- Email: [email protected]
Assessing preference for social interactions
Research in Developmental Disabilities 34 (2013) 362–371
Contents lists available at SciVerse ScienceDirect
Research in Developmental Disabilities
Assessing preference for social interactions Casey J. Clay a, Andrew L. Samaha b,*, Sarah E. Bloom a, Bistra K. Bogoev b, Megan A. Boyle a a b
Department of Special Education and Rehabilitation, Utah State University, United States Department of Psychology, Utah State University, United States
A R T I C L E I N F O
A B S T R A C T
Article history: Received 9 March 2012 Received in revised form 26 July 2012 Accepted 26 July 2012 Available online 23 September 2012
We examined a procedure to assess preference for social interactions in individuals with intellectual and developmental disabilities. Preferences were identified in five individuals using a paired-choice procedure in which participants approached therapists who provided different forms of social interactions. A subsequent tracking test showed that participants’ approaches were under control of the form of social interaction provided as opposed to idiosyncratic features of the therapists. Results of a reinforcer assessment found that the social interaction identified as preferred also functioned as a reinforcer for all five participants. ß 2012 Elsevier Ltd. All rights reserved.
Keywords: Preference assessment Social interaction Reinforcer assessment
1. Introduction A number of techniques have been described that allow for the identification of preferred and reinforcing stimuli in individuals with language and communication deficits (see Hagopian, Long, & Rush, 2004 for a review). Several of these approaches involve a two-phase process (e.g., DeLeon & Iwata, 1996; Fisher et al., 1992; Pace, Ivancic, Edwards, Iwata, & Page, 1985; Roane, Vollmer, Ringdahl, & Marcus, 1998). In the first phase, an individual’s approaches toward or interactions with a set of stimuli are evaluated. In the second phase, stimuli approached most often are then provided contingent on performance of a response. Increases in that response relative to a baseline condition provide evidence that the stimulus functions as a reinforcer. Preference assessments are useful because they are relatively brief and predictive reinforcing efficacy, of performance under both increased work requirements (DeLeon, Frank, Gregory, & Allman, 2009) and extended work periods (Keyl-Austin, Samaha, Bloom, & Boyle, 2012). For example, DeLeon et al. (2009) identified high, moderate, and low-preferred stimuli before providing them contingent on behavior under progressive ratio schedules. Progressive ratio schedules involve systematically increasing responserequirements within a session and performance on progressive ratio schedules is often characterized in terms of breakpoints, or the highest ratio requirement participants complete successfully for a given stimulus. Intuitively, stimuli associated with greater breakpoints are more effective reinforcers. The authors reported that high-preference stimuli resulted in larger obtained break points on progressive ratio schedules, as compared to moderate and low-preferred stimuli. Likewise, Keyl-Austin et al. (2012) evaluated responding during extended 30-min sessions when both highly and moderately preferred stimuli were delivered response-contingently. Results suggested that although response rates were similar at the beginning of sessions, rates declined more rapidly for moderate-preference stimuli as compared to high-preference stimuli.
* Corresponding author at: Department of Psychology, Utah State University, 2810 Old Main Hill, Logan, UT 84322-2810, United States. Tel.: +1 435 797 1633. E-mail address: [email protected] (A.L. Samaha). 0891-4222/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ridd.2012.07.028
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
363
Findings from both of these studies suggest that stimulus preference can not only predict reinforcer efficacy in general, but also predict important aspects of performance. A seminal study by Fisher et al. (1992) evaluated a procedure for identifying preferred stimuli in a forced-choice format. Over successive trials, stimuli are presented to participants in pairs until every stimulus has been paired with every other stimulus. Approaches to one of the stimuli from each pair results in brief access and relative preference is inferred from the proportion of trials in which each stimulus is selected. As Fisher et al.’s data suggested, individuals might show indistinguishable patterns of approach to two stimuli presented in a single-item format even when they consistently select one over the other when the stimuli are presented in a forced- or paired-choice format. The latter approach was a technological improvement over previous single-item formats (Pace et al., 1985) because of its sensitivity to differences in degree of preference. Such relative sensitivity is often described by ranking the stimuli according to degree of preference to produce a preference hierarchy (Fisher et al., 1992; Piazza, Fisher, Hagopian, Bowman, & Toole, 1996). In addition to the clinical utility of being able to identify the most preferred stimulus in a given set, the assessment of preference through choice procedures allows for the subsequent investigation of a number of preference-related effects. Other techniques have been described such as the multiple-stimulus without replacement (MSWO) and free-operant (FO; Roane et al., 1998) preference assessments, which take less time to conduct and are associated with less problem behavior. However, neither the MSWO nor the FO produce more consistent results across subsequent tests than the paired-choice preference assessment (DeLeon & Iwata, 1996; Windsor, Piche, & Locke, 1994). DeLeon and Iwata evaluated the consistency of results obtained using the MSWO, paired-choice preference assessment, and a multiple stimulus format similar to the FO in that a set of stimuli was presented and stimuli continued to remain available after selection by participants. DeLeon and Iwata found that participants’ resulting preference hierarchies were more consistent across five repeated administrations of the pairedchoice procedure as compared to the others assessments. Similarly, Windsor et al. compared the consistency obtained across six administrations of a paired-choice format and multiple-stimulus format and found consistency was greater across successive administration of the paired-choice format, replicating the results of DeLeon and Iwata. The test–retest reliability of paired-choice format may be due to a number of factors. Unlike multiple-stimulus formats that may require individuals to select among a large set of stimuli, paired-choice preference assessments only require participants make selections between two choices on any given trial. As such, reliable results of multiple stimulus methods may be dependent on certain prerequisite skills on the part of the person being assessed, e.g., the ability and self-control to scan all of the available stimuli before making a selection, the sensory capabilities to distinguish between the stimuli, and perhaps the ability to ignore characteristics that draw one’s attention but are not correlated with any responsestrengthening aspects of the stimuli. Thus, paired-choice formats may be more appropriate for a broader set of individuals. Indeed, the paired-choice preference assessment has been shown to be capable of detecting preference for stimuli in individuals exhibiting a broad range of intellectual abilities and skills including children (Fisher et al., 1992; Piazza et al., 1996), adults with intellectual and developmental disabilities (DeLeon & Iwata, 1996), children with visual impairments (Paclawskyj & Vollmer, 1995), and typically developing toddlers (Cote, Thompson, Hanley, & McKerchar, 2007). Thus, the reliability and broad applicability of the paired-choice preference assessment make it appealing for extension into understudied areas. One particularly understudied area is the examination of methods to identify preferred forms of social interaction. Methods to identify preferred forms of social interaction may be useful if food or edibles are contraindicated (e.g., as in individuals with Prader–Willi Syndrome), if individuals lack the prerequisite skills to play appropriately with toys or activities, in cases when carrying and delivering potential reinforcers are impractical, or if the target behavior warrants a form of attention specifically (e.g., a request for attention). Further, there is accumulating evidence in typically developing individuals and nonhuman animals that repeated exposure to energy-rich food may be associated with an increased risk of obesity (see Anzman, Rollins, & Birch, 2010; Kenny, 2011 for reviews). In addition to the immediate health effects of eating high-caloric foods, their delivery might be accompanied by long-term changes in food preferences that in turn lead to health challenges later in life. Such evidence might suggest that reinforcement-based interventions relying heavily on food could affect long-term food preferences and have unintended downstream health effects. Health effects not withstanding, brainimaging studies in obese and overweight humans have identified changes in the dorsal striatum indicative of reward hyposensitivity, the net result of which may be increased seeking of palatable foods (e.g., Stice, Spoor, Bohon, & Small, 2008). Although studies on the relative preference for food versus social interactions have not been conducted, increases in food seeking might exacerbate existing deficits in communication or social skills (to the extent that behavior maintained by food displace behavior maintained by social interactions) such as those present in individuals with autism. Therefore, overuse of food in reinforcement-based interventions may be associated with both long-term health effects and food preferences, the latter of which may be contraindicated for particular populations. Both possibilities invite a renewed need to identify effective reinforcers in other classes of stimuli, including social interactions. Unfortunately, a number of difficulties may be encountered when attempting to evaluate individuals’ preferences for social interactions. Whereas the appearance of food items is strongly related to their taste, social interactions do not necessarily have antecedents, such as physical appearance, that reliably signal their availability. Likewise, two tangible items can easily be presented at the same time by the same person. However, it may be difficult for the same therapist to signal the availability of two or more forms of social interaction. One approach to ameliorating these problems may be to present choices between picture-cards or icons that result in brief access to some other extended activity (Graff & Gibson, 2003; Hanley, Iwata, & Lindberg, 1999). However, this approach
364
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
