ASSESSING PREFERENCE FOR TYPES OF SOCIAL INTER A CTION By SAMUEL L. MORRIS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 201 9
Â© 201 9 Samuel L. Morris
3 ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Timothy Vollmer, for guiding this research. I would also like to thank the other members of my committee: Dr. Iser DeLeon and Dr. Maureen Conroy . I also would like to thank my colleagues in the Vollmer Lab and the staf f at the Florida Autism Centers, who helped me in various ways throughout this project.
4 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 3 LIST OF TABLES ................................ ................................ ................................ ........................... 5 LIST OF FIGURES ................................ ................................ ................................ ......................... 6 A BSTRACT ................................ ................................ ................................ ................................ ..... 7 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................... 8 2 METHOD ................................ ................................ ................................ ............................... 14 Participants and Setting ................................ ................................ ................................ .......... 14 Response Measurement and Reliability ................................ ................................ .................. 14 Procedure ................................ ................................ ................................ ................................ 15 3 RESULTS ................................ ................................ ................................ ............................... 21 Pre Test ................................ ................................ ................................ ................................ ... 21 Social Interaction Preference Assessments ................................ ................................ ............ 21 Concurrent Operant Reinforcer Assessments ................................ ................................ ......... 23 4 DISCUSSION ................................ ................................ ................................ ......................... 26 APPENDIX: TABLES AND FIGURES ................................ ................................ ....................... 31 LIST OF RE FERENCES ................................ ................................ ................................ ............... 34 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ......... 36
5 LIST OF TABLES Table page A 1 Correspondence between the SIPA and Reinforcer Assessment ................................ ....... 31
6 LIST OF FIGURES Figure page A 1 Results of the Social Interaction Preference Assessment . ................................ ................. 32 A 2 Results of the Reinforcer Assessment ................................ ................................ ................ 33
7 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ASSESSING PREFERENCE FOR TYPES OF SOCIAL INTERACTION By Samuel L. Morris Ma y 201 9 Chair: Timothy R. Vollmer Major: Psychology To date, few researchers have evaluated methods for assessing preference for social interactions. Due to concerns that commonly used stimulus preference assessment methods may be inappropriate, or at least cumbersome, for the assessment of social reinforcement, we developed and evaluated a new method of assessing pre ference for social interactions. A Social Interaction Preference Assessment (SIPA) and a concurrent operant reinforcer assessment were conducted with 5 participants diagnosed with autism spectrum disorder. A differentially preferred and reinforcing social interaction was identified for all 5 participants. The SIPA procedures, results, and the implications of these results are discussed.
8 CHAPTER 1 INTRODUCTION Methods of identifying reinforcing stimuli have proven to be useful in working with individuals who cannot effectively communicate their own preferences. Several variations of stimulus preference assessments have been developed (e.g., Pace, Ivancic, Edward s, Iwata, &Page 1985; Fisher et al., 1992; DeLeon & Iwata, 1996; Hanley, Iwata, Lindberg, & Conners, 2003). To date, the preference assessment literature has primarily focused on identifying leisure and edible stimuli. There has been relatively little rese various types of social interaction. There are likely several variables contributing to this discrepancy. First, the nature of social interactions is such that they cannot be placed on a surface and directly appr oached or selected (like edible or leisure items). However, potential solutions such as using picture cards (e.g., Kelly, Roscoe, Hanley, & Schlichenmeyer, 2014), therapists (e.g., Clay, Sa maha, Bloom, Bogoev, &Boyle, 2013) , or videos (e.g., Clark, Donalds on, & Kahng, 2015 ) as selection stimuli have been evaluated. Second, the idiosyncratic differences in preferred social interactions across participants increase the response effort involved with planning and conducting such assessments. Third, many partici pants in preference assessment research are diagnosed with autism spectrum disorder (ASD), which is characterized by a relative lack of interest in social interaction ( American Psychiatric Association, 2013) . Thus, some researchers may assume that individu als with ASD would not find social interaction appealing and, hence, may not include such stimuli in assessments. Fourth, it is possible that when individuals have generalized skill deficits with communication, the applicability of preference assessments f or social interaction is not immediately evident. Despite these potential difficulties, some researchers have in fact evaluated methods of identifying reinforcing social interactions.
