dependent variable of interest in the study was derived textual control, which was measured by the percent of correct schedule following responses made in the presence of printed words during pre and posttests. Responses during textual control probes were scored as correct when a participant responded by selecting and engaging in an activity in the presence of written words on the activity schedule. Conversely, responses on tests of textual control were scored as incorrect in the event that participants either selected or engaged in activities not currently depicted on the schedule, or if participants failed to respond in the presence of textual stimuli. Additional dependent variables included the emergence of textual behavior (i.e., reading written words aloud) and emergent relations indicative of reading comprehension, such as accurately matching pictures to printed words and vice versa. Pre and posttests of emergent stimulus relations measured the percentage of correct independent responses made in the presence of the sample stimulus. Emergent relations trials were scored as correct when a participant selected the printed word that was a positive comparison to the sample picture (BC) or selected the picture that was a positive comparison to the sample 19 printed word (CB). Tests for emergent relations (e.g., BC and CB) were conducted in a basic matching to sample (MTS) format (see below). Textual behavior trials were scored as correct when the participants produced the correct vocal response in the presence of the printed word (CD), and were conducted in a simple discrimination format. In addition, both the number of trials to meet training criterion and the total number of errors made were examined as a measure of the relative efficacy of the two teaching methods. Experimental Design An alternating treatments design was employed (Barlow & Hayes, 1979) in order to compare both procedures, with pre and posttests measuring changes in dependent measures (i.e., measures of textual control and emergent relations). Each of six activities was assigned to one of two sets for each participant. Each of these activity sets was comprised of three activities which was assigned to one treatment condition (i.e., conditional discrimination or superimposition and fading).Please see Appendix C for pre and posttests stimulus sets for each participant. The same linear schedule was used for all training conditions, and stimulus cards corresponding to activity sets (e.g., Sets CDT and SF) were varied based on condition. Pre- and posttests were used to evaluate the emergence of textual control, textual behavior, and emergent stimulus relations. Please refer to Table 3 for a complete list of all conditions in the order in which they were presented. 20 Table 3 Order of Experimental Conditions Order Condition Testing/Teaching Procedure ________________________________________________________________________ 1 Pretests of Schedule Following Task Analysis 2 Training Schedule Following* Task Analysis 3 Pretests of Textual Control Task Analysis 4 Textual Behavior Pretests Simple Discrimination 5 Emergent Relations Pretests Matching to Sample 6 Alternating Treatment Conditions Superimposition and Fading; Conditional Discrimination Training 7 Posttests of Textual Control Task Analysis 8 Posttests of Textual Behavior Simple Discrimination 9 Posttests of Emergent Relations Matching to Sample * Activity schedule following trained only if participants failed to demonstrate schedule following during assessment. Interobserver Agreement (IOA) An additional independent observer scored testing and training sessions via either live observation video recording during 72.9% of all sessions. An agreement was defined as both the experimenter and the independent observer scoring either an incorrect, correct or 21 prompted response on a single trial. A disagreement was defined as a difference in scoring by the experimenter and the second observer on any given trial. IOA was calculated by dividing the number of agreements by the total number of agreements plus disagreements and multiplied by 100. Mean IOA was 97.1% across participants. Table 1 contains individual IOA values for each participant. Table 1 Interobserver Agreement (IOA) Laura Virginia Larry All % Blocks with IOA 78 72 67.8 72.9 IOA% 99.6 93.8 97.9 97.1 77 – 100 54 – 100 IOA Range 77 – 100 55 – 100 Treatment Integrity Treatment integrity was a measure of correct implementation of training as designed. Trials were scored as correct by the independent observer if the experimenter presented stimuli according to the pre-randomized arrangement of the stimuli, presented the correct discriminative stimulus during MTS, and delivered reinforcement or implemented error correction procedures according to participants’ response. An incorrect trial was recorded if the experimenter failed to complete any of the above as outlined. Treatment integrity was assessed by dividing the number of correctly 22 implemented trials by the number of correctly implemented plus incorrectly implemented trials multiplied by 100. Treatment integrity was assessed for an average of 66.9% of training and testing blocks across participants. Mean treatment integrity was 99.4% across participants. Please see Table 2 for individual participant values. Table 2 Treatment Integrity (TI) Laura Virginia Larry All % Blocks 78 55 67.8 66.9 TI 99 99.6 99.7 99.4 89 – 100 89 – 100 TI Range 89 – 100 89 – 100 Procedures Stimulus Preference Assessment The purpose of this procedure was to identify potential reinforcers for each current experimental session. This assessment was conducted once at the beginning of each experimental session. Parents or caregivers of each participant completed a modified version of the Reinforcer Assessment Inventory for Individuals with Severe Disabilities (RAISD; Fisher, Piazza, Bowman & Almari, 1996) prior to the beginning of the study (see Appendix B). Based on the RAISD, a number of potential reinforcers were selected to be presented to each participant in a brief multiple stimulus preference assessment 23 without replacement (MSWO; Higbee, Carr & Harrison, 2000). Only edible items were included in this assessment. Prior to the beginning of each session, participants were exposed to a brief MSWO (i.e., one trial) including edible items from the RAISD and were allowed to choose their preferred reinforcer. A minimum of three preferred edible items were randomly placed in a line approximately 10 cm. apart on the table in the experimental area. Each participant was instructed to select one item from the array. For each training session, the first item selected was used as the reinforcer for the remainder of that session. In the event that a participant vocally requested a different edible item throughout the session, the request was honored. The RAISD was also used to identify items and activities to be included within activity schedules. Tangible, non-edible items identified as preferred by each participant’s primary caregiver were counterbalanced across sets for length of words/total number of letters in words within each set. See Table 5 (Appendix C) for stimulus sets for each participant. Pre-assessment of Schedule Following Behavior This phase was conducted in order to ensure that all participants were able to follow a picture activity schedule. During this pre-assessment phase, participants were presented with a minimum of two, three item picture activity schedules, so that each picture used throughout the course of the study was presented once during this preassessment phase. In the event that participants correctly followed schedules, they were exposed to additional three item schedules to ensure consistent demonstration of this skill. Pictures of items identified via the RAISD were used in this phase. Each 24 participant was presented with pictures corresponding to items in both previously identified activity sets (i.e., Set SF and Set CDT). An experimenter sat in the room for data collection purposes, but did not interact with the participant following the presentation of the initial instruction, “Do your schedule.” Correct responses consisted of participants approaching the schedule, referencing and selecting the item or activity, completing the activity, removing the picture, and repeating the process until all activities were completed. A response was defined as incorrect when a participant failed to respond in the presence of the presented schedule, or completed any of the above outlined steps for an activity not corresponding to that currently depicted on the schedule. When a participant completed 80% or more of all required steps to follow a picture activity schedule in the absence of prompting, they then progressed to the next phase of the study. When participants failed to demonstrate schedule following behavior at criterion levels, they were trained to complete the steps listed above. Pretests for activity schedule following were terminated when the participant failed to respond correctly, or did not respond within two minutes of the presentation of the instruction. Training Schedule Following During this phase, participants were trained to complete all steps of the PAS, including approaching the schedule, referencing and selecting the item or activity, completing the activity, returning to the schedule and repeating the process until all activities were completed, in a manner adapted from that described by MacDuff et al (1993). 