may require extended experience or sophisticated discriminative repertoires not present in all individuals (see for example Clevenger & Graff, 2005). Other data suggest that use of pictures as opposed to tangible items may result in greater variation across selections (Higbee, Carr, & Harrison, 1999). If certain individuals are consistent in the forms of social interaction they provide to others, an alternative approach may be to assess individuals’ preferences for different staff members (Rapp, 2005; Sturmey, Lee, Reyer, & Robek, 2003). Rapp (2005) examined an individual’s preference for staff and caregivers by allowing the participant to approach one of the two adults present in a room. Approach responses resulted in attention from the adult. Although the author did not attempt to control or equate the attention provided by the adults, the participant showed clear preferences for particular staff and caregivers. Results from this study suggest that assessing preferences for forms of social interaction might be possible in the context of a paired-choice assessment. Smaby, MacDonald, Ahearn, and Dube (2007) evaluated an alternative method for examining the reinforcing effectiveness of different forms of social interaction in children with autism. The experimenters used a single-operant arrangement to assess the efficacy of three forms of social interaction (i.e., tickle, head rub, and praise). One of the three social interactions was delivered each time participants handed over a corresponding token to the therapist. Responding (handing over a token) in a social interaction condition was compared to responding in an extinction condition in which social interaction was not provided. Response rates were differentiated across the forms of social interaction, revealing a hierarchy across the social stimuli. However, previous research evaluating single-operant schedules has found that absolute rates of responding may not be well correlated with preference (Roscoe, Iwata, & Kahng, 1999). In addition, Smaby et al.’s inconsistent session termination criteria might have artificially produced relative increases in rate during social consequence sessions. This would have been due to experimenters continuing to carry out extinction sessions for 5 min or until 1 min elapsed without responding (the latter of which might tend to produce relative decreases during extinction) but terminating social consequences sessions after 1 min regardless of the level of responding. Finally, Smaby et al.’s procedures included two potential points of comparison that make comparing rates across social consequence sessions difficult: response rate during the extinction periods prior to each social consequence session and response rate during other social consequence sessions. Results may be ambiguous if, for example, levels of responding in social consequence sessions alone suggest one form of social interaction is most-preferred (e.g., high-five) but the level of responding during the preceding extinction session for that social consequence was highest (indicating a potential false-positive reinforcement-effect). Said another way, the procedure does not easily distinguish between forms of social interaction when baseline levels of responding vary across the forms. The use of social interactions as reinforcers has a number of potential advantages over tangible and edible stimuli in skill acquisition and behavior management programs. First, once identified, social interactions are unlikely to be as expensive as tangible or edible stimuli in terms of costs associated with purchase or storage. Second, social interactions may be more practical to deliver in community settings where it may be inconvenient for therapists to carry necessary edibles or tangibles. Third, the delivery of social interactions may be less stigmatizing in educational and community settings than edibles and tangibles. Fourth, the use of social interactions as reinforcers may promote generalization because their availability as reinforcers is not dependent on the presence of objects that may exert discriminative control over responding. Fifth, the inclusion of reinforcing forms of social interactions may strengthen social skill training programs, such as those used in early intensive behavioral intervention programs for children with autism. Indeed, the use of social interactions as reinforcers for individuals with autism seems particularly well suited given the nature of their social deficits (Volkmar, Lord, Bailey, Schultz, & Klin, 2004). Sixth, the identification of reinforcing forms of social interactions increases the array of available stimuli for individuals on caloric or food restricted diets. Finally, the inclusion of effective reinforcers is an essential component of any reinforcement-based intervention. Skill acquisition programs and reinforcement-based behavior management programs can fail for a variety of reasons, only one of which is the misidentification of an effective reinforcer. Therefore, identification and inclusion of reinforcing stimuli in reinforcement-based interventions may have two benefits: (1) they may increase the likelihood that interventions are effective and (2) they may reduce the time and effort related to repairing interventions that are initially ineffective. Taken together, there would seem to be a clear need for a renewed investigation of methods for the identification of preferred forms of social interaction. The purpose of this study was to examine a method of identifying individuals’ preferences for forms of social interaction in a paired-choice format and evaluate whether those forms identified as highly preferred serve as reinforcers. In an initial phase, we presented participants with pairs of therapists who, when approached, delivered a specific form of social interaction. In a second phase, we alternated therapists assigned to deliver the recently established highest and lowest preferred form of social interaction to determine whether participant approaches were under the control of the form of social interaction or idiosyncratic features of the therapist. Finally, we evaluated whether the form of social interaction identified as most preferred functioned as a reinforcer. 2. Methods 2.1. Participants and setting Five children referred for assessment and treatment of severe problem behavior participated. Alex was a 12-year-old boy diagnosed with autism and cerebral palsy. He was proficient at making choices, but had ambulatory difficulties due to
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
365
cerebral palsy. He was referred for non-compliant and aggressive problem behavior. Sofia was a 10-year-old girl diagnosed with Attention Deficit Hyperactivity Disorder (ADHD), Pervasive Developmental Disability Not Otherwise Specified (PDDNOS), and Bipolar Disorder. She attended an inclusive elementary school classroom. She was referred for verbal threats and aggressive, self-injurious, and destructive behavior. Chris was a 6-year-old boy diagnosed with ADHD. He attended an inclusive elementary school classroom and did school work at the level of his peers. He was referred for aggressive behavior. Kyle was a 17-year-old boy diagnosed with Cognitive Disorder, Mood Disorder, Mild Intellectual Disability, and Lennox– Gastaut Syndrome. He attended a self-contained high school special education classroom. He was referred for treatment of aggressive and destructive behavior. Rutherford was a 6-year-old boy diagnosed with ADHD. He was referred for aggressive and non-compliant behavior. He attended a self-contained elementary school special education classroom. All participants could follow full sentence instructions, and exhibited vocal verbal behavior. Alex, Chris, and Sofia’s assessments were conducted in 2 m 3 m room in a university-based outpatient client. The room included a small table, two chairs, and a oneway mirror. Kyle’s assessments were conducted in a 3 m 3 m area of his school classroom that contained a table and two chairs. No other students were present at the time of the assessment. Rutherford’s assessments were conducted in a 2 m 2.5 m area in his school auditorium containing one chair. 2.2. Preference assessment We identified four forms of attention to be included in the assessment through an open-ended informal interview of participants’ teachers or caregivers about interactions that they seemed to enjoy. Each form of social interaction consisted of both physical contact (e.g., a high-five gesture) and a vocal statement (e.g., ‘‘High-Five!’’) presented in an exaggerated and playful tone (see Table 1 for operational definitions; durations are specified for cases in which the physical interaction did not end immediately after it began [cf. Hugs vs. High-Five]). For Alex, we evaluated Noodle Arms, High-Five, Head Rubs, and Hugs. For Kyle, we evaluated Thumbs Up, High-Five, Knuckles, and Rascal. For Chris, we evaluated Tickles, High-Five, Noodle Arms, and Squeezes. For Rutherford, we evaluated Thumbs Up, Head Rubs, Noodle Arms, and High-Five. For Sofia, we evaluated Hugs, High-Five, Knuckles, and Noodle Arms. During this phase of the assessment each of four therapists was assigned one of the four forms of social interaction identified for that participant. The therapist only delivered the form of interaction they were assigned. A total of 9 therapists were used to deliver forms of social interaction for the study. Therapists included students at the undergraduate, masters, and doctoral levels. 2.2.1. Trial blocks Assessments were conducted over a series of six blocks of 5 trials. Blocks were presented until each combination of therapist and form of social interaction had been paired with every other. This resulted in a total of 6 blocks or 30 trials. Each block consisted of an initial pre-exposure to two forms of social interaction delivered individually by one therapist each followed by five paired-choice trials in which participants could choose between those therapists and forms of social interaction. The following procedures were used with Sofia, Chris, Kyle, and Rutherford. Slightly different procedures (described below) were used with Alex to accommodate a physical disability due to cerebral palsy. 2.2.2. Pre-exposures In order for participants to contact the contingency for approaching a particular therapist to receive a particular form of social interaction, a pre-exposure was provided at the beginning of each trial block. During the pre-exposure, a combination of vocal instructions and manual guidance was used to prompt participants to the center of the room, and turn to face a wall where one therapist stood. Participants were then prompted to approach that therapist who delivered the assigned form of social interaction. Approach was defined as movement to within .3 m of either therapist for all participants except Alex. Participants were prompted to approach each therapist relevant for that block of trials and receive the form of social interaction relevant for that therapist. Pre-exposures continued until participants experienced each form of social interaction once.
Table 1 Operational definitions of social interactions. Label
Description
Therapist statement
Noodle Arms High Five Head Rubs Hugs Thumbs Up Knuckles Rascal Tickles Squeezes
Holding the participants’ hands, and wiggling their arms up and down for 2 s Holding up one hand with an open palm and making contact with participant’s hand once Rubbing the participant’s head lightly for 2 s Wrapping both arms around the participant for 2 s Raising both hands in a thumbs-up position for 2 s Extending a closed fist toward the participant’s hand and making light contact once Tapping the participant lightly on the shoulder once Tickling the participant under the arms for 2 s Wrapping both arms around the participant, and rocking side to side gently for 2 s
‘‘Noodle arms.’’ ‘‘High five.’’ ‘‘Head rubs.’’ ‘‘Hugs.’’ ‘‘Thumbs up.’’ ‘‘Knuckles.’’ ‘‘You little rascal.’’ ‘‘Tickles!’’ ‘‘Squeezes.’’