9 orcing efficacy of different social interactions using a concurrent operant arrangement. Results showed that one participant allocated more time to playing with toys that resulted in reprimands compared to playing with toys that resulted in praise, and ano ther participant preferred tickles over praise. Smaby, MacDonald, Ahearn, and Dube (2007) conducted brief reinforcer assessments in which different social interactions were delivered across one minute sessions contingent on a single response. Although this method efficiently assessed if social interactions were reinforcing, it did not allow researchers to assess relative preference between those that were shown to be reinforcing. More recently, researchers have applied procedures from the edible and leisure preference assessment literature to assessing preference for different types of social interaction. Clay et al. (2013) conducted a paired stimulus preference assessment (PSPA) in which the stimuli were different therapists who each delivered a single type of social interaction during a given PSPA. Although the researchers identified reinforcing social interactions for all five participants, the use of therapists as the stimuli to be approached was resource intensive and required repeated preference assessm ents to test for idiosyncratic effects across different therapists. For the most part, however, such extension of the preference assessment literature has been accomplished by substituting picture card selection for the approach or selection responses that are typically recorded. For example, Kelly et al. (2014) assessed preferences for seven different types of social interaction with single stimulus preference assessments (SSPA) and PSPA and found that the PSPA more accurately and reliably identified prefe rence and relative reinforcing efficacy for the different types of social interaction for all participants. Nuernberger, Smith, Czapar, and Klatt (2012) used multiple stimulus without replacement (MSWO)
10 procedures to assess preference for several different types of social interaction and assessed their efficacy by delivering each type of social interaction contingent on a sorting task in separate sessions. The MSWO procedures produced a hierarchy of reinforcing social interactions for 2 of 3 participants. L ang et al. (2014) assessed preferences for different types of attention in PSPA and MSWO preference assessments. They found that both assessments produced the same highest and lowest preferred types of social interaction for all participants. These high pr eferred types of social interaction were shown to be relatively more reinforcing when provided as consequences in discrete trial training . The preference assessment studies previously discussed identified the relative reinforcing efficacies of different social interactions. Most of these studies required participants to discriminate between pictures of therapist participant interactions based on their previous experience with such stimuli or the different social interactions delivered as consequ ences for their selection (cf. Clay et al., 2013). Lang et al. (2014) used pre training sessions that lasted 20 and 40 min while the other studies (Nuernberger et al., 2012; Kelly et al., 2014) conducted exposures to the contingencies prior to the beginnin g of the assessment to aid the development of stimulus control. To reiterate, these procedures were adequate for the participants in the respective studies because the participants had previous history with the picture cards to be selected or because the p rocedures facilitated the development of stimulus control between these picture cards and the different social interactions delivered as consequences for their selection. However, there are two potential shortcomings of these procedures. First, conducting lengthy pre training sessions is potentially an inefficient method of developing or ensuring the stimulus control required by these assessments. Second, the more traditional alternative of conducting an exposure to the
11 contingencies prior to the assessment may prove inadequate for some individuals with intellectual and developmental disabilities (IDD). These procedures may be problematic, not due to a lack of social reinforcers, but because individuals simply lack a reinforcement history with the stimuli to be selected or the history required for such rapid development of stimulus control. However, it may be possible to increase the accuracy and accessibility of such preference assessments for social interaction by providing exposure to the contingencies thr oughout the assessment, allowing the individuals to emit more selection responses before a given stimulus and consequence are removed, and increasing the difference between the stimuli to be selected. Such changes could potentially increase the efficiency and accuracy of these types of preference assessments when conducted with some individuals with IDD. One preference assessment procedure that has yet to be applied to social interaction is the response restricted preference assessment (RRPA; Hanley et al., 2003). Researchers assessed preference for tangible items by measuring the percentage of 5 s intervals in which a participant interacted with any of the available stimuli in 5 min sessions. Items to which a participant consistently allocated the majority of responding were removed from the array contingent on meeting the criteria of predetermined rules. For example, if the participant interacted with an item for 60% or more of intervals in two consecutive sessions, that item was removed. Similar procedures c ould be applied to assessing preferences for types of social interaction by recording picture card selections instead of leisure item engagement . While it might be informative to operant type of arrangement, the previous research o n assessing preference for social interaction has shown that trial based assessments such as the MSWO and PSPA have been effective in ensuring or obtaining stimulus control. It may be useful to combine aspects of each of these assessments.