25 Teaching began with a vocal instruction (i.e., “check your schedule”, etc.). The instructor then waited five seconds for the participant to begin responding to the instruction. In the event that the participant did not respond within five seconds, or responded incorrectly to the instruction by engaging in any response other than referencing the schedule, the instructor then prompted the participant to begin the sequence of activities. All prompts were delivered from behind the participant, and were initially manual prompts (e.g., hand over hand). The hierarchy of prompts included manual, spatially faded and proximity prompts. Prompts were first faded from manual prompts to spatial fading (e.g., the location of the manual prompt was changed), and then to shadowing (e.g., experimenters moved with participants but did not provide physical guidance), following which physical proximity of the researcher to the learner was faded to distances of two feet, and finally, five feet. No prescribed vocal or gesture prompts were provided during training. Prompts were faded according to the outlined progression when participants demonstrated on-task and on-schedule behavior for a minimum of 80% of steps for each prompting progression, for two consecutive schedules. The activity schedule training phase was considered complete following two consecutive demonstrations of schedule following, including a minimum of 80% of responses being both on-task and on-schedule, with researchers faded to five feet from participants. Participants were trained to complete activity schedules consisting of three items. The order of activities was pre-randomized for each presented schedule in such a way that no one activity consistently preceded or followed another. Steps required to 26 complete each schedule included: 1) approaching the schedule 2) pointing to the first picture, 3) gathering necessary materials, 4) completing the activity, 5) putting the materials away, and 6) removing the picture and placing it in the all done pocket. Steps 2 through 6 were then completed for each of the subsequent activities, for a total of 16 steps required to complete one full schedule. Data on participants’ on-schedule and on-task behavior was collected during each step of each schedule completed. Correct responses for on-task behavior consisted of participants engaging in at least one of the following: attending to materials or the schedule, manipulating materials appropriately, and transitioning between activities. Ontask responses were scored as incorrect if participants engaged in any response other than those listed above (e.g., stereotypy or other forms of behavior excesses) for longer than five seconds. On-schedule behavior was scored as correct when participants were engaged in the activity, or were completing any steps required to engage in the activity currently depicted on the schedule (e.g., selecting the item). On-schedule responses were scored as incorrect when participants engaged in other responses, including behavior excesses or when they engaged in activities not currently depicted on the schedule for longer than five seconds. In the event that participants engaged in inappropriate behavior (e.g., stereotypy), or paused for more than five seconds they were immediately returned to the previous step in the schedule, and were then prompted to initiate the following (i.e., previously incorrect) step. Participants were initially prompted to engage in both on-task and on-schedule behavior while completing schedules. 27 Textual Control Tests The purpose of this phase was to test if written words evoked activity completion prior to and after training with stimulus fading and conditional discrimination. During this condition, all pictures in the participants’ activity schedules were replaced with corresponding written words. All items corresponding to printed words were present in the room within four feet of the participant. The experimenter presented the activity schedule and instructed the participant by stating, “Do your schedule”. Schedules were placed on a table (Virginia and Larry) or attached to the wall (Laura) in the work area, based on requests by participants’ parents/caregivers within two feet of the location from which the experimenter issued the initial instruction The experimenter remained in the room for data collection purposes, but did not interact with the participant in any way other than to deliver the initial instruction. Correct responses consisted of participants approaching the schedule, referencing and selecting the item or activity when presented with the printed word, completing the activity, removing the picture, and repeating the process until all activities were completed. Incorrect responses consisted of the participants failing to respond in the presence of the presented written schedule, or completing any of the required steps for an activity not corresponding to that currently depicted on the schedule. In the event that the participant did not begin to complete the schedule (i.e., initiated schedule following) within two minutes from the delivery of the instruction, the pretest for textual control was terminated and responses were scored as incorrect. Pretests for textual control were not terminated for any participant, as all 28 participants approached the schedule. No participant selected or engaged in activities after referencing written words. Textual Behavior Tests Pretests and posttests for textual behavior, or reading a printed word aloud were completed for stimulus sets used for both conditional discrimination training and stimulus fading procedures. These tests consisted of the experimenter presenting the participant with the printed word and requiring each participant to read it (CD). This relation was tested successively by presenting one card at a time. The experimenter held up one card containing a printed word, and asked “What is it?” Responses were scored as correct if participants produced a recognizable vocal approximation of the printed word, (e.g.,, recognizable vocalization corresponding to the sample), or correctly read the word aloud. Responses were scored as incorrect if participants either failed to respond, or if they did not accurately read the word. All responses during pretesting, regardless of accuracy, resulted in neutral social feedback (e.g., “OK”). In order to ensure continued engagement on the part of the participant, maintenance trials were conducted after every two to three test trials. These maintenance trials consisted of previously mastered instructions (e.g., “touch your nose.”) and edible items were delivered contingent upon correct responses. Textual behavior testing was conducted in 9-trial blocks for each set of stimuli (i.e., Set SF and Set CDT); one nine-trial block was conducted per set, with one block of pretesting conducted for each set of stimuli. Stimuli were presented in a pre-randomized order, so that each sample stimulus was presented a total of three times in a simple discrimination format. 