366
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
2.2.3. Trials At the beginning of each trial, the participant was re-positioned to the center of the room, both therapists associated with the current block of trials stood in opposite corners along the same wall, and the participant was instructed to ‘‘pick one.’’ Manual guidance was not used following the vocal cue. Therapists ignored participants (i.e., did not make eye contact) until an approach occurred. Approaches resulted in brief delivery of the social interaction associated with that therapist and initiation of a new trial. Therapists alternated left and right positions prior to each new trial. Observers recorded participants’ selections in each trial and the position (left or right) of the therapist. Observers scored ‘‘no selection’’ and a new trial was presented if participants did not make an approach response within 10 s. Participants were re-prompted to the center of the room within 4 s after the social interaction terminated. For Alex, due to the locomotor challenges associated with his cerebral palsy, pre-exposures and paired-choice trials took place with him seated at a table and two therapists seated across from him. Approaches were defined as touching one of the therapists with any part of his hand. As with other participants, approaches resulted in brief interaction and therapists switched positions after every trial. 2.3. Tracking test Following the preference assessment, we re-evaluated participants’ preferences for the high- and low-preferred forms of social interaction by switching the therapists associated with these forms of social interaction. This was done in order to determine if participants’ choices in the previous phase were under the control of the form of attention provided or idiosyncratic features of the specific therapist (e.g., appearance, gender). Specifically, the therapist previously associated with the most-preferred form of social interaction now delivered the least-preferred form and vice versa. Each block of trials began with pre-exposures as described above. As in the previous phase, blocks were preceded by a pre-exposure to each social-interaction/therapist and included five paired-choice trials. Therapists alternated left and right positions every trial. Two to four blocks were conducted in this phase due to the reduction in attention forms assessed from 4 (during the preference assessment) to 2. 2.3.1. Response measurement and reliability Trained observers recorded participants’ choices and therapist position on each trial. An agreement was defined as both observers having recorded the same selection for each trial. Reliability was calculated by dividing agreements by agreements plus disagreements and multiplying by 100 to produce a percentage. A second observer collected data independently on 59% of the trials. Observers agreed on 100% of trials. 2.4. Reinforcer assessment One form of social interaction was evaluated in a reinforcer assessment for each participant. To be included in the reinforcer assessment, a given form of social interaction needed to be selected most often during the preference assessment and continue to be preferred during the tracking test. One of the two target responses (defined below) was selected for each participant. Extinction baseline and fixed-ratio 1 (FR1) were used in ABAB design to evaluate the reinforcing efficacy of the highest preferred social interaction for each participant. 2.4.1. Session protocol All sessions were 5 min and preceded by a single exposure to the programmed contingency: The therapist said, ‘‘When you do this’’, manually guided the participant to complete the response, then said, ‘‘you get this,’’ and delivered either the most preferred form of attention during FR1, or nothing during baseline. All target responses resulted in delivery of the programmed consequence for that condition. 2.4.2. Response definitions 2.4.2.1. Block-in-bowl. Participants were seated at a table and presented with two identical bowls, one of which contained a small block. A response was recorded when the participant picked up the block, placed it in the other bowl, and released it. This response was used for Kyle, Alex, and Sofia. 2.4.2.2. Spot touching. Participants were seated at a table and presented with a laminated colored sheet of construction paper placed approximately one foot in front of them. Pre-session exposure and session conditions were identical to block-inbowl. A response was recorded any time a part of the participant’s hand touched the construction paper. This response was used for Chris and Rutherford. 2.5. Response measurement and reliability Trained observers recorded the occurrence of each participant response using !Observe software on a hand-held computer. The software automatically time-tags and stores events as they are scored, and is capable of reporting those data
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
367
as either percentage of intervals or frequency. A second observer collected data independently during 38% of sessions. We compared observer’s records using both average agreement within 10-s intervals and Pearson’s r. Average agreement within intervals was calculated by dividing the smaller number by the larger number scored by observers in each interval, averaging across all intervals, and multiplying by 100 to obtain a percentage. Intervals in which neither observer recorded a response were scored as an agreement (1.0). Intervals in which only one observer recorded no occurrences of behavior were scored as a disagreement (0.0). Average agreement within intervals was 94.2% (range, 77.8–100%). Pearson’s r was also calculated to compare session rates obtained by both observers and was r(27) = .998, p < .0001. 3. Results Results of the attention preference assessment are shown in Fig. 1. Noodle Arms was the highest preferred form of attention (80% of selections) for Alex. Thumbs Up was the highest preferred form of attention (100% of selections) for Kyle. Tickles was the highest preferred form of attention (60% of selections) for Chris. Thumbs Up was the highest preferred form of attention for Rutherford. Hugs was the highest preferred form of attention (73% of selections) for Sofia. Results of the preference assessment and tracking test are shown together in Fig. 2, however baseline data include only those trials from the initial preference assessment in which the most- and least-preferred forms of social interaction were presented together. Baseline was composed of 2 blocks of sessions (10 trials). Closed and open symbols indicate the forms of social interaction that were most- and least-preferred during the initial preference assessment. Results show that the relative preference of the most- and least-preferred forms of social interaction remained consistent across therapists, indicating that participants’ selections were under the control of the form of attention provided as opposed to idiosyncratic features of the therapists. Data for Chris and Sofia indicate that they initially approached the therapist previously associated with the most-preferred form of attention in the first block of trials following the therapist switch. However, choices in subsequent trial blocks indicate they began to track the form of social interaction.
Fig. 1. Results from phase 1 of the preference assessment. Data reflect the percentage of selections to each form of social interaction.
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
368
BL
BL
TRACKING TEST 100
100
Noodle arms
80
TRACKING TEST Thumbs Up
80
60
60
40
40
Rutherford 20
20
Hugs Alex
0
PERCENTAGE OF SELECTIONS
1
2
High Five
0
3
1
100
100
80
80
2
3
4
5
Hugs Thumbs Up
60
60
40
40
Rascal
20
Kyle
0
20
Noodle Arms
0 1
2
3
1
2
3
Sofia 4
5
TRIAL BLOCKS 100
Tickles 80 60 40
Squeezes
20
Chris
0 1
2
3
TRIAL BLOCKS Fig. 2. Results from phases 1 and 2 of the attention preference assessment. Data from the BL condition reflect the percentage of selections toward the mostand least-preferred form of social interaction from phase 1. Data from the therapist switch condition reflect selections made after the therapists associated with the most- and least-preferred forms of social interaction in phase 1 switched roles.
Results from the reinforcer assessment are shown in Fig. 3. The most-preferred form of attention functioned as a reinforcer for all participants. For Alex, we saw little or no responding during the initial baseline phase. Upon switching to the FR1 phase we saw a modest increase in responding. Reversing back to baseline, we again saw little or no responding. In the second FR1 phase we saw a relatively high level of responding (M = 5.5) indicating that his highest preferred form of attention, Noodle Arms, served as a reinforcer. For Kyle, we saw a variable decreasing trend in the initial baseline phase, followed by an increasing trend in responding during the first FR1 phase. Reversing back to baseline we saw near zero response rates. In the second FR1 phase we saw a relatively high level of responding (M = 37.6) indicating that his highest preferred form of attention, Thumbs Up, served as a reinforcer. For Chris, we did not see any responding in the initial baseline phase. In the first FR1 phase we saw high levels of responding, followed by no responding in the second baseline phase. In the second FR1 phase we saw an increasing trend in responding at a level comparable to the first FR1 phase indicating that his highest preferred form of attention, Tickles, served as a reinforcer. For Rutherford, we saw low to zero response rates in the initial baseline phase, followed by elevated levels of responding in the FR1 phase (M = 10.7). In the second baseline phase we saw a return of low to zero rates of responding. In the final FR1 phase we saw elevated rates of responding (M = 7.6) indicating that his highest preferred form of attention, Thumbs Up, served as a reinforcer. For Sofia, we saw low to zero rates of responding in the initial baseline phase, followed by elevated levels of responding in the FR1 phase (M = 5.73). In the second baseline phase we saw a return to low to zero rates of responding. In the final FR1 phase we saw an increasing trend in responding, indicating that her highest preferred form of attention, Hugs, served as a reinforcer. Overall, for each of the five participants, the highest preferred form of attention identified through the preference assessment functioned as a reinforcer.
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371 10
FR1
BL
BL
18
FR1
BL
FR1
BL
369 FR1
8
12 6 4
6 2 Alex
0 5
RESPONSESPERMINUTE
50
10
FR1
BL
15
BL
Rutherford
0
20
5
10
FR1
40
8
30
6
20
4
10
BL
10
FR1
BL
15
20
FR1
2 Kyle
0 5
10
15
20
Sofia
0 5
10
15
20
SESSIONS 5
BL
FR1
BL
FR1
4 3 2 1 Chris
0 5
10
15
20
SESSIONS Fig. 3. Response rates in the reinforcer assessment for block-in-bowl (Alex, Kyle, and Sofia) and spot touching (Rutherford).
4. Discussion This study evaluated a three-phase process for identifying preferred and reinforcing forms of social interaction. In the first phase, a paired-choice preference assessment was used to identify a hierarchy of preferred social interactions. A subsequent tracking test provided evidence that participants’ approaches were under the control of the form of social interaction delivered, as opposed to idiosyncratic features of the therapists. Finally, a third phase demonstrated that forms of social interaction shown to be both preferred and tracked functioned as reinforcers for all five participants. Like previous studies, a paired-choice format successfully led to the identification of a preference hierarchy (Fisher et al., 1992; Piazza et al., 1996). Also like previous studies, events identified as highly preferred via preference assessment functioned as reinforcers. These results extend previous research by evaluating different forms of social interactions in a paired-choice format and evaluating the generality of those preferences (i.e., the range of conditions in which maintain) across therapists. A number of the seminal studies on preference assessments included at least one form of social interaction (e.g., Fisher et al., 1992; Pace et al., 1985). However, social interactions were not highly ranked by participants in those studies. In addition, research on preference assessment outcomes has found that stimuli in particular classes (e.g., edibles, activities) tend to be similarly preferred (DeLeon, Iwata, & Roscoe, 1997). That is, food tends to be more highly preferred than activities. Additional research should be conducted on whether similar patterns are obtained with social interactions and, if so, to determine the relative preference for social interactions versus other stimuli. One limitation of the current study is that we did not compare the reinforcing efficacy of high- and low-preferred social interactions. Given the results of previous work (e.g., DeLeon et al., 2009), we would expect highly ranked social interactions to be more effective reinforcers than comparatively lower ranked social interactions. Future research should examine the generality of such patterns with social interactions.