12 There are severa l potential benefits of identifying preferred types of social interaction. First, almost by diagnostic definition, individuals diagnosed with ASD could benefit from increased social interaction. Second, social reinforcers can supplement tangible and edible reinforcers and allow for greater reinforcer variation which may maintain more effortful or higher rates of responding (Egel, 1981). Third, social reinforcers are often more practical, socially acceptable, and readily available relative to edible and tang ible stimuli (DeLeon, Bullock, & Catania, 2013) and consequently may promote the generalization and maintenance of responses in the natural environment. Fourth, the use of social reinforcers could lead to a decrease in the consumption of high calorie foods (commonly used as reinforcers) that contain large amounts of sugar and salt. Fifth, social reinforcers could reduce the amount of time spent engaging in sedentary leisure activities such as watching videos (also commonly used as reinforcers). For these re asons, among others, methods of accurately and efficiently identifying preferred forms of social interaction for less advanced learners would be useful for caregivers, clinicians, and researchers. The purpose of the current study was to identify reinforci ng social interactions and to development of the social interaction preference assessment (SIPA) which combines aspects of a RRPA and a MSWO such that sessi ons are divided into trials, selections are recorded for each trial, and response options and their consequences are removed from the array based on the percentage of selections across consecutive sessions. In this assessment, exposure to the contingencies prior to each session and dissimilar stimuli were used to aid in the development of stimulus control if it was not already present. Picture cards were chosen as the stimuli to be selected because for practitioners they would require little time and resour ces to make, and
13 because several of the participants in the current study had an extensive history learning and communicating via picture card selection. The current study advanced previous research by evaluating an alternative method of assessing preferen ce for social interactions that may be more accurate and more accessible to individuals with less sophisticated communicative repertoires relative to those methods previously described.
14 CHAPTER 2 METHOD Participants and Setting Participants were five children diagnosed with ASD. Romeo was 8 years old and used a form of augmentative and alternative communication (AAC). Romeo had an extensive verbal instruction following repertoire, but often required gestural or environmental prompts to follow most instructions. Alice was 11 years old and used a form of AAC. Alice had a very limited receptive identification and instruction following repertoire. Avi was 4 years old; he had just begun using a form of AAC at the onset of the study. He had a very limite d repertoire of mands and instruction following skills. Beckham was 5 years old and communicated vocally with a limited repertoire of tacts, mands, and intraverbals. He occasionally spoke in sentences but often required prompting to do so. Joel was 5 years old and communicated vocally with a small repertoire of mostly mands. All sessions were conducted within an early intensive behavioral intervention (EIBI) clinic. Session rooms contained the necessary stimuli to conduct the assessment, a table, and chairs . Response Measurement and Reliabilit y In the pre responses. Data were used to calculate the percentage of selections on each stimulus by dividing the number of times a given stimulus was selected by the total number of selections possible per session. In reinforcer assessment sessions, observers collected data on the number of responses and then rate of responding was calculated by dividing the frequency of responding by the total session duration after subtracting the duration of each consequence delivered. A second observer independently collected data during 41% of total sessions. In the pre by trial basis.
15 Agreement was scored as having t he same selection written for a given trial of a given session. Interobserver agreement (IOA) was calculated by dividing the number of agreement trials by the tests and pre ference assessments. Reinforcer assessment sessions were divided into 10 s intervals and by interval basis. Agreement was defined as any interval in which both observers scored the same number of responses (comp leting the responses defined above or not responding). IOA was calculated by dividing the number of agreement intervals by the total number of intervals and multiplying by 100. IOA for the reinforcer assessment averaged 93.69% (range, 86.67 to 100). Proced ure Questionnaire Potentially reinforcing social interactions were identified for each participant through questionnaires given to clinicians who worked frequently with the participant. Questions were selected from the Reinforce r Assessment for Individuals with Severe Disabilities (RAISD ; Fisher, Piazza, Bowman, & Amari, 1996). Respondents were asked to provide detailed answers, but no other examples or prompts were delivered other than those provided in as a part of the questions. The most commonly reporte d distinct types of social interaction were included in the preference assessment. team were used to develop the definitions of each type of social interaction outlined above; here they will be listed as the simple names for each types of social interaction without the details rubs, and praise all functioned as reinforcers. Jo
16 as reinforcers. The number of reported social reinforcers ranged from 2 5 with an average of 3.6.For all participants, praise along with two types of social interaction f rom these reports were selected to be included in the preference assessment. The inclusion of similar interactions or two interactions that were not mutually exclusive was avoided (e.g., up and dips were not both included for Joel because both required pic king him up in a similar manner). Sequence of Assessments Three assessments were conducted with each participant in the current study. The pre test test determined if they would continue in the stud y. Second, a SIPA was conducted with all participants whose responding in the pre assessment met inclusion criteria. The SIPA provided a hierarchy of most to least preferred types of social interaction. Finally, a concurrent operant reinforcer assessment w as conducted with all participants. This assessment evaluated if social interactions would function as reinforcers for more effortful tasks and provided data that allowed for a clearer interpretation and validation of the SIPA results. Pre T es t A pre test was conducted to ensure that the participants could discriminate between picture cards that were different shapes and colors. Tangible reinforcers were delivered contingent on the selection of three picture cards, whereas there were no programmed consequen ces for selecting a fourth picture card. Once we observed differential selection of picture cards which received tangible reinforcers it was assured that each participant could learn