29 Emergent Relations Tests Pretests and posttests for emergent relations were completed for stimuli sets used for both conditional discrimination training and within stimulus fading procedures, both before and after training, and after textual control tests. Tests of emergent relations were conducted using a typical MTS format, during which the participant selected the printed word in the presence of the picture (CB), or in the presence of the picture, selected the printed word (BC). During MTS tests of emergent relations, participants were required to engage in an observing response consisting of touching the presented sample stimulus card (Dinsmoor, 1985). The observing response then produced an array of three comparison stimuli, which were positioned equally in each position on the board, and oriented to the other cards an equal number of trials. Responses were scored as correct when participants retrieved the corresponding comparison. Trials were scored as incorrect when the participant selected an incorrect comparison stimulus, or failed to respond within five seconds from the presentation of the comparison stimuli. All responses during pretesting, regardless of accuracy, resulted in neutral social feedback (e.g., “OK”). In order to ensure continued engagement on the part of the participant, maintenance trials were conducted after every two to three test trials. These maintenance trials consisted of previously mastered instructions (e.g., “touch your nose.”) and edible items were delivered contingent upon correct responses, and occurred in addition to the 9-trial block completed for each relation. Testing was conducted in 9-trial blocks for each relation described above (e.g., BC, CB, CD for each set of stimuli); each participant was exposed to a block comprised of nine testing trials, plus three to four maintenance 30 trials. In the event that a participant scored higher than chance levels during a given pretest, a second identical pretest was administered to ensure that previous responding was solely due to chance. Conditional Discrimination Training Conditional discrimination training was conducted in a manner consistent with that described by Miguel et al. (2009). Each participant was exposed to conditional discrimination training for one set of activities (i.e., Set CDT). Figure 1 contains a sample procedure for conditional discrimination training. In this example, the relation taught is selection of the picture in the presence of the dictated name (AB). The experimenter required an observing response (e.g., eye contact, correct responding to a simple instruction such as “Give me five”). Following the observing response, the experimenter presented the auditory stimulus (e.g., dictated name), and presented the three stimulus array. The participant then selected the positive comparison. Selection was either independent or occurred following a prescribed gesture prompt. 31 Relation (AB): Selection of picture in presence of spoken name Observing response: Eye contact by participant or simple instruction following (e.g., “Give 5”) Dictated name stimulus delivered by experimenter: Puzzle Comparison Stimulus Array: Selection Response: Remove comparison corresponding to sample from array. Figure 1. Sample procedure for conditional discrimination training. Relation displayed is selection of the positive comparison picture to the sample dictated name stimulus (AB). 32 During conditional discrimination training, participants were taught to select either the picture (B) or printed word (C) when presented with the dictated name of a given activity or item (A). A progressive one second prompt delay procedure was used Each training phase consisted of six steps, with each step incorporating a one second increase in the prompt delay. In step 1, the experimenter immediately provided a gesture prompt (i.e., pointing) to the correct comparison stimulus following the presentation of the dictated name. In step 2, the experimenter delivered the dictated name and then waited 1 second before delivering the gesture prompt. The delay was systematically increased by one second per step so that for step 3, the delay was two seconds, for step 4 the delay was three seconds, and so forth. In other words, the delivery of the prescribed prompt was systematically delayed by one second per step until no prompt was prescribed (i.e., independent responding was expected). This increase in the time between the presentation of the stimulus and the gesture prompt continued until step 6. In step 6, no prompts were provided. The experimenter provided the dictated name, and then waited five seconds. Trials for step 6 were scored based on whether the participants selected a positive or negative comparison to the sample. If participants failed to respond within five seconds, the trial was scored as incorrect. The experimenter delivered social praise (e.g., “Good job” or “That’s right”) contingent upon each prompted or unprompted correct response, or selection of correct comparison in the presence of the sample. In addition, tokens were delivered contingently upon responses that occurred prior to the delivery of the prescribed prompt. These tokens were drawn on to a token board by the experimenter following each 33 positive comparison in the presence of the sample. For example, the experimenter would quickly draw a star on a token board while providing social praise. Preferred edibles were delivered following a specific number of unprompted responses that occurred prior to the delivery of the prescribed prompt. If a participant gave an incorrect response, an error correction procedure was implemented. This error correction procedure consisted of the experimenter representing the dictated name of the stimulus and immediately prompting (i.e., zero delay) selection of the positive comparison using a gestural prompt (i.e., pointing to the correct response). No participant failed to respond to the initial gesture prompt provided in the error correction procedure. For steps 2 through 5, a differential reinforcement procedure was implemented following the first independent response made by each participant (Karsten & Carr, 2009). Once each participant responded correctly in the absence of prompting, only independent responses were followed by the delivery of a secondary reinforcer. All responses emitted following the prescribed prompt were followed by the delivery social praise. All relations were trained using a three stimulus array. Three cards were presented to the participant, and were positioned on each trial in such a way that each stimulus was presented an equal number of trials (e.g., three trials per block) in each of three positions (e.g., left middle and right). Each relation was trained in 9 trial blocks until each participant achieved 89% (8 out of 9) or greater accuracy across two consecutive blocks in the absence of prompting. 34 Superimposition and Fading Procedure Stimulus fading was conducted in a manner adapted from that described by Birkan et al (2007). Each participant was exposed to this procedure for one set of activities (i.e., Set SF). The stimulus fading procedure was comprised of 12 steps. In the first step, the entire picture of an item or activity was presented on a card, with the textual label corresponding to that item superimposed over the picture. Additionally, three items, each corresponding to one picture with superimposed text were presented to the participant in a three stimulus array. These stimuli were positioned on each trial in such a way that each stimulus was presented for three trials per block in each position (e.g., left, center and right). Participants were shown the picture with superimposed text, were required to engage in an observing response consisting of touching or removing the picture from the stimulus board, and were provided with the instruction to “find it” or “get it.” Participants were then required to select the item corresponding to the picture. For step one, participants were immediately provided with a gesture prompt (e.g., the experimenter pointed to the correct item/activity), with 100% of the picture presented. For step 2, the top and bottom five percent of each picture was removed (i.e., total of 10%), while the superimposed textual label remained constant. The experimenter presented the visual stimulus and then waited one second prior to delivering the gesture prompt. For each consecutive step, 10% of the picture was removed, and a progressive one second delay between the delivery of the visual stimulus and the gesture prompt occurred. Gesture prompts from the experimenter were completely faded by step 6, during which participants were required to select the correct item or activity within five 35 seconds. For example, in step 3, 20% of the picture was removed (i.e., 80% of the picture of the item will be visible), and a two second delay will occurred between the delivery of the visual stimulus and the gesture prompt. In step 4, 30% of the picture was removed, and the gesture prompt was delayed by three seconds. In step 5, 40% of the picture was removed and a four second delay occurred between presentation of the visual stimulus and the prompt. No prompt was delivered beginning in step 6, when 50% of the picture was removed. For each consecutive step, an additional 10% of the picture was removed, so that or step 7, 60% of the picture was removed, in step 8, 70% of the picture was removed, in step 9, 80% of the picture was removed, and in step 10, 90% of the picture was removed. This pattern continued until step 11, during which only 5% of the picture was shown, and the textual stimulus remained constant. In step 12, no portion of the picture corresponding to the item was visible, and no prompts were delivered. Figure 2 contains a visual display of a sample fading progression for the superimposition and fading procedure. 