370
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
Similarly, we did not evaluate the necessity of the tracking test. In preference assessments for edible or tangible stimuli, no analogous test exists. Indeed, with edible and tangible stimuli, the formal qualities of their consumption tend to be highly and uniquely correlated with their appearance (e.g., the best avocados are dark and slightly tender whereas the best bananas are bright yellow with few marks or dark spots). It is easy to imagine how the appearance and smell of certain foods can become discriminative for the properties that make them delicious. Evidence exists that certain human preferences may be driven by features (e.g., facial symmetry) that predict reproductive fitness (Prokosch, Yeo, & Miller, 2005). Any such preferences would likely interfere with the social interactions we were trying to assess. In addition, phenomenon such as stimulus overselectivity, or the tendency for behavior to come under the control of a narrow subset of potentially relevant stimuli, has been associated with certain intellectual and developmental disabilities such as autism (Schreibman & Lovaas, 1973) and general cognitive deficits like low mental age (Schover & Newsom, 1976). Therefore, it seemed reasonable to be skeptical of the variables controlling participants’ choices during the preference assessments especially given the brief nature of the assessment. For the purposes of this study, the tracking test provided evidence that social interactions evaluated during the reinforcer assessment were preferred. However, we did not evaluate the necessity of the tracking test. Future studies may wish to examine the degree to which phenomena like overselectivity affect the interpretation of preference assessments. Although we did not evaluate it, it is also possible that the value of a given social interaction was related to the history of reinforcement associated with the therapist delivering it (e.g., one might prefer a high-five from person A more than person B). Results of the tracking test suggest such preferences in the current study were not strong enough to override differences between the high and low-preferred forms of social interaction. Future researchers may wish to examine this interaction. It is worth nothing that the assessment of preference for forms of social interaction was fairly personnel intensive. Four therapists were involved in the preference assessment: one for each form of social interaction. Although this may be a practical limitation of our procedure in clinical settings, assessing the generality of and individual’s preferences for forms of social interaction across different therapists (as in the tracking test) necessarily requires more than one therapist. Having each therapist deliver only one kind of social interaction was designed to increase the likelihood that approaches to specific therapists were more likely under the control the form of social interaction (i.e., we felt it would lead to an increases in the reliability of the assessment). Previous research with paired-choice procedures suggests their findings are highly reliable across time (DeLeon & Iwata, 1996). Future research may examine the reliability of the procedures described here across time or compare the reliability of these procedures to those involving choices between social interactions represented by other tangible stimuli. Future research might also examine whether similar results can be obtained when forms of social interactions are delivered by the same therapist and represented using picture cards. Another promising approach may be to represent social interactions using brief video clips (cf. Snyder, Higbee, & Dayton, 2012). A potential application of the procedures described here is in the treatment of severe problem behavior maintained by attention. In some cases, it may be impractical or unethical to place problem behavior on extinction (e.g., aggression toward peers or severe self-injury). An alternative approach may be to provide attention following both appropriate and problem behavior, but vary the quality of attention provided following each. For example, highly preferred forms of attention may be provided following appropriate requests and low-preferred forms following problem behavior. In this way, the contingencies may still favor appropriate requests while not placing problem behavior on extinction (cf. Athens & Vollmer, 2010). The procedures described in the current study contribute to the literature on preference assessment by applying a variation of the paired-choice preference assessment to forms of social interaction. We also demonstrated a procedure that can distinguish between selections that are under control of a form of social interaction versus features of the individual delivering said social interaction. Finally, we showed that this approach to preference assessment identified social stimuli that were effective reinforcers. These procedures may be especially useful in cases in which it is necessary to identify preferred forms of social interactions for use in acquisition programs as well as interventions for problem behavior. References Anzman, S. L., Rollins, B. Y., & Birch, L. L. (2010). Parental influence on children’s early eating environments and obesity risk: Implications for prevention. International Journal of Obesity, 34, 1116–1124. Athens, E. S., & Vollmer, T. R. (2010). An investigation of differential reinforcement of alternative behavior without extinction. Journal of Applied Behavior Analysis, 43, 569–589. Clevenger, T. M., & Graff, R. B. (2005). Assessing object-to-picture and picture-to-object matching as prerequisite skills for pictorial preference assessments. Journal of Applied Behavior Analysis, 38, 543–547. Cote, C. A., Thompson, R. H., Hanley, G. P., & McKerchar, P. M. (2007). Teacher report and direct assessment of preferences for identifying reinforcers for young children. Journal of Applied Behavior Analysis, 40, 157–166. DeLeon, I. G., Frank, M. A., Gregory, M. K., & Allman, M. J. (2009). On the correspondence between preference assessment outcomes and progressive-ratio schedule assessments of stimulus value. Journal of Applied Behavior Analysis, 42, 729–733. DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a multiple-stimulus presentation format for assessing reinforcer preferences. Journal of Applied Behavior Analysis, 29, 519–532. DeLeon, I. G., Iwata, B. A., & Roscoe, E. M. (1997). Displacement of leisure reinforcers by food during preference assessments. Journal of Applied Behavior Analysis, 30, 457–484. 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. Graff, R. B., & Gibson, L. (2003). Using pictures to assess reinforcers in individuals with developmental disabilities. Behavior Modification, 27, 470–483.
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
371
Hagopian, L. P., Long, E. S., & Rush, K. S. (2004). Preference assessment procedures for individuals with developmental disabilities. Behavior Modification, 28, 668–677. Hanley, G. P., Iwata, B. A., & Lindberg, J. S. (1999). Analysis of activity preferences as a function of differential consequences. Journal of Applied Behavior Analysis, 32, 419–435. Higbee, T. S., Carr, J. E., & Harrison, C. D. (1999). The effects of pictorial versus tangible stimuli in stimulus-preference assessments. Research in Developmental Disabilities, 20, 63–72. Kenny, P. J. (2011). Reward mechanisms in obesity: New insights and future directions. Neuron, 69, 664–679. Keyl-Austin, A. A., Samaha, A. L., Bloom, S. E., & Boyle, M. A. (2012). Effects of preference and reinforcer variation on within-session patterns of responding. Journal of Applied Behavior Analysis, 45, 637–641. Pace, G. M., Ivancic, M. T., Edwards, G. L., Iwata, B. A., & Page, T. J. (1985). Assessment of stimulus preference and reinforcer value with profoundly retarded individuals. Journal of Applied Behavior Analysis, 18, 249–255. Paclawskyj, T. R., & Vollmer, T. R. (1995). Reinforcer assessment for children with developmental disabilities and visual impairments. Journal of Applied Behavior Analysis, 28, 219–224. Piazza, C. C., Fisher, W. W., Hagopian, L. P., Bowman, L. G., & Toole, L. (1996). Using a choice assessment to predict reinforcer effectiveness. Journal of Applied Behavior Analysis, 29, 1–9. Prokosch, M. D., Yeo, R. A., & Miller, G. F. (2005). Intelligence tests with higher g-loadings show higher correlations with body symmetry: Evidence for a general fitness factor mediated by developmental stability. Intelligence, 33, 203–213. Rapp, J. T. (2005). An assessment of preference for caregivers and antecedents for problem behavior. Behavioral Interventions, 20, 301–311. Roane, H. S., Vollmer, T. R., Ringdahl, J. E., & Marcus, B. A. (1998). Evaluation of a brief stimulus preference assessment. Journal of Applied Behavior Analysis, 31, 605–620. Roscoe, E. M., Iwata, B. A., & Kahng, S. W. (1999). Relative versus absolute reinforcement effects: Implications for preference assessments. Journal of Applied Behavior Analysis, 32, 479–493. Schover, L. R., & Newsom, C. D. (1976). Overselectivity, developmental level, and overtraining in autistic and normal children. Journal of Abnormal Child Psychology, 4, 289–298. Schreibman, L., & Lovaas, O. I. (1973). Overselective response to social stimuli by autistic children. Journal of Abnormal Child Psychology, 1, 152–168. 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. Snyder, K., Higbee, T. S., & Dayton, E. (2012). Preliminary investigation of a video-based stimulus preference assessment. Journal of Applied Behavior Analysis, 45, 413–418. Stice, E., Spoor, S., Bohon, C., & Small, D. M. (2008). Relation between obesity and blunted striatal response to food is moderated by TaqlA A1 allele. Science, 322, 449–452. Sturmey, P., Lee, R., Reyer, H., & Robek, A. (2003). Assessing preferences for staff: Some pilot data. Behavioural and Cognitive Psychotherapy, 31, 103–107. Volkmar, F. R., Lord, C., Bailey, A., Schultz, R. T., & Klin, A. (2004). Autism and pervasive developmental disorders. Journal of Child Psychology and Psychiatry, 45, 135–170. Windsor, J., Piche, L. M., & Locke, P. A. (1994). Preference testing: A comparison of two presentation methods. Research in Developmental Disabilities, 15, 439–455.