17 to discriminate between picture cards that were different shapes and colo rs; and we moved on to the SIPA. Social Interaction Preference Assessment types of social interaction. Each assessment consisted of several sessions and each session co nsisted of five trials. In each trial the participant was presented with an array of four stimuli in the delivery of one type of social interaction or no progra mmed consequence in the case of control cards. If no selection was made the experimenter waited 5 s and then re presented the array and instruction. If no selection was made within 5 s of re presentation no selection was recorded and the 5 s inter trial in terval began. (During this study, participants selected a picture card on the first presentation of every trial). After the selection, the array of stimuli was removed or covered while the relevant consequence was delivered. The array of stimuli was four different shapes, which were each a different color (red circle, blue square, yellow heart, and green triangle). The length and width of each shape was approximate 7.62 cm. Pictures (2.54cm X 2.54 cm) of an arrow pointing upward (up), a tornado (spins), a face (praise) were printed on the cards. A white square was placed in the middle of the control card (blue circle). Participants were prompted to select each card three times prior to the first session of that day and once prior to each subsequent session. The relevant consequence was delivered for selecting that card. The order each consequence was sampled was determined quasi randomly. The stimuli and their associated consequ ences were removed from the array after meeting the
18 criteria for one of four pre determined rules adapted from those described by Hanley et al. (2003). If the participant selected one stimulus for 80% or more of trials for two consecutive sessions or 60% o r more of trials for three consecutive sessions it was removed from the array (Rule 1). If selections were consistentl y and evenly distributed among two or more stimuli then those stimuli, and their consequences were removed from the array (Rule 2). Removal of the according to rules discussed above, until selections occurred on 20% or less of total trials for t w o consecutive sessions (Rule 3), or until the participant displayed a pattern of responding characterized as undiscriminated for more than five sessions (Rule 4). A less stringent version of Rule 1was used with the first two participants in this study (Rom eo and Alice). Stimuli and consequences were removed from their arrays contingent on selecting one stimulus for 60% or more of trials for two consecutive sessions. (Participants in this study only met criteria for Rule 1.) The types of social interaction i ncluded in the assessments included up, spins, tickles, and moving around the room (Beckham and Joel). Each instance of this social interaction lasted approximately 5 s. Spins were defined as the experimenter holding the participant parallel to the floor with both arms and slowly spinning around. Spins lasted appr oximately 10 s for Romeo and (Romeo, Avi, Beckham, and Joel), torso (Romeo, Beckham, and Joel), arms (Romeo, Alice, and Avi), and armpits (Joel). Tickles were deliv ered for 10 s for Romeo and 5 s for all other participants. Hugs were defined as the participant standing up while the experimenter turned and
19 Praise was includ delivery of several praise statements from a list of 10 statements that were used for all awes participants. During all interactions the experimenter attempted to make eye contact was made for at least part of the time and the facial expressions and tone typic ally associated with that type of social interaction accompanied it. Concurrent Operant Reinforcer Assessment After a hierarchy of preferences for the three types of social interaction was obtained from the SIPA a concurrent operant reinforcer assessment w as conducted. The concurrent operant reinforcer assessment served to validate the SIPAs accuracy in identifying the relative reinforcing efficacy of the different types of social interaction. Four similar response options were made concurrently available d were selected based on skill deficits or recommendations provided by clinicians who worked d as making a mark larger than a dot on a blank sheet of paper with one of four different colored markers. Participants were prompted to respond on each option prior to all sessions. Each response option was accompanied by coordinated color stimuli in an a ttempt to make discrimination between
20 them easier and the location of the four response options in the array varied were varied systematically across sessions. All reinforcer assessment sessions lasted 5 min. During the baseline sessions, four response opt ions were placed on the table in front of the participant and no consequences were associated with responding on any of the options. During the first reinforcement phase, the same four response options were available; three of the options were associated w ith each of the types of social interaction from the SIPA and one was associated with no programmed consequences. After the first reinforcement phase, two more were conducted in which the response option to which the participant had allocated the majority of their responding in the preceding phase was removed. Reinforcement phases were terminated and response options were removed based on visual analysis. For Alice, after the second reinforcement phase, we reversed back to the first reinforcement phase due to the low rates of responding observed in the second phase. The removal of each response option and its consequence allowed us to assess the accuracy of the hierarchy provided by the SIPAs as well as whether or not each interaction functioned as a reinfor assessment responses. Finally, a follow up reinforcer assessment session was conducted one month after the last reinforcer assessment session to assess whether preference had changed over time. The follow up session wa s conducted in the same manner as sessions in the first reinforcement phase. In the reinforcer assessment, social interactions provided contingent on responding lasted approximately the same amount of time as they did in the SIPA and were delivered accordi ng to the same definitions given above.