36 Step 1. Full picture Step 2. 10% picture removed PUZZLE PUZZLE Step3. 20% picture removed Step 4. 30% picture removed PUZZLE PUZZLE Step 5. 40% picture removed Step 6. 50% picture removed PUZZLE PUZZLE Step 7. 60% picture removed Step 8. 70% picture removed PUZZLE Step 9. 80% picture removed PUZZLE Step 10. 90% picture removed PUZZLE Step 11. 95% picture removed PUZZLE Step 12. 100% picture removed PUZZLE Figure 2. Sample superimposition and fading progression. PUZZLE 37 For steps 6 through 12, when participants failed to respond within the five second time period, the trial was scored as incorrect, an error correction procedure was implemented. This procedure consisted provision of the experimenter representing the sample stimulus and immediately providing a gesture prompt indicating the selection of the positive comparison. Reinforcement and error correction procedures for stimulus fading procedures were identical to those described above for conditional discrimination training. Additionally, responding via selection of an item in the presence of a given visual stimulus was trained in nine trial blocks. Mastery criteria of each step were 89% on two consecutive blocks. The criterion to return to the previous step was three consecutive incorrect responses. 38 Chapter 3 RESULTS Training Schedule Following Prior to the introduction of training procedures, participants schedule following behavior was assessed. Larry followed a picture activity schedule and completed all steps with an average accuracy of 97% across six separate pretests. As such, he did not complete training of schedule following, and progressed to pretesting of textual control. Laura and Virginia did not follow the picture schedule and were taught to complete the required steps via graduated guidance prior to completing pretests of textual control. Please see Figure 3 for results of picture activity schedule training for Laura (upper panel) and Virginia (lower panel). Solid squares indicate on-task responding and open diamonds indicate on-schedule behavior. Laura and Virginia required 13 and ten trials (i.e., presentations of picture schedules) respectively to learn to follow a picture activity schedule to criterion performance. Both performed at criterion levels at a two week follow up. 39 Figure 3. Block – by – block data of acquisition of activity schedule following for Laura and Virginia. Percentage of correct responses for both on-task and on-schedule behavior for Laura and Virginia. 40 Training Conditions Figure 4 contains results of training for Laura, Virginia and Larry. Solid circles represent the percent per block of positive comparisons to the sample when participants selected pictures in the presence of the dictated name in conditional discrimination training (i.e., relation AB). Solid triangles indicate the percentage of positive selections of the printed word in the presence of the dictated name (AC). Open diamonds indicate the percent of selections of the positive comparison to the sample in the stimulus fading procedure. Table 4 depicts the number of trials to meet criterion performance levels and the number of errors made for each procedure. Table 4 Trials to Criterion and Errors - Conditional Discrimination Training (CDT) and Superimposition and Fading (SF) Trials to Criterion _________________________ Participant CDT Total Errors _______________________ SF CDT SF ________________________________________________________________________ Laura 261 225 16 11 Virginia 216 243 3 9 Larry 216 603 4 75 41 Laura acquired the selection of picture in the presence of the spoken name (AB) in 15 blocks, and acquired selection of the printed word in the presence of the spoken name (AC) in 14 blocks of teaching using the conditional discrimination procedure. She learned to select items in the presence of stimuli corresponding to Set SF (i.e., pictures with superimposed written words faded to only written words) in 25 teaching blocks. Laura completed training for both procedures within the outlined 12 steps. Virginia acquired selection of picture in the presence of the dictated name (AB) and selection of printed word in the presence of the spoken name (AC) in 12 blocks of training for each relation. She required 27 blocks of teaching to achieve mastery criteria for the stimulus fading procedure, and completed both training procedures in the above outlined 12 steps. Larry acquired the selection of the picture (AB) and printed word (AC) in the presence of the spoken name in 12 blocks each. Conversely, he did not meet mastery criteria using only stimulus fading procedures; when required to select the correct item/activity at step 12 when only the written word was presented, Larry’s performance decreased significantly. Per the outlined error correction procedures which moved Larry to the previously learned step, he was returned to step 11 (5% picture shown), during which time, he again selected the correct item or activity for at least 8 of 9 trials within two blocks. Larry was then provided with an additional fading step, during which 2.5% of the picture was shown with the printed word superimposed. While Larry met mastery criteria for this step, performance again decreased below criterion levels when Larry was 42 presented with only the printed word, and was required to select the corresponding stimulus. An additional modification for Larry included the presentation of the 2.5% picture, with the superimposed printed word in colored text, with each stimulus (i.e., printed word) printed in a different color (e.g., blue, red, black). This measure was implemented based on the hypothesis that Larry was selecting activities/items when viewing the small percentage of the colored picture remaining on the stimulus card. If Larry was selecting based on picture , reinforcing selection responses in the presence of colored text and picture, with one color assigned to each stimulus, could have facilitated correct responding when pictures were removed and only the printed word was present. When pictures were faded and only the colored printed word was presented however, Larry’s performance was again observed to decrease below criterion levels. Larry had already met performance criteria for set CDT, and continued to select stimuli corresponding to the dictated name. As such, the stimuli corresponding to Set SF were then taught via conditional discrimination training using stimulus fading. For example, Larry was presented with the written word and the corresponding dictated name of the stimulus and then required to select the positive comparison from an array of actual items/activities. Following the introduction of this procedure, Larry achieved mastery criteria within 81 trials (i.e., nine blocks). 43 Figure 4. Block – by – block acquisition data for Laura, Virginia and Larry. * indicate mastery criterion achieved for given relation or teaching procedure (i.e., CDT - AB, CDT - AC, or SF). 