Contents lists available at SciVerse ScienceDirect
Research in Developmental Disabilities
Assessing preference for social interactions Casey J. Clay a, Andrew L. Samaha b,*, Sarah E. Bloom a, Bistra K. Bogoev b, Megan A. Boyle a a b
Department of Special Education and Rehabilitation, Utah State University, United States Department of Psychology, Utah State University, United States
A R T I C L E I N F O
A B S T R A C T
Article history: Received 9 March 2012 Received in revised form 26 July 2012 Accepted 26 July 2012 Available online 23 September 2012
We examined a procedure to assess preference for social interactions in individuals with intellectual and developmental disabilities. Preferences were identified in five individuals using a paired-choice procedure in which participants approached therapists who provided different forms of social interactions. A subsequent tracking test showed that participants’ approaches were under control of the form of social interaction provided as opposed to idiosyncratic features of the therapists. Results of a reinforcer assessment found that the social interaction identified as preferred also functioned as a reinforcer for all five participants. ß 2012 Elsevier Ltd. All rights reserved.
Keywords: Preference assessment Social interaction Reinforcer assessment
1. Introduction A number of techniques have been described that allow for the identification of preferred and reinforcing stimuli in individuals with language and communication deficits (see Hagopian, Long, & Rush, 2004 for a review). Several of these approaches involve a two-phase process (e.g., DeLeon & Iwata, 1996; Fisher et al., 1992; Pace, Ivancic, Edwards, Iwata, & Page, 1985; Roane, Vollmer, Ringdahl, & Marcus, 1998). In the first phase, an individual’s approaches toward or interactions with a set of stimuli are evaluated. In the second phase, stimuli approached most often are then provided contingent on performance of a response. Increases in that response relative to a baseline condition provide evidence that the stimulus functions as a reinforcer. Preference assessments are useful because they are relatively brief and predictive reinforcing efficacy, of performance under both increased work requirements (DeLeon, Frank, Gregory, & Allman, 2009) and extended work periods (Keyl-Austin, Samaha, Bloom, & Boyle, 2012). For example, DeLeon et al. (2009) identified high, moderate, and low-preferred stimuli before providing them contingent on behavior under progressive ratio schedules. Progressive ratio schedules involve systematically increasing responserequirements within a session and performance on progressive ratio schedules is often characterized in terms of breakpoints, or the highest ratio requirement participants complete successfully for a given stimulus. Intuitively, stimuli associated with greater breakpoints are more effective reinforcers. The authors reported that high-preference stimuli resulted in larger obtained break points on progressive ratio schedules, as compared to moderate and low-preferred stimuli. Likewise, Keyl-Austin et al. (2012) evaluated responding during extended 30-min sessions when both highly and moderately preferred stimuli were delivered response-contingently. Results suggested that although response rates were similar at the beginning of sessions, rates declined more rapidly for moderate-preference stimuli as compared to high-preference stimuli.
* Corresponding author at: Department of Psychology, Utah State University, 2810 Old Main Hill, Logan, UT 84322-2810, United States. Tel.: +1 435 797 1633. E-mail address: [email protected] (A.L. Samaha). 0891-4222/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ridd.2012.07.028
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
363
Findings from both of these studies suggest that stimulus preference can not only predict reinforcer efficacy in general, but also predict important aspects of performance. A seminal study by Fisher et al. (1992) evaluated a procedure for identifying preferred stimuli in a forced-choice format. Over successive trials, stimuli are presented to participants in pairs until every stimulus has been paired with every other stimulus. Approaches to one of the stimuli from each pair results in brief access and relative preference is inferred from the proportion of trials in which each stimulus is selected. As Fisher et al.’s data suggested, individuals might show indistinguishable patterns of approach to two stimuli presented in a single-item format even when they consistently select one over the other when the stimuli are presented in a forced- or paired-choice format. The latter approach was a technological improvement over previous single-item formats (Pace et al., 1985) because of its sensitivity to differences in degree of preference. Such relative sensitivity is often described by ranking the stimuli according to degree of preference to produce a preference hierarchy (Fisher et al., 1992; Piazza, Fisher, Hagopian, Bowman, & Toole, 1996). In addition to the clinical utility of being able to identify the most preferred stimulus in a given set, the assessment of preference through choice procedures allows for the subsequent investigation of a number of preference-related effects. Other techniques have been described such as the multiple-stimulus without replacement (MSWO) and free-operant (FO; Roane et al., 1998) preference assessments, which take less time to conduct and are associated with less problem behavior. However, neither the MSWO nor the FO produce more consistent results across subsequent tests than the paired-choice preference assessment (DeLeon & Iwata, 1996; Windsor, Piche, & Locke, 1994). DeLeon and Iwata evaluated the consistency of results obtained using the MSWO, paired-choice preference assessment, and a multiple stimulus format similar to the FO in that a set of stimuli was presented and stimuli continued to remain available after selection by participants. DeLeon and Iwata found that participants’ resulting preference hierarchies were more consistent across five repeated administrations of the pairedchoice procedure as compared to the others assessments. Similarly, Windsor et al. compared the consistency obtained across six administrations of a paired-choice format and multiple-stimulus format and found consistency was greater across successive administration of the paired-choice format, replicating the results of DeLeon and Iwata. The test–retest reliability of paired-choice format may be due to a number of factors. Unlike multiple-stimulus formats that may require individuals to select among a large set of stimuli, paired-choice preference assessments only require participants make selections between two choices on any given trial. As such, reliable results of multiple stimulus methods may be dependent on certain prerequisite skills on the part of the person being assessed, e.g., the ability and self-control to scan all of the available stimuli before making a selection, the sensory capabilities to distinguish between the stimuli, and perhaps the ability to ignore characteristics that draw one’s attention but are not correlated with any responsestrengthening aspects of the stimuli. Thus, paired-choice formats may be more appropriate for a broader set of individuals. Indeed, the paired-choice preference assessment has been shown to be capable of detecting preference for stimuli in individuals exhibiting a broad range of intellectual abilities and skills including children (Fisher et al., 1992; Piazza et al., 1996), adults with intellectual and developmental disabilities (DeLeon & Iwata, 1996), children with visual impairments (Paclawskyj & Vollmer, 1995), and typically developing toddlers (Cote, Thompson, Hanley, & McKerchar, 2007). Thus, the reliability and broad applicability of the paired-choice preference assessment make it appealing for extension into understudied areas. One particularly understudied area is the examination of methods to identify preferred forms of social interaction. Methods to identify preferred forms of social interaction may be useful if food or edibles are contraindicated (e.g., as in individuals with Prader–Willi Syndrome), if individuals lack the prerequisite skills to play appropriately with toys or activities, in cases when carrying and delivering potential reinforcers are impractical, or if the target behavior warrants a form of attention specifically (e.g., a request for attention). Further, there is accumulating evidence in typically developing individuals and nonhuman animals that repeated exposure to energy-rich food may be associated with an increased risk of obesity (see Anzman, Rollins, & Birch, 2010; Kenny, 2011 for reviews). In addition to the immediate health effects of eating high-caloric foods, their delivery might be accompanied by long-term changes in food preferences that in turn lead to health challenges later in life. Such evidence might suggest that reinforcement-based interventions relying heavily on food could affect long-term food preferences and have unintended downstream health effects. Health effects not withstanding, brainimaging studies in obese and overweight humans have identified changes in the dorsal striatum indicative of reward hyposensitivity, the net result of which may be increased seeking of palatable foods (e.g., Stice, Spoor, Bohon, & Small, 2008). Although studies on the relative preference for food versus social interactions have not been conducted, increases in food seeking might exacerbate existing deficits in communication or social skills (to the extent that behavior maintained by food displace behavior maintained by social interactions) such as those present in individuals with autism. Therefore, overuse of food in reinforcement-based interventions may be associated with both long-term health effects and food preferences, the latter of which may be contraindicated for particular populations. Both possibilities invite a renewed need to identify effective reinforcers in other classes of stimuli, including social interactions. Unfortunately, a number of difficulties may be encountered when attempting to evaluate individuals’ preferences for social interactions. Whereas the appearance of food items is strongly related to their taste, social interactions do not necessarily have antecedents, such as physical appearance, that reliably signal their availability. Likewise, two tangible items can easily be presented at the same time by the same person. However, it may be difficult for the same therapist to signal the availability of two or more forms of social interaction. One approach to ameliorating these problems may be to present choices between picture-cards or icons that result in brief access to some other extended activity (Graff & Gibson, 2003; Hanley, Iwata, & Lindberg, 1999). However, this approach
364
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