21 CHAPTER 3 RESULTS Pre Test All participants with whom the pre test was conducted met criteria for continuing in the study. The pre test results and more detailed procedures are available as supplemental materi al online. Social Interaction Preference Assessments Romeo Romeo SIPA results are depicted in the first panel of Figure A 1 . Romeo selected spins on 100% and 80% of trials during the first and second sessions, respectively. This met the criteria specified by Rule 1, so the spins picture card was removed from the array. He met the criteria of Rule 1 again on sessions 6 and 7 by selecting the tickles card on 100% and 80% of trials, respectively. Therefore tickles was removed. Finally, he selected the praise card for 60% and 80% of the eleventh and twelfth sessions, respectively. This met criteria for the less stringent version of Rule 1. The resulting preference hierarchy (highest to lowest) was spins, tickles, and praise. His SIPA consisted of 12 sessions, was conducted over three days, and lasted 39 min. Joel A 1. J oel selected the up card on 100% of trials in the third and fourth session of the preference assessment; therefore, it was removed (Rule 1). On sessions 5 and 6, he met the criteria of Rule 1 again by selecting the praise card on 100% of trials in both ses sions; therefore it was removed. During the final two sessions of the preference assessment, he selected the tickles card 100% and 80% of trials in the seventh and eighth sessions, respectively (Rule 1).The resulting preference hierarchy (highest to
22 lowest two days, and lasted 20.5 min. Avi SIPA results are depicted in the third panel of Figure A 1 . Avi selected the spins card for 100% and 80% of trials in the f irst and second preference assessment sessions, respectively. Therefore the spins card was removed (Rule 1). The same pattern of responding was observed for selection of the tickles card; he selected it for 100% and 80% of trials in sessions 3 and 4, respe ctively (Rule 1). Avi selected the praise card for 100% of trials during the fifth and sixth sessions of the SIPA; therefore it was removed. The resulting preference hierarchy sessions, was conducted in one day, and lasted 15.5 min. Beckham Figure A 1 . He met criteria for removal of the up card by selecting it for 60%, 100%, and 80% of sessions 3, 4, and 5 respectively (Rule 1). In sessions 6, 7, and 8, he met criteria again by selecting the tickles card for 80%, 60%, and 100% of trials in each session, respectively. One outcome that sets f the control card by selecting it for 80% of trials in sessions 12 and 13 (Rule 1). This suggested that not only was praise not preferred, but possibly aversive. His preference assessment results indicated that up was more preferred than tickles, and noth ing or no programmed consequences may be preferred to praise. The resulting preference hierarchy (highest to lowest) was up and tickles. Alice
23 ts are depicted in the fifth panel of Figure A 1 Alice selected the tickles card for 100% of trials in the first and second preference assessment sessions; therefore it was removed (Rule 1). In the fifth and sixth sessions, Alice selected the praise card f or 60% and 100% of trials in each session, respectively. This met criteria for the less stringent form of Rule 1. The SIPA was finished after the hugs card was removed after she selected it for 100% of trials in sessions 8 and 9 (Rule 1). The resulting pre ference hierarchy (highest to lowest) was tickles, lasted 22.5 min. Concurrent Operant Reinforcer Assessments Romeo wn in the top panel of Figure A 2 . During the reinforcer assessment, he allocated the majority of his responding to the bell associated with the highest preferred type of social interaction available, as identified by his SIPA. The results of the reinforce r assessment showed that being spun, tickled, and praised each would differentially reinforce his appropriate use of the bells. Joel. Figure A 2 . During the reinforcer assessment, he al located the majority of his responding to the marker associated with the highest preferred type of social interaction available, as identified by his SIPA. The results of the reinforcer assessment showed that being tossed in to the air or placed on the the sheet with markers. Avi
24 Figure A 2 . During the first two phases of the reinforcer assessme nt, he allocated the majority of his responding to the bin associated with the highest preferred type of social interaction available, as identified by his SIPA. The results of the reinforcer assessment showed that being spun and tickled would differential ly reinforce his cleaning up of the blocks. When only praise and control were left in the third reinforcement phase of the reinforcer assessment responding became undiscriminated between the response options suggesting that praise was not functioning as a reinforcer. Anecdotally, during this phase, we saw negative vocalizations and some escape behavior that had previously very rarely happened. Beckham Figure A 2 . During the first two phases of his reinforcer assessment, he allocated the majority of his responding to the group of beads associated with the highest preferred type of social interaction available as identified by his SIPA. The results of the reinforcer assessment verified t hat being picked up was more reinforcing than tickles and that both of these types of social interaction would reinforce putting beads on a string. The reinforcer assessment did not validate that praise was aversive. When only praise and control were left in the third reinforcement phase of the reinforcer assessment, responding became undiscriminated between the response options associated with praise and control. These data suggest that praise did not function as reinforcer for putting beads on a string. During the third reinforcement phase, we also observed a few instances of aggression, which was the only time that this behavior was observed. Alice