44 Testing Textual Control Figure 5 depicts performance on pre and posttests of textual control for Laura, Virginia and Larry. These tests measured the percentage of correct schedule following responses made in the presence of printed words. Light grey bars represent pretest scores, while charcoal bars represent posttest scores. Exact scores for pre and posttests are located directly above the corresponding bar. Pre and posttest percentages of correct responding in the presence of printed words are separately indicated for stimuli trained via both conditional discrimination training and superimposition and fading. During pretesting, Laura did not follow a textual/printed schedule. Both training conditions resulted in derived textual schedule following. Laura correctly completed 94% of steps required to follow a printed schedule following stimulus fading, and 88% of steps following conditional discrimination. Virginia did not follow a schedule consisting of printed words during pretests of textual control. She correctly completed 100% of steps required (i.e., referencing each presented activity card, selecting and completing the activity corresponding to the card, returning to the schedule, removing the activity card, and repeating the process for all subsequent activities) to follow a schedule consisting of printed words following superimposition and fading procedures, and also completed 93% of steps required to follow a textual schedule following conditional discrimination training. Both Laura and Virginia demonstrated similarly accurate scores for derived textual control following exposure to both procedures. 45 Figure 5. Pre and posttests of textual control for Laura, Virginia and Larry. Scores are grouped by training procedure. 46 Larry did not follow a textual/written activity schedule following during pretests of textual control. In contrast with the similar scores of derived textual control across training methods demonstrated by both Laura and Virginia, Larry showed a greater percentage of correct responses on tests of textual control after exposure to conditional discrimination training (100% correct responding across activities) than following superimposition and fading (22% correct responding across activities). Textual Behavior/Reading Figure 6 contains pre and posttest scores of textual behavior (CD) for all participants. These tests measure the percentage of correct sight word reading responses. Light grey bars represent scores on pretest 1. Medium grey bars represent scores for pretest 2. Pretest 2 was conducted if participants performed above chance levels during pretest 1. No participant performed above chance levels on pretest of textual behavior. Solid black bars represent posttest scores and dark grey bars represent posttest 2 scores (Laura only). Scores are presented with emergent stimulus relations tests, with each relation presented separately for each teaching procedure. During pretests, Laura did not accurately read printed words for either set of stimuli. Following conditional discrimination training, Laura read printed words with 100% accuracy. She did not read printed words following the superimposition and fading procedure. During pretests of textual behavior, Virginia did not read words accurately. Following conditional discrimination training, she accurately read printed words (CD) on all posttest trials. Virginia accurately read printed words aloud on 67% of trials during posttesting for stimuli taught via the superimposition and fading procedure. Larry did not 47 read words during pretests including stimuli for either Set SF or Set CDT. He reliably read words (i.e., demonstrated textual behavior) for 89% of trials following conditional discrimination as compared to 11% of trials following stimulus fading. Emergent Relations Testing Figure 6 contains pre and posttest scores of emergent stimulus relations for all participants. These tests measured percentage of correct matches to sample for BC and CB relations. Light grey bars represent scores on pretest 1. Medium grey bars represent scores for pretest 2 (Larry and Virginia). Solid black bars represent posttest scores and dark grey bars represent posttest 2 scores (Laura only). Scores for single blocks of tests of matching picture to printed word (BC) and printed word to picture (CB). Each relation (i.e., BC, CB) is presented separately for the corresponding training procedure. During pretests, Laura scored between 22% and 33% on matching pictures to printed words (BC) and matching printed words to pictures (CB) for both Set CDT and Set SF, respectively. During posttests, Laura matched picture to printed words and printed words to pictures with 100% accuracy following conditional discrimination training. Laura matched pictures to printed words with 100% accuracy, and printed words to pictures with 77% accuracy after exposure to stimulus fading. Virginia matched pictures to printed words with 33% accuracy during pretest for stimuli assigned to both training conditions. For stimuli taught via conditional discrimination training, she initially showed 44% accuracy matching printed words to pictures, but a second pretest showed only 22% accuracy, indicating that accuracy of pretest responses was likely due to chance. Pretest scores of matching printed words to 48 pictures for set SF also indicated chance responding (i.e., 33% accuracy). Following conditional discrimination training, Virginia matched pictures to printed words (BC) and vice versa (CB) with 100% accuracy. Virginia’s data also show similar results for Set SF; she matched pictures to printed words with 100% accuracy and matched printed words to pictures with 89% during posttesting for stimuli taught via the superimposition and fading procedure. Larry matched pictures to printed words (BC) and printed words to pictures (CB) with 33% accuracy during pretests of stimulus relations for stimuli trained via superimposition and fading. For stimuli trained via conditional discrimination, he matched pictures to printed words (BC) with 44% accuracy, and matched printed words to pictures (CB) with 67% accuracy during the initial pretest. A second pretest of this relation yielded scores of 33%, indicating chance responding. During the first pretest of this relation, Larry’s first three responses corresponding to a particular position in the stimulus array (i.e., left side) regardless of the picture placed in that location. Following the first three responses during this test, he then allocated all subsequent pretest responses to the middle position of the stimulus array. Following training, Larry matched pictures to printed words (BC) and vice versa (CB) with 100% accuracy across stimuli and training procedures. 49 Figure 6. Pre and posttest scores for measures of stimulus relations. Relations tested include matching picture to printed word (BC), printed word to picture (CB) and reading printed words aloud (CD), and are grouped by training procedure. 50 Chapter 4 DISCUSSION The current study replicates and extends previous research related to establishing textual control over responding in activity schedules. Specifically, the current study replicated the work of Miguel et al (2009) by showing the emergence of derived textual control, derived textual behavior and derived stimulus relations resulting from conditional discrimination training. The findings of Birkan et al (2007) were extended to show that superimposition and fading alone (i.e., in the absence of vocal verbal behavior by the participant) sometimes, though not always, results in textual control through direct training. Laura and Virginia both followed written/textual schedules with a slightly higher percentage of correct responses during posttests of textual control following superimposition and fading as compared to conditional discrimination training. For Laura, 6% more accuracy during posttests was observed for Set SF than for Set CDT. Similarly, Virginia showed 7% greater accuracy of step completion for Set SF than for Set CDT. Larry however, made far more correct responses during posttests following conditional discrimination training. He responded correctly during posttests of textual control to 78% more steps for Set CDT than for Set SF. Laura and Virginia’s results indicate that both superimposition and fading and conditional discrimination training resulted in derived stimulus control over responding in the context of activity schedules, which consisted of schedule following behavior in the presence of printed words only. 51 Conversely, Larry’s results indicate that superimposition and fading did not result in textual control over schedule following. Larry’s results indicate the superiority of conditional discrimination training related to textual control. In terms of trials required to meet mastery criteria and number of errors made, Laura and Virginia’s data again demonstrate the similarity of both procedures. During training, Laura required slightly fewer trials (i.e., four blocks less) to meet mastery criteria during superimposition and fading procedure than during conditional