may require extended experience or sophisticated discriminative repertoires not present in all individuals (see for example Clevenger & Graff, 2005). Other data suggest that use of pictures as opposed to tangible items may result in greater variation across selections (Higbee, Carr, & Harrison, 1999). If certain individuals are consistent in the forms of social interaction they provide to others, an alternative approach may be to assess individuals’ preferences for different staff members (Rapp, 2005; Sturmey, Lee, Reyer, & Robek, 2003). Rapp (2005) examined an individual’s preference for staff and caregivers by allowing the participant to approach one of the two adults present in a room. Approach responses resulted in attention from the adult. Although the author did not attempt to control or equate the attention provided by the adults, the participant showed clear preferences for particular staff and caregivers. Results from this study suggest that assessing preferences for forms of social interaction might be possible in the context of a paired-choice assessment. Smaby, MacDonald, Ahearn, and Dube (2007) evaluated an alternative method for examining the reinforcing effectiveness of different forms of social interaction in children with autism. The experimenters used a single-operant arrangement to assess the efficacy of three forms of social interaction (i.e., tickle, head rub, and praise). One of the three social interactions was delivered each time participants handed over a corresponding token to the therapist. Responding (handing over a token) in a social interaction condition was compared to responding in an extinction condition in which social interaction was not provided. Response rates were differentiated across the forms of social interaction, revealing a hierarchy across the social stimuli. However, previous research evaluating single-operant schedules has found that absolute rates of responding may not be well correlated with preference (Roscoe, Iwata, & Kahng, 1999). In addition, Smaby et al.’s inconsistent session termination criteria might have artificially produced relative increases in rate during social consequence sessions. This would have been due to experimenters continuing to carry out extinction sessions for 5 min or until 1 min elapsed without responding (the latter of which might tend to produce relative decreases during extinction) but terminating social consequences sessions after 1 min regardless of the level of responding. Finally, Smaby et al.’s procedures included two potential points of comparison that make comparing rates across social consequence sessions difficult: response rate during the extinction periods prior to each social consequence session and response rate during other social consequence sessions. Results may be ambiguous if, for example, levels of responding in social consequence sessions alone suggest one form of social interaction is most-preferred (e.g., high-five) but the level of responding during the preceding extinction session for that social consequence was highest (indicating a potential false-positive reinforcement-effect). Said another way, the procedure does not easily distinguish between forms of social interaction when baseline levels of responding vary across the forms. The use of social interactions as reinforcers has a number of potential advantages over tangible and edible stimuli in skill acquisition and behavior management programs. First, once identified, social interactions are unlikely to be as expensive as tangible or edible stimuli in terms of costs associated with purchase or storage. Second, social interactions may be more practical to deliver in community settings where it may be inconvenient for therapists to carry necessary edibles or tangibles. Third, the delivery of social interactions may be less stigmatizing in educational and community settings than edibles and tangibles. Fourth, the use of social interactions as reinforcers may promote generalization because their availability as reinforcers is not dependent on the presence of objects that may exert discriminative control over responding. Fifth, the inclusion of reinforcing forms of social interactions may strengthen social skill training programs, such as those used in early intensive behavioral intervention programs for children with autism. Indeed, the use of social interactions as reinforcers for individuals with autism seems particularly well suited given the nature of their social deficits (Volkmar, Lord, Bailey, Schultz, & Klin, 2004). Sixth, the identification of reinforcing forms of social interactions increases the array of available stimuli for individuals on caloric or food restricted diets. Finally, the inclusion of effective reinforcers is an essential component of any reinforcement-based intervention. Skill acquisition programs and reinforcement-based behavior management programs can fail for a variety of reasons, only one of which is the misidentification of an effective reinforcer. Therefore, identification and inclusion of reinforcing stimuli in reinforcement-based interventions may have two benefits: (1) they may increase the likelihood that interventions are effective and (2) they may reduce the time and effort related to repairing interventions that are initially ineffective. Taken together, there would seem to be a clear need for a renewed investigation of methods for the identification of preferred forms of social interaction. The purpose of this study was to examine a method of identifying individuals’ preferences for forms of social interaction in a paired-choice format and evaluate whether those forms identified as highly preferred serve as reinforcers. In an initial phase, we presented participants with pairs of therapists who, when approached, delivered a specific form of social interaction. In a second phase, we alternated therapists assigned to deliver the recently established highest and lowest preferred form of social interaction to determine whether participant approaches were under the control of the form of social interaction or idiosyncratic features of the therapist. Finally, we evaluated whether the form of social interaction identified as most preferred functioned as a reinforcer. 2. Methods 2.1. Participants and setting Five children referred for assessment and treatment of severe problem behavior participated. Alex was a 12-year-old boy diagnosed with autism and cerebral palsy. He was proficient at making choices, but had ambulatory difficulties due to
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
365
cerebral palsy. He was referred for non-compliant and aggressive problem behavior. Sofia was a 10-year-old girl diagnosed with Attention Deficit Hyperactivity Disorder (ADHD), Pervasive Developmental Disability Not Otherwise Specified (PDDNOS), and Bipolar Disorder. She attended an inclusive elementary school classroom. She was referred for verbal threats and aggressive, self-injurious, and destructive behavior. Chris was a 6-year-old boy diagnosed with ADHD. He attended an inclusive elementary school classroom and did school work at the level of his peers. He was referred for aggressive behavior. Kyle was a 17-year-old boy diagnosed with Cognitive Disorder, Mood Disorder, Mild Intellectual Disability, and Lennox– Gastaut Syndrome. He attended a self-contained high school special education classroom. He was referred for treatment of aggressive and destructive behavior. Rutherford was a 6-year-old boy diagnosed with ADHD. He was referred for aggressive and non-compliant behavior. He attended a self-contained elementary school special education classroom. All participants could follow full sentence instructions, and exhibited vocal verbal behavior. Alex, Chris, and Sofia’s assessments were conducted in 2 m 3 m room in a university-based outpatient client. The room included a small table, two chairs, and a oneway mirror. Kyle’s assessments were conducted in a 3 m 3 m area of his school classroom that contained a table and two chairs. No other students were present at the time of the assessment. Rutherford’s assessments were conducted in a 2 m 2.5 m area in his school auditorium containing one chair. 2.2. Preference assessment We identified four forms of attention to be included in the assessment through an open-ended informal interview of participants’ teachers or caregivers about interactions that they seemed to enjoy. Each form of social interaction consisted of both physical contact (e.g., a high-five gesture) and a vocal statement (e.g., ‘‘High-Five!’’) presented in an exaggerated and playful tone (see Table 1 for operational definitions; durations are specified for cases in which the physical interaction did not end immediately after it began [cf. Hugs vs. High-Five]). For Alex, we evaluated Noodle Arms, High-Five, Head Rubs, and Hugs. For Kyle, we evaluated Thumbs Up, High-Five, Knuckles, and Rascal. For Chris, we evaluated Tickles, High-Five, Noodle Arms, and Squeezes. For Rutherford, we evaluated Thumbs Up, Head Rubs, Noodle Arms, and High-Five. For Sofia, we evaluated Hugs, High-Five, Knuckles, and Noodle Arms. During this phase of the assessment each of four therapists was assigned one of the four forms of social interaction identified for that participant. The therapist only delivered the form of interaction they were assigned. A total of 9 therapists were used to deliver forms of social interaction for the study. Therapists included students at the undergraduate, masters, and doctoral levels. 2.2.1. Trial blocks Assessments were conducted over a series of six blocks of 5 trials. Blocks were presented until each combination of therapist and form of social interaction had been paired with every other. This resulted in a total of 6 blocks or 30 trials. Each block consisted of an initial pre-exposure to two forms of social interaction delivered individually by one therapist each followed by five paired-choice trials in which participants could choose between those therapists and forms of social interaction. The following procedures were used with Sofia, Chris, Kyle, and Rutherford. Slightly different procedures (described below) were used with Alex to accommodate a physical disability due to cerebral palsy. 2.2.2. Pre-exposures In order for participants to contact the contingency for approaching a particular therapist to receive a particular form of social interaction, a pre-exposure was provided at the beginning of each trial block. During the pre-exposure, a combination of vocal instructions and manual guidance was used to prompt participants to the center of the room, and turn to face a wall where one therapist stood. Participants were then prompted to approach that therapist who delivered the assigned form of social interaction. Approach was defined as movement to within .3 m of either therapist for all participants except Alex. Participants were prompted to approach each therapist relevant for that block of trials and receive the form of social interaction relevant for that therapist. Pre-exposures continued until participants experienced each form of social interaction once.
Table 1 Operational definitions of social interactions. Label
Description
Therapist statement
Noodle Arms High Five Head Rubs Hugs Thumbs Up Knuckles Rascal Tickles Squeezes
Holding the participants’ hands, and wiggling their arms up and down for 2 s Holding up one hand with an open palm and making contact with participant’s hand once Rubbing the participant’s head lightly for 2 s Wrapping both arms around the participant for 2 s Raising both hands in a thumbs-up position for 2 s Extending a closed fist toward the participant’s hand and making light contact once Tapping the participant lightly on the shoulder once Tickling the participant under the arms for 2 s Wrapping both arms around the participant, and rocking side to side gently for 2 s
‘‘Noodle arms.’’ ‘‘High five.’’ ‘‘Head rubs.’’ ‘‘Hugs.’’ ‘‘Thumbs up.’’ ‘‘Knuckles.’’ ‘‘You little rascal.’’ ‘‘Tickles!’’ ‘‘Squeezes.’’