25 Figure A 2 . During the first phase of her reinforcer assessment, she allocated the majority of her responding to the highest preferred type of social interaction available, as identifie d by her SIPA. The results of the reinforcer assessment showed that being tickled would reinforce placing silverware into cylinder containers, but that praise and hugs would not. When the response option reinforced with tickles was removed in the second re inforcement phase, responding dropped to very low rates (2 total responses in sessions 14 and 15). When we reversed back to the first reinforcement phase, it was shown again that tickles reinforced sorting the silverware. Unlike Avi and te of responding decreased to near zero levels when no reinforcing social interactions were available.
26 CHAPTER 4 DISCUSSION A relatively more preferred and more reinforcing type of social interaction was identified for all participants, two differentially reinforcing social interactions were identified for four out of five participants (all except Alice), and three differential ly reinforcing social interactions were identified for Romeo and Joel. These outcomes were inferred from the SIPA and reinforcer assessment in combination, but could not have been obtained from the SIPA alone. Table A 1 displays the hierarchies produced by Although the SIPA produced a hierarchy of preference for social interactions for all participants, it was shown that one (Avi) or two (Alice) of these types of social interactions did not reinforce their responding in the reinforcer assessment despite having maintained selection responses in the SIPA. There may be several variables contributing to this type of discrepancy. Responses used in the reinforcer assessment were at least somewhat more effortful th an the selection of picture cards. It may be the case that some types of social interaction would reinforce very low effort responses such as picture card selection, but not more effortful responses. Skjoldager , Pierre, and Mittleman ( 1993) demonstrated th at increases in response effort were correlated with decreases in progressive ratio schedule breakpoints, and that more rapid decreases were obtained with relatively lower magnitude reinforcers. A similar effect may have occurred during the current study, but as a result of relative quality, rather than magnitude, of the reinforcers delivered. Additionally, we prompted our participants to respond before each trial in the preference assessment, which may have increased the likelihood that responding in this assessment was under instructional control and at least partially maintained by negative reinforcement. However,
27 consequences of picture card selection were also diffe rentially reinforcing and maintained responding to some extent. The number of times a given consequence was delivered within a session was much greater in the reinforcer assessment compared to the preference assessment, which could potentially lead to sati ation effects or the breakdown of contingencies in which praise or other more verbal types of social interaction function as discriminative stimuli rather than conditioned reinforcers (see Dozier, Iwata, Thomason Sassi, Worsdell, & Wilson , 2012 for a discu ssion). Despite these discrepancies, social interactions that reinforce responses equally as effortful as picture card selections may be useful in several ways. Examples could include training mands for social interactions, training low effort skills, or s upporting some skills that have already been acquired. For these reasons there may still be utility in the entire hierarchy produced by the SIPA even in the case that one or more of those social interactions do not maintain responding in a reinforcer asses sment context. The results of the current study have several implications. The SIPA and reinforcer assessment may be useful as a method for caregivers, clinicians, and researchers to identify reinforcing social interactions. Furthermore, the finding that s ocial interaction of various types functioned as social reinforcers for individuals diagnosed with ASD is potentially useful given the common characterization of children with ASD as being socially avoidant ( American Psychiatric Association, 2013). With th at said, the types of social interaction that did reinforce seen in prior research. For example, social interactions such as praise, high fives, smiles, fist bumps, and thumbs up were included in preference assessments and consistently selected by participants across several preference assessments in previous studies (e.g., Kelly et al., 2014; Lang et al., 2014). On the other hand, being tickled was the only c onsequence shown to be