discrimination training. Conversely, Virginia required fewer trials (i.e., three blocks less) to meet criterion for conditional discrimination training. During training, Larry required nearly three times as many trials to meet criteria for Set SF (i.e., 67 blocks) than for Set CDT (i.e., 24 blocks). Emergent relations posttest performance showed that participants could match pictures to printed words and vice versa. Conditional discrimination training produced posttest results indicating that all stimuli belonged to an equivalence class. In addition, posttests showed the emergence of textual behavior (i.e., reading, or relation CD) across participants. Superimposition and fading resulted in the correct selection of printed words in the presence of corresponding pictures (BC) across all participants. The correct selection of pictures in the presence of printed words (CB) was also observed, though less consistently across participants (Laura – 77%, Virginia – 89% and Larry – 100%). Textual behavior or reading (CD), however, rarely emerged following superimposition and fading (Laura – 0% and Larry – 11%), with the exception of Virginia, who demonstrated some reading follow this procedure (67% correct responses). For Virginia 52 especially, these procedures were equally effective in establishing derived stimulus relations. Following both training procedures, she correctly related pictures to printed words and vice versa, and also read printed words, though textual behavior was demonstrated at a lesser extent following superimposition and fading. Laura also conditionally related pictures to printed words and printed words to pictures. She read printed words following conditional discrimination training, but was not able to do so following stimulus fading. Thus, conditional discrimination training produced BC, CB and CD (i.e., reading) performances, and superimposition and fading produced BC and CB performances, but not CD for Laura. As such, conditional discrimination training was determined to be slightly superior in terms of producing textual behavior. Larry’s results indicate that conditional discrimination training was far superior to superimposition and fading, in terms of trials to criterion, number of errors, and derived textual control. Following exposure to conditional discrimination procedures, he conditionally related pictures to printed words and vice versa, read written words, and followed written activity schedules. Conversely, superimposition and fading did not result in derived textual control, and while Larry matched picture to printed words and vice versa following this procedure, he did not read the printed words. It is possible that Larry was able to match picture to printed words and vice versa following stimulus fading simply due to the repeated pairings of pictorial and textual stimuli. Following exposure to over 600 trials in which written words and pictures were paired, it is possible that Larry was able to match pictures to printed words and vice versa solely as a result of this extensive pairing. This would support the findings of previous research that has 53 shown that when selection of stimulus compounds (i.e., pairs) results in the formation of a stimulus class, and components of those compounds are then recombined to form novel compounds, participants reliably select new compounds comprised of members of the original class (Debert, Matos & McIlvane, 2007). Larry’s results also indicate that the formation of stimulus classes does not necessarily result in the formation of functional classes. While his posttest results indicate that Larry matched pictures and printed words, he did not follow a schedule comprised of printed words. When stimuli are equivalent or are members of the same equivalence class, they are substitutable for each other within a given context (Sidman, 1994). While stimulus equivalence can be demonstrated via matching of pictures and printed words, these stimuli are not necessarily functionally equivalent (Catania, 2007). Functional equivalence among stimuli indicates that each stimulus belonging to an equivalence class will exert the same control over responding via direct or derived means (Goldiamond, 1962). Laura and Virginia’s results support previous research that indicates that training a particular response in the presence of one member of an equivalence class results in the derived control over that response by another member of the class (Wirth & Chase, 2002). Larry’s results, however, contrast this finding. When Larry successfully matched pictures to printed words and printed words to pictures in emergent relations posttests, he demonstrated that these stimuli were members of the same equivalence class. He also demonstrated the functional equivalence of the stimuli trained via conditional discrimination training when he showed textual control of responding in activity schedule completion. He did not, however, demonstrate textual control for stimuli for stimuli trained via superimposition and fading, even though 54 he was able to match pictures and printed words for theses stimuli. These results indicate that even though stimuli became equivalent or members of the same equivalence class they did not serve the same function. These results support earlier findings that suggest that in some cases, transfer of functional control may depend on the difficulty or complexity of the task over which functional control is to be transferred (McVeigh & Keenan, 2009). In the current study, participants tacted items/activities prior to training and played in the presence of picture activity schedules following teaching of activity schedule following. During conditional discrimination training, participants were taught to select pictures and printed words in the presence of dictated names. Posttests examined the emergence of textual behavior and the transformation of function of printed words. Results showed that conditional discrimination training resulted in this transfer of function through membership in an established equivalence class. Specifically, in the absence of stimulus pairing, printed words evoked the same response (i.e., schedule following) that had previously been evoked by other members (i.e., pictures) of the same stimulus class. In superimposition and fading procedures, participants who had previously been taught to follow schedules in the presence of pictures were trained to select items when presented with stimuli comprised of both pictures and superimposed printed words. Pictures were extensively paired with printed words during training, with pictures being slowly faded. Following this pairing, two of three participants followed textual activity schedules. Rather than indicate derived functional transfer, these results show that repeated pairings combined with the fading of picture prompts produced 55 textual schedule following through direct training. Equivalence classes were formed via compound stimulus training (Debert, et al, 2007); the repeated presentation of stimulus pairs (i.e., pictures and printed words) led to matching corresponding printed words and pictures when presented separately. In summary, conditional discrimination training procedures established derived textual control, textual behavior and the formation of equivalence classes across participants. Stimulus fading procedures produced similar results in terms of stimulus equivalence, and directly established textual control for two participants. Textual behavior, however, was not reliably established following this procedure. The results of the current study indicate that conditional discrimination training may be superior to superimposition and fading procedures in terms of amount of derived relations. Limitations While the current study demonstrates strong experimental control, some limitations must be considered. Firstly, while Larry eventually met criterion for the superimposition and fading procedure, he required multiple trials and troubleshooting measures to do so, such as an additional fading step and the use of colored text. Even with the additional fading steps, Larry did not demonstrate criterion performance in the absence of the dictated name of each activity/item by the experimenter. Only after Larry was provided with the dictated name as well as the printed word of the sample stimulus was he able to accurately select the item/activity in the absence of the picture. Since Larry’s posttests were completed following mastery of Set SF using