366
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
2.2.3. Trials At the beginning of each trial, the participant was re-positioned to the center of the room, both therapists associated with the current block of trials stood in opposite corners along the same wall, and the participant was instructed to ‘‘pick one.’’ Manual guidance was not used following the vocal cue. Therapists ignored participants (i.e., did not make eye contact) until an approach occurred. Approaches resulted in brief delivery of the social interaction associated with that therapist and initiation of a new trial. Therapists alternated left and right positions prior to each new trial. Observers recorded participants’ selections in each trial and the position (left or right) of the therapist. Observers scored ‘‘no selection’’ and a new trial was presented if participants did not make an approach response within 10 s. Participants were re-prompted to the center of the room within 4 s after the social interaction terminated. For Alex, due to the locomotor challenges associated with his cerebral palsy, pre-exposures and paired-choice trials took place with him seated at a table and two therapists seated across from him. Approaches were defined as touching one of the therapists with any part of his hand. As with other participants, approaches resulted in brief interaction and therapists switched positions after every trial. 2.3. Tracking test Following the preference assessment, we re-evaluated participants’ preferences for the high- and low-preferred forms of social interaction by switching the therapists associated with these forms of social interaction. This was done in order to determine if participants’ choices in the previous phase were under the control of the form of attention provided or idiosyncratic features of the specific therapist (e.g., appearance, gender). Specifically, the therapist previously associated with the most-preferred form of social interaction now delivered the least-preferred form and vice versa. Each block of trials began with pre-exposures as described above. As in the previous phase, blocks were preceded by a pre-exposure to each social-interaction/therapist and included five paired-choice trials. Therapists alternated left and right positions every trial. Two to four blocks were conducted in this phase due to the reduction in attention forms assessed from 4 (during the preference assessment) to 2. 2.3.1. Response measurement and reliability Trained observers recorded participants’ choices and therapist position on each trial. An agreement was defined as both observers having recorded the same selection for each trial. Reliability was calculated by dividing agreements by agreements plus disagreements and multiplying by 100 to produce a percentage. A second observer collected data independently on 59% of the trials. Observers agreed on 100% of trials. 2.4. Reinforcer assessment One form of social interaction was evaluated in a reinforcer assessment for each participant. To be included in the reinforcer assessment, a given form of social interaction needed to be selected most often during the preference assessment and continue to be preferred during the tracking test. One of the two target responses (defined below) was selected for each participant. Extinction baseline and fixed-ratio 1 (FR1) were used in ABAB design to evaluate the reinforcing efficacy of the highest preferred social interaction for each participant. 2.4.1. Session protocol All sessions were 5 min and preceded by a single exposure to the programmed contingency: The therapist said, ‘‘When you do this’’, manually guided the participant to complete the response, then said, ‘‘you get this,’’ and delivered either the most preferred form of attention during FR1, or nothing during baseline. All target responses resulted in delivery of the programmed consequence for that condition. 2.4.2. Response definitions 2.4.2.1. Block-in-bowl. Participants were seated at a table and presented with two identical bowls, one of which contained a small block. A response was recorded when the participant picked up the block, placed it in the other bowl, and released it. This response was used for Kyle, Alex, and Sofia. 2.4.2.2. Spot touching. Participants were seated at a table and presented with a laminated colored sheet of construction paper placed approximately one foot in front of them. Pre-session exposure and session conditions were identical to block-inbowl. A response was recorded any time a part of the participant’s hand touched the construction paper. This response was used for Chris and Rutherford. 2.5. Response measurement and reliability Trained observers recorded the occurrence of each participant response using !Observe software on a hand-held computer. The software automatically time-tags and stores events as they are scored, and is capable of reporting those data
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
367
as either percentage of intervals or frequency. A second observer collected data independently during 38% of sessions. We compared observer’s records using both average agreement within 10-s intervals and Pearson’s r. Average agreement within intervals was calculated by dividing the smaller number by the larger number scored by observers in each interval, averaging across all intervals, and multiplying by 100 to obtain a percentage. Intervals in which neither observer recorded a response were scored as an agreement (1.0). Intervals in which only one observer recorded no occurrences of behavior were scored as a disagreement (0.0). Average agreement within intervals was 94.2% (range, 77.8–100%). Pearson’s r was also calculated to compare session rates obtained by both observers and was r(27) = .998, p < .0001. 3. Results Results of the attention preference assessment are shown in Fig. 1. Noodle Arms was the highest preferred form of attention (80% of selections) for Alex. Thumbs Up was the highest preferred form of attention (100% of selections) for Kyle. Tickles was the highest preferred form of attention (60% of selections) for Chris. Thumbs Up was the highest preferred form of attention for Rutherford. Hugs was the highest preferred form of attention (73% of selections) for Sofia. Results of the preference assessment and tracking test are shown together in Fig. 2, however baseline data include only those trials from the initial preference assessment in which the most- and least-preferred forms of social interaction were presented together. Baseline was composed of 2 blocks of sessions (10 trials). Closed and open symbols indicate the forms of social interaction that were most- and least-preferred during the initial preference assessment. Results show that the relative preference of the most- and least-preferred forms of social interaction remained consistent across therapists, indicating that participants’ selections were under the control of the form of attention provided as opposed to idiosyncratic features of the therapists. Data for Chris and Sofia indicate that they initially approached the therapist previously associated with the most-preferred form of attention in the first block of trials following the therapist switch. However, choices in subsequent trial blocks indicate they began to track the form of social interaction.
Fig. 1. Results from phase 1 of the preference assessment. Data reflect the percentage of selections to each form of social interaction.
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
368
BL
BL
TRACKING TEST 100
100
Noodle arms
80
TRACKING TEST Thumbs Up
80
60
60
40
40
Rutherford 20
20
Hugs Alex
0
PERCENTAGE OF SELECTIONS
1
2
High Five
0
3
1
100
100
80
80
2
3
4
5
Hugs Thumbs Up
60
60
40
40
Rascal
20
Kyle
0
20
Noodle Arms
0 1
2
3
1
2
3
Sofia 4
5
TRIAL BLOCKS 100
Tickles 80 60 40
Squeezes
20
Chris
0 1
2
3
TRIAL BLOCKS Fig. 2. Results from phases 1 and 2 of the attention preference assessment. Data from the BL condition reflect the percentage of selections toward the mostand least-preferred form of social interaction from phase 1. Data from the therapist switch condition reflect selections made after the therapists associated with the most- and least-preferred forms of social interaction in phase 1 switched roles.
Results from the reinforcer assessment are shown in Fig. 3. The most-preferred form of attention functioned as a reinforcer for all participants. For Alex, we saw little or no responding during the initial baseline phase. Upon switching to the FR1 phase we saw a modest increase in responding. Reversing back to baseline, we again saw little or no responding. In the second FR1 phase we saw a relatively high level of responding (M = 5.5) indicating that his highest preferred form of attention, Noodle Arms, served as a reinforcer. For Kyle, we saw a variable decreasing trend in the initial baseline phase, followed by an increasing trend in responding during the first FR1 phase. Reversing back to baseline we saw near zero response rates. In the second FR1 phase we saw a relatively high level of responding (M = 37.6) indicating that his highest preferred form of attention, Thumbs Up, served as a reinforcer. For Chris, we did not see any responding in the initial baseline phase. In the first FR1 phase we saw high levels of responding, followed by no responding in the second baseline phase. In the second FR1 phase we saw an increasing trend in responding at a level comparable to the first FR1 phase indicating that his highest preferred form of attention, Tickles, served as a reinforcer. For Rutherford, we saw low to zero response rates in the initial baseline phase, followed by elevated levels of responding in the FR1 phase (M = 10.7). In the second baseline phase we saw a return of low to zero rates of responding. In the final FR1 phase we saw elevated rates of responding (M = 7.6) indicating that his highest preferred form of attention, Thumbs Up, served as a reinforcer. For Sofia, we saw low to zero rates of responding in the initial baseline phase, followed by elevated levels of responding in the FR1 phase (M = 5.73). In the second baseline phase we saw a return to low to zero rates of responding. In the final FR1 phase we saw an increasing trend in responding, indicating that her highest preferred form of attention, Hugs, served as a reinforcer. Overall, for each of the five participants, the highest preferred form of attention identified through the preference assessment functioned as a reinforcer.
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371 10
FR1
BL
BL
18
FR1
BL
FR1
BL
369 FR1
8
12 6 4
6 2 Alex
0 5
RESPONSESPERMINUTE
50
10
FR1
BL
15
BL
Rutherford
0
20
5
10
FR1
40
8
30
6
20
4
10
BL
10
FR1
BL
15
20
FR1
2 Kyle
0 5
10
15
20
Sofia
0 5
10
15
20
SESSIONS 5
BL
FR1
BL
FR1
4 3 2 1 Chris
0 5
10
15
20
SESSIONS Fig. 3. Response rates in the reinforcer assessment for block-in-bowl (Alex, Kyle, and Sofia) and spot touching (Rutherford).
4. Discussion This study evaluated a three-phase process for identifying preferred and reinforcing forms of social interaction. In the first phase, a paired-choice preference assessment was used to identify a hierarchy of preferred social interactions. A subsequent tracking test provided evidence that participants’ approaches were under the control of the form of social interaction delivered, as opposed to idiosyncratic features of the therapists. Finally, a third phase demonstrated that forms of social interaction shown to be both preferred and tracked functioned as reinforcers for all five participants. Like previous studies, a paired-choice format successfully led to the identification of a preference hierarchy (Fisher et al., 1992; Piazza et al., 1996). Also like previous studies, events identified as highly preferred via preference assessment functioned as reinforcers. These results extend previous research by evaluating different forms of social interactions in a paired-choice format and evaluating the generality of those preferences (i.e., the range of conditions in which maintain) across therapists. A number of the seminal studies on preference assessments included at least one form of social interaction (e.g., Fisher et al., 1992; Pace et al., 1985). However, social interactions were not highly ranked by participants in those studies. In addition, research on preference assessment outcomes has found that stimuli in particular classes (e.g., edibles, activities) tend to be similarly preferred (DeLeon, Iwata, & Roscoe, 1997). That is, food tends to be more highly preferred than activities. Additional research should be conducted on whether similar patterns are obtained with social interactions and, if so, to determine the relative preference for social interactions versus other stimuli. One limitation of the current study is that we did not compare the reinforcing efficacy of high- and low-preferred social interactions. Given the results of previous work (e.g., DeLeon et al., 2009), we would expect highly ranked social interactions to be more effective reinforcers than comparatively lower ranked social interactions. Future research should examine the generality of such patterns with social interactions.