28 preferred and reinforcing for more than half the participants in the current study. Additionally, similar types of potential social reinforcers were identified for all participants in the current study during the RAISD conducted with members of their clinical teams. Potential social reinforcers could simply be the most common social interactions delivered to individuals attending an EIBI clinic, or it may be related to the questions and the examples provided within them (e.g., hug, sa effortful and require more time, there is utility in attempting to identify reinforcing social interactions for learners with less well developed communicative reper toires and few reported social reinforcers. Although the goal of the current study was very similar to that of previous studies on social reinforcement, we used a less traditional approach. Based on previous experience with these participants and the repo rts of clinicians it seemed that the procedures described in previous research would be inefficient and potentially inaccurate if used with our participants. As a result of these assumptions, the current study used exposure to the contingencies throughout the assessment to develop or ensure stimulus control while also requiring a degree of consistency in responding before a given picture card and consequence would be removed from the array. The inclusions of these atypical procedures necessitated the develo pment of the SIPA. Due to the several ways in which the current study differs from previous research, the results of this study and conclusions derived from them should be interpreted as preliminary until adequate replications have been conducted. There a re also several limitations to the current study. While we assessed the validity of the SIPA results in the reinforcer assessment, we did not assess the test retest reliability of the SIPAs. Also, the SIPAs conducted in the current study only included thre e types of social
29 interaction and a control. While this was adequate for the participants in this study it may be useful to include more types of social interaction for future participants. In the current study we only conducted SIPAs and did not conduct P SPAs or MSWOs as a point of comparison or reference. The results and limitations of the current study suggest several directions for future research. Future research could evaluate the reliability of the SIPA; evaluate the effects of expanding the array of stimuli to include more social interactions; and compare the efficiency, social interaction. It seems likely that the result of such comparisons will depend on the characteristics and repertoires of the participants with whom it is conducted. For all participants in the current study, at least one type of social interaction reinforced responding during their reinforcer assessment. Although these responses were more effortful than picture card selection, future research should evaluate the efficacy of such social interactions when delivered contingent on more effortful responses. Specifically, evaluating the utility of social interactions as consequences during skill acquisition seems particularly useful, as this may ultimately be the type of application used by most therapists. Future research evaluating variations in the rules used to restrict response options, the amount of forced exposure, the duration of con sequences, and the number of trials per session could all prove useful. It is possible that extensions and modifications could make the SIPA more appropriate for certain populations and generally more accurate, reliable, and efficient. Additionally, due to the clinical relevance of identifying reinforcing social interactions and increasing the number and efficacy of such interactions for individuals with ASD, it may be beneficial to conduct such assessments on a longitudinal basis. When conducted across tim e, these types of assessments could function as a method of
30 evaluating how preference changes over time (e.g. if another type of interaction replaces being picked up when the child becomes too old or too heavy), progress in the number and potency of social reinforcers, the outcomes of specific interventions, or some combination.
31 A PPENDIX T ABLES AND FIGURES T able A 1. Correspondence between the SIPA and Reinforcer Assessment Participants SIPA Reinforcer Assessment Romeo Spins, Tickles, Praise Spins, Tickles, Praise Joel Up, Praise, Tickles Up, Praise, Tickles Beckham Up, Tickles Up, Tickles Avi Spins, Tickles, Praise Spins, Tickles Alice Tickles, Praise, Hugs Tickles
32 Figure A 1. The results of the social interaction preference assessment are shown (from top to bottom) for Romeo, Joel, Avi, Beckham, and Alice.
33 Figure A 2. The results of the reinforcer assessment are shown (from top to bottom) for Romeo, Joel, Avi, Beckham, and Alice. Dashed lines denote points at which response options were removed or added back in.