multiple teaching procedures, the effects of the stimulus fading procedure cannot be separated from those 56 of conditional discrimination training. Due to the use of multiple teaching procedures, the results of Larry’s posttests should be interpreted with caution in regards to stimuli associated with the superimposition and fading procedure. It is possible that all derived relations, including matching pictures to printed words, printed words to pictures and reading printed words resulted from the brief (i.e., nine block) exposure to the dictated name by the experimenter provided immediately prior to posttesting. As a possible control for this in future research, if similar performances are observed, posttests of textual control, textual behavior and emergent relations should be conducted prior to initiating conditional discrimination training. After completion of conditional discrimination training, a second posttest of textual control, textual behavior and emergent relations should be conducted. This would allow researchers to observe the effects of the superimposition and fading procedure separately from those of conditional discrimination. A second limitation that should be noted is the additional learning that may have occurred in the participants’ natural environments. It is possible that additional learning, such as sight word reading, letter recognition, activity/item naming or other skills could have been acquired through other teaching methods. For example, Virginia attended a mainstream preschool class and might have been exposed to some of the written words used during the various procedures over the course of the study, which could account for the elevated performance in posttests of textual behavior following the superimposition and fading procedure that was observed in Virginia only. Both Laura and Larry did not demonstrate textual behavior following this procedure, but Virginia did for 67% of trials. 57 Further limitations include both the number of steps and number of stimuli required to complete the superimposition and fading procedure in this study. While 12 steps were developed to counterbalance the number of steps required to complete each procedure, there were more steps than would typically occur in a clinical environment. Each step required separate stimuli, which resulted in a larger number of stimuli required to complete the superimposition and fading procedure. In addition, no criteria for independence probes were implemented for either procedure. It may be possible that all participants would have met mastery criteria more rapidly for both procedures with such criteria in place. Anecdotally, all participants began responding correctly prior to the prescribed prompt when prompts were delayed by only two seconds in the conditional discrimination training. This means that independent responding was observed for at least 8 out of 9 trials by the third prescribed step of this training procedure. It is possible, that with criteria for independence probes in place, one procedure would have been shown to be far superior to the other in terms of trials required to meet mastery criteria. A final factor to consider includes each participant’s learning history. Considerations such as each participant’s reinforcement history and experience with a particular teaching strategy may have affected their performance in the current study (Pipkin & Vollmer, 2009; Coon, 2010), especially for those participants who had an extensive history with learning via methods of applied behavior analysis. For example, the participants in the current study had varying amounts of exposure to teaching. Larry had extensively participated in behaviorally based interventions for over 7.5 years at the start of the study. Laura and Virginia had participated in behaviorally based interventions 58 for nearly 6.5 and less than one year, respectively, prior to the start of the study. Without a thorough review of previous teaching methods, it is impossible to determine each participant’s experience with a particular teaching methodology. Larry’s results in particular suggest that extra-experimental history may have had an effect on his performance in the current study. Future Research Future research will surely seek to replicate the current study. In addition, future research should examine the effects of a prompt fading procedure more likely to be encountered in clinical practice. The prompt fading procedure in the current study did not include criteria for probing of independence of either teaching procedure. While this was done intentionally to eliminate any potential confounds related to differences in requirements to meet mastery criterion, it does not resemble independence probes common to applied practice. Future replications of the current study might seek to probe independence for one method (e.g., superimposition and fading) when independence is demonstrated for the other (e.g., conditional discrimination training). Such independence might be considered as demonstrated when participants respond correctly prior to the prescribed prompt on a given number of trials within a nine trial block. Additionally, future research should seek to further clarify the relationship between stimulus and functional classes by comparing conditional discrimination training and stimulus fading procedures in terms of established functional and stimulus equivalence Future research should examine the transfer of control by members of equivalence classes, in addition to the transfer of function of systematically more complex tasks (e.g., 59 two item schedules versus five item schedules). A more detailed understanding of the transfer of control over various types of responding would further assist practitioners in modifying teaching methods to maximize learning and derived control as well as derived stimulus relations. Applied Implications The current study has multiple implications for applied practice. Of primary importance is the finding that, for two of three participants, both conditional discrimination training and superimposition and fading procedures produce textual control within activity schedules. For children with autism, use of a less intrusive, more flexible and less potentially socially stigmatizing schedule (e.g., written schedule) is preferred. Both procedures (i.e., CDT and SF) examined in this study can establish following such as schedule, though performance may vary based on student’s learning history. If a student has a strong reinforcement history with one type of procedure (e.g., stimulus pairing versus conditional discrimination training), practitioners would do well to teach via methods that complement and/or capitalize on that history. Both teaching procedures produced picture and printed word matching, though performance differed slightly across participants and training procedures, with conditional discrimination training indicated as slightly superior. Additionally, the behavioral mechanism behind these matching skills varies across procedures. Conditional discrimination training produced derived stimulus relations, while relations shown during posttests following stimulus fading procedures were directly taught via repeated stimulus pairings. Also, practitioners striving to establish vocal verbal 60 repertoires in their students should prioritize teaching via conditional discrimination training, as that procedure reliably produced textual behavior (i.e., reading) while superimposition and fading did not. As a final note, practitioners should develop programming using teaching methods that maximize learning in the least amount of time, with the greatest potential for resultant derived skills, while taking individual learning and reinforcement histories into consideration. 61 APPENDIX A Sample Data Sheets Figure A1. Sample data sheet for activity schedule training, and pre and posttests of textual activity schedule following (i.e., textual control). 62 Figure A2. Sample data sheet used for conditional discrimination training, and pre and posttests of stimulus relations (BC) and (CB). 63 Figure A3. Sample data sheet used for training with superimposition and fading procedure. 64 Figure A4. Sample data sheet used for testing of textual behavior (i.e., relation CD). 