370
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
Similarly, we did not evaluate the necessity of the tracking test. In preference assessments for edible or tangible stimuli, no analogous test exists. Indeed, with edible and tangible stimuli, the formal qualities of their consumption tend to be highly and uniquely correlated with their appearance (e.g., the best avocados are dark and slightly tender whereas the best bananas are bright yellow with few marks or dark spots). It is easy to imagine how the appearance and smell of certain foods can become discriminative for the properties that make them delicious. Evidence exists that certain human preferences may be driven by features (e.g., facial symmetry) that predict reproductive fitness (Prokosch, Yeo, & Miller, 2005). Any such preferences would likely interfere with the social interactions we were trying to assess. In addition, phenomenon such as stimulus overselectivity, or the tendency for behavior to come under the control of a narrow subset of potentially relevant stimuli, has been associated with certain intellectual and developmental disabilities such as autism (Schreibman & Lovaas, 1973) and general cognitive deficits like low mental age (Schover & Newsom, 1976). Therefore, it seemed reasonable to be skeptical of the variables controlling participants’ choices during the preference assessments especially given the brief nature of the assessment. For the purposes of this study, the tracking test provided evidence that social interactions evaluated during the reinforcer assessment were preferred. However, we did not evaluate the necessity of the tracking test. Future studies may wish to examine the degree to which phenomena like overselectivity affect the interpretation of preference assessments. Although we did not evaluate it, it is also possible that the value of a given social interaction was related to the history of reinforcement associated with the therapist delivering it (e.g., one might prefer a high-five from person A more than person B). Results of the tracking test suggest such preferences in the current study were not strong enough to override differences between the high and low-preferred forms of social interaction. Future researchers may wish to examine this interaction. It is worth nothing that the assessment of preference for forms of social interaction was fairly personnel intensive. Four therapists were involved in the preference assessment: one for each form of social interaction. Although this may be a practical limitation of our procedure in clinical settings, assessing the generality of and individual’s preferences for forms of social interaction across different therapists (as in the tracking test) necessarily requires more than one therapist. Having each therapist deliver only one kind of social interaction was designed to increase the likelihood that approaches to specific therapists were more likely under the control the form of social interaction (i.e., we felt it would lead to an increases in the reliability of the assessment). Previous research with paired-choice procedures suggests their findings are highly reliable across time (DeLeon & Iwata, 1996). Future research may examine the reliability of the procedures described here across time or compare the reliability of these procedures to those involving choices between social interactions represented by other tangible stimuli. Future research might also examine whether similar results can be obtained when forms of social interactions are delivered by the same therapist and represented using picture cards. Another promising approach may be to represent social interactions using brief video clips (cf. Snyder, Higbee, & Dayton, 2012). A potential application of the procedures described here is in the treatment of severe problem behavior maintained by attention. In some cases, it may be impractical or unethical to place problem behavior on extinction (e.g., aggression toward peers or severe self-injury). An alternative approach may be to provide attention following both appropriate and problem behavior, but vary the quality of attention provided following each. For example, highly preferred forms of attention may be provided following appropriate requests and low-preferred forms following problem behavior. In this way, the contingencies may still favor appropriate requests while not placing problem behavior on extinction (cf. Athens & Vollmer, 2010). The procedures described in the current study contribute to the literature on preference assessment by applying a variation of the paired-choice preference assessment to forms of social interaction. We also demonstrated a procedure that can distinguish between selections that are under control of a form of social interaction versus features of the individual delivering said social interaction. Finally, we showed that this approach to preference assessment identified social stimuli that were effective reinforcers. These procedures may be especially useful in cases in which it is necessary to identify preferred forms of social interactions for use in acquisition programs as well as interventions for problem behavior. References Anzman, S. L., Rollins, B. Y., & Birch, L. L. (2010). Parental influence on children’s early eating environments and obesity risk: Implications for prevention. International Journal of Obesity, 34, 1116–1124. Athens, E. S., & Vollmer, T. R. (2010). An investigation of differential reinforcement of alternative behavior without extinction. Journal of Applied Behavior Analysis, 43, 569–589. Clevenger, T. M., & Graff, R. B. (2005). Assessing object-to-picture and picture-to-object matching as prerequisite skills for pictorial preference assessments. Journal of Applied Behavior Analysis, 38, 543–547. Cote, C. A., Thompson, R. H., Hanley, G. P., & McKerchar, P. M. (2007). Teacher report and direct assessment of preferences for identifying reinforcers for young children. Journal of Applied Behavior Analysis, 40, 157–166. DeLeon, I. G., Frank, M. A., Gregory, M. K., & Allman, M. J. (2009). On the correspondence between preference assessment outcomes and progressive-ratio schedule assessments of stimulus value. Journal of Applied Behavior Analysis, 42, 729–733. DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a multiple-stimulus presentation format for assessing reinforcer preferences. Journal of Applied Behavior Analysis, 29, 519–532. DeLeon, I. G., Iwata, B. A., & Roscoe, E. M. (1997). Displacement of leisure reinforcers by food during preference assessments. Journal of Applied Behavior Analysis, 30, 457–484. 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. Graff, R. B., & Gibson, L. (2003). Using pictures to assess reinforcers in individuals with developmental disabilities. Behavior Modification, 27, 470–483.
C.J. Clay et al. / Research in Developmental Disabilities 34 (2013) 362–371
371
Hagopian, L. P., Long, E. S., & Rush, K. S. (2004). Preference assessment procedures for individuals with developmental disabilities. Behavior Modification, 28, 668–677. Hanley, G. P., Iwata, B. A., & Lindberg, J. S. (1999). Analysis of activity preferences as a function of differential consequences. Journal of Applied Behavior Analysis, 32, 419–435. Higbee, T. S., Carr, J. E., & Harrison, C. D. (1999). The effects of pictorial versus tangible stimuli in stimulus-preference assessments. Research in Developmental Disabilities, 20, 63–72. Kenny, P. J. (2011). Reward mechanisms in obesity: New insights and future directions. Neuron, 69, 664–679. Keyl-Austin, A. A., Samaha, A. L., Bloom, S. E., & Boyle, M. A. (2012). Effects of preference and reinforcer variation on within-session patterns of responding. Journal of Applied Behavior Analysis, 45, 637–641. Pace, G. M., Ivancic, M. T., Edwards, G. L., Iwata, B. A., & Page, T. J. (1985). Assessment of stimulus preference and reinforcer value with profoundly retarded individuals. Journal of Applied Behavior Analysis, 18, 249–255. Paclawskyj, T. R., & Vollmer, T. R. (1995). Reinforcer assessment for children with developmental disabilities and visual impairments. Journal of Applied Behavior Analysis, 28, 219–224. Piazza, C. C., Fisher, W. W., Hagopian, L. P., Bowman, L. G., & Toole, L. (1996). Using a choice assessment to predict reinforcer effectiveness. Journal of Applied Behavior Analysis, 29, 1–9. Prokosch, M. D., Yeo, R. A., & Miller, G. F. (2005). Intelligence tests with higher g-loadings show higher correlations with body symmetry: Evidence for a general fitness factor mediated by developmental stability. Intelligence, 33, 203–213. Rapp, J. T. (2005). An assessment of preference for caregivers and antecedents for problem behavior. Behavioral Interventions, 20, 301–311. Roane, H. S., Vollmer, T. R., Ringdahl, J. E., & Marcus, B. A. (1998). Evaluation of a brief stimulus preference assessment. Journal of Applied Behavior Analysis, 31, 605–620. Roscoe, E. M., Iwata, B. A., & Kahng, S. W. (1999). Relative versus absolute reinforcement effects: Implications for preference assessments. Journal of Applied Behavior Analysis, 32, 479–493. Schover, L. R., & Newsom, C. D. (1976). Overselectivity, developmental level, and overtraining in autistic and normal children. Journal of Abnormal Child Psychology, 4, 289–298. Schreibman, L., & Lovaas, O. I. (1973). Overselective response to social stimuli by autistic children. Journal of Abnormal Child Psychology, 1, 152–168. 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. Snyder, K., Higbee, T. S., & Dayton, E. (2012). Preliminary investigation of a video-based stimulus preference assessment. Journal of Applied Behavior Analysis, 45, 413–418. Stice, E., Spoor, S., Bohon, C., & Small, D. M. (2008). Relation between obesity and blunted striatal response to food is moderated by TaqlA A1 allele. Science, 322, 449–452. Sturmey, P., Lee, R., Reyer, H., & Robek, A. (2003). Assessing preferences for staff: Some pilot data. Behavioural and Cognitive Psychotherapy, 31, 103–107. Volkmar, F. R., Lord, C., Bailey, A., Schultz, R. T., & Klin, A. (2004