34 LIST OF REFERENCES American Psychiatric Association. (20 13). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Association. Clark, D. R., Donaldson, J. M., & Kahng, S. (2015). Are video based preference assessments without access to selected stimuli effective? Journal of A pplied B ehavior A nalysis , 48 , 895 900. doi : 10.1002/jaba.246 Clay, C. J., Samaha, A. L., Bloom, S. E., Bogoev, B. K., & Boyle, M. A. (2013). Assessing preference for social interactions. Research in developmental disabilities , 34 , 3 62 371. doi: 10.1016/j.ridd.2012.07.028 DeLeon, I. G., Bullock, C. E., & Catania, A. C. (2013). Arranging reinforcement contingencies in applied settings: Fundamentals and implications of recent basic and applied research. In G. J. Madden, W.V. Dube, G. Hanley, T. Hackenberg, and K.A. Lattal (Eds.) , American Psychological Association Handbook of Behavior Analysis . Washington, DC: American Psychological Association. DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a multiple stim ulus presentation format for assessing reinforcer preferences. Journal of Applied Behavior Analysis, 29 , 519 53 3 . doi: 10.1901/jaba.1996.29 519 Dozier, C. L., Iwata, B. A., Thomason Sassi, J., Worsdell , A. S., & Wilson, D. M. (2012). A comparison of two pairing procedures to establish praise as a reinforcer. Journal of Applied Behavior Analysis , 45 , 721 735. doi: 10.1901/jaba.2012.45 721 Egel, A. L. (1981). Reinforcer variation: Implications for motiva ting developmentally disabled children. Journal of Applied Behavior Analysis, 14 , 345 350. doi: 10.1901/jaba.1981.14 345 Fisher, W . , Piazza, C. C., Bowman, L. G., Hagopian, L. P., Owens, J. C., & Slevin , I. (1992). A comparison of two approaches for identifying reinforcers for persons with severe and profound disabilities. Journal of Applied Behavior Analysis, 25, 491 498. doi: 10.1901/jaba.1992.25 491 Fisher, W. W., Piazza, C. C., Bowman, L. G., & Amar i, A. (1996). Integrating caregiver report with a systematic choice assessment to enhance reinforcer identification . American Journal on Mental Retardation, 101, 15 25. Hanley, G. P., Iwata, B. A., Lindberg, J. S., & Conners, J. (2003). R esponse r e stricti on a nalysis: I. Assessment of activity preferences. Journal of Applied Behavior Analysis, 36, 47 58. doi: 10.1901/jaba.2003.36 47
35 Kelly, M. A., Roscoe, E. M., Hanley, G. P., & Schlichenmeyer, K. (2014). Evaluation of assessment methods for identifying soc ial reinforcers. Journal of Applied Behavior Analysis, 47, 113 135. doi: 10.1016/j.rasd.2014.06.012 Lang, R., va , G. (2014). Comparison of high and low preferred topographies o f contingent attention during discrete trial training. Research in Autism Spectrum Disorders, 8, 1279 1286. doi: 10.1016/j.rasd.2014.06.012 N uer nberger, J. E., Smith, C. A., Czapar, K. N., & Klatt, K. P. (2012). Assessing preference for social interaction in children diagnosed with autism. Behavioral Interventions, 27, 33 44. doi: 10.1002/bin.1336 Pace, G. M., Ivancic, M. T., Edwards, G. L., Iwata, B. A., & Page, T. J. (1985). Assessment of s timulus preference and reinforcer value with profoundly retarded individuals. Journal of Applied Behavior Analysis, 18, 249 255. doi: 10.1901/jaba.1985.18 249 Piazza, C. C., Bowman, L. G., Contrucci, S. A., Delia, M. D., Adelinis, J. D., & Goh, H . L. (199 9). An evaluation of the properties of attention as reinforcement for destructive and appropriate behavior. Journal of Applied Behavior Analysis , 32 , 437 449 . doi: 10.1901/jaba.1999.32 437 Skjoldager, P., Pierre, P. J., & Mittleman , G. (1993). Reinforcer magnitude and progressive ratio responding in the rat: effects of increased effort, prefeeding, and extinction. Learning and Motivation , 24 , 303 343. doi: 10.1006/lmot.1993.1019 Smaby, K., MacDonald, R. P. F. , Ahearn, W. H., &Dube, W. V. (2007). Assessment protocol for identifying preferred social consequences. Behavioral Interventions , 22 , 311 318. doi: 10.1002/bin.242
36 BIOGRAPHICAL SKETCH Samuel Morris the University of Florida i n 20 16 . He 9 from the University of Florida. He is currently pursuing a doctoral degree in p sychology, in the area of b ehavior a nalysis, under the advisement of Dr. Timothy Vollmer.