65 APPENDIX B Modified RAISD Modified RAISD – to be completed by parents/caregivers prior to the beginning of the study Preferred Items Assessment The purpose of this survey is to obtain information about the foods that you believe would be useful as rewards for your child. If you have signed the consent form, please answer the following questions regarding your child’s preferences: 1. Some children really enjoy foods like crackers, chips, pretzels, cereal, cookies, candy, etc. What are the specific foods your child likes to eat the most? _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ 2. Please go back to the list in question number one, and place a number in each box to rank these foods from most favorite (1) to least favorite (10). 66 3. Are there any foods that your child is allergic to? _____________________________ ________________________________________________________________________ 4. Are there any foods that you would prefer your child not eat during our study? ________________________________________________________________________ ________________________________________________________________________ 5. Some children really enjoy playing with toys like blocks, books, musical toys, figurines, balls, puzzles, toy cars, etc. What are the specific toys your child likes to play with? _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ _____________________ 6. Please go back to the list in question number one, and place a number in each box to rank these foods from most favorite (1) to least favorite (10). 67 APPENDIX C Stimulus Sets for Participants Table 5. Stimulus Sets per Participant Participant Training Condition ______________________________________________________ CDT Printed Word Laura Virginia Larry SF Picture Printed Word coloring sticker puzzle cutting books paste machine music sing stickers drawing puzzle DS toys blocks books dinosaur puzzles Picture 68 REFERENCES American Psychological Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Anderson, M. D., Sherman, J. A., Sheldon, J. B. & McAdam, D. (1997). Picture activity schedules and engagement of adults with mental retardation in a group home. Research in Developmental Disabilities, 18, 231-250. Barlow, D. H. & Hayes, S. C. (1979). Alternating treatments design: One strategy for comparing the effects of two treatments on a single behavior. Journal of Applied Behavior Analysis, 12, 199-210. Birkan, B., McClannahan, L. E., & Krantz, P. J. (2007). Effects of superimposition and fading on the sight-word reading of a boy with autism. Research in Autism Spectrum Disorders, 1, 117-125. Bondy, A & Frost, L. (2002). The picture exchange communication system. Newark, DE: Pryamid Education Products. Bryan, L. C., & Gast, D. L. (2000). Teaching on-task and on-schedule behaviors to highfunctioning children with autism via picture activity schedules. Journal of Autism and Developmental Disorders, 30, 553 – 567. Catania, A.C. (2007). Learning (4th ed.). Cornwall-on-Hudson, NY: Sloan. Coon, J. T. (2010). The role of increased exposure to transfer of stimulus control procedures on the acquisition of intraverbals behavior. Unpublished master’s thesis, California State University, Sacramento. 69 Corey, J. R. & Shamow, J. (1972). The effects of fading on the acquisition and retention of oral reading. Journal of Applied Behavior Analysis, 5, 311-315. Debert, P., Matos, M. A. & McIlvane, W. (2007). Conditional relations with compound abstract stimuli using a go/no-go procedure. Journal of the Experimental Analysis of Behavior, 87, 89-96. Dinsmoor, J. A. (1985). The role of observation and attention in establishing stimulus control. Journal of the Experimental Analysis of Behavior, 43, 365-381. Fisher, W. W., Piazza, C. C., Bowman, L. G., & Almari, A. (1996). Integrating caregiver report with a systematic choice assessment to enhance reinforcer identification. American Journal of Mental Retardation, 101, 15 – 25. Goldiamond, I. (1962). Perception. In A. J. Bachrach (Ed.) Experimental foundations of clinical psychology (pp. 280-340). New York: Basic Books. Higbee, T. S., Carr, J. E. & Harrison, C. D. (2000). Further evaluation of the multiple stimulus preference assessment. Research in Developmental Disabilities, 21, 6173. Horne, P. J. & Lowe, C. F. (1996). On the origins of naming and other symbolic behavior. Journal of the Experimental Analysis of Behavior, 65, 185-241. Karsten, A. M., & Carr, J. E. (2009). The effects of differential reinforcement of unprompted responding on the skill acquisition of children with autism. Journal of Applied Behavior Analysis, 42, 327 – 334. 70 Krantz, P. J., & McClannahan, L. E. (1998). Social interaction skills for children with autism: A script fading procedure for beginning readers. Journal of Applied Behavior Analysis, 31, 191 – 202. Lynch, D. C. & Cuvo, A. J. (1995). Stimulus equivalence instruction of fraction-decimal relations. Journal of Applied Behavior Analysis, 28, 115-126. MacDuff, G. S., Krantz, P. J., & McClannahan, L. E. (1993). Teaching children with autism to use photographic activity schedules: Maintenance and generalization of complex response chains. Journal of Applied Behavior Analysis, 26, 89-97. Massey, N. G., & Wheeler, J. J. (2000). Acquisition and generalization of activity schedules and their effects on task engagement in a young child with autism in a pre-school classroom. Education and Training in Mental Retardation and Developmental Disabilities, 35, 326 – 335. McClannahan, L. E., & Krantz, P. J. (1999). Activity schedules for children with autism: Teaching independent behavior. Bethesda, MD: Woodbine House. McClannahan, L. E., & Krantz, P. J. (2010). Activity schedules for children with autism: Teaching independent behavior (2nd ed.). Bethesda, MD: Woodbine House. McClannahan, L. E., MacDuff, G. S., & Krantz, P. J. (2002). Behavior analysis and intervention for adults with autism. Behavior Modification, 26, 9-27. McVeigh, V. & Keenan, M. (2009). Multiple functions in equivalence classes. The Psychological Record, 59, 93-118. Michael, J. L. (2004). Concepts and principles of behavior analysis. Kalamazoo, MI: Association for Behavior Analysis International. 71 Miguel, C. F., Yang, H., Finn, H. & Ahearn, W. (2009) Establishing derived textual control in activity schedules with children with autism. Journal of Applied Behavior Analysis, 42, 703-709. Mueller, M. M., Palkovic, C. M., & Maynard, C. S. (2007). Errorless learning: Review and practical application for teaching children with pervasive development disorders. Psychology in the Schools, 44, 691-700. Pipkin, C. S. P. & Vollmer, T. R. (2009). Applied implications of reinforcement history effects. Journal of Applied Behavior Analysis, 42, 83-103. Rehfeldt, R. A., & Root, S. L. (2005). Establishing derived requesting skills in adults with severe developmental disabilities. Journal of Applied Behavior Analysis, 38, 101-105. Rosales, R., & Rehfeldt, R. A. (2007). Contriving transitive conditioned establishing operations to establish derived manding skills in adults with sever developmental disabilities. Journal of Applied Behavior Analysis, 40, 105-121. Sidman, M. (1971). Reading and auditory-visual equivalences. Journal of Speech and Hearing Research, 14, 5-13. Sidman, M. (1994). Equivalence relations: A research story. Boston, MA: Authors Cooperative. Sidman, M., Wynne, C. K., Maguire, R. W., & Barnes, T. (1989). Functional classes and equivalence relations. Journal of the Experimental Analysis of Behavior, 52, 261274. Skinner, B. F. (1957). Verbal Behavior. Acton, MA: Copley Publishing. 72 Toussaint, K. A. & Tiger, J. H. (2010). Teaching early braille literacy skills within a stimulus equivalence paradigm to children with degenerative visual impairments. Journal of Applied Behavior Analysis, 43, 181-194. Vause, T., Martin, G. L., Yu, C. T., Marion, C. & Sakko, G. (2005). Teaching equivalence relations to individuals with minimal verbal repertoires: Are visual and auditory-visual discriminations predictive of stimulus equivalence? The Psychological Record, 55, 197-218. Walker, B. D., Rehfeldt, R. A. & Ninness, C. (2010). Using the stimulus equivalence paradigm to teach course material in an undergraduate rehabilitation course. Journal of Applied Behavior Analysis, 43, 615-633. Wirth, O. & Chase, P. N. (2002). Stability of functional equivalence and stimulus equivalence: Effects of baseline reversals. Journal of the Experimental Analysis of Behavior, 77, 29-47. Zimbelman, M., Paschal, A., Hawley, S. R., Molgaard, C. A. & St.Romain, T. (2006). Addressing physical inactivity among developmentally disabled students through visual schedules and social stories. Research in Developmental Disabilities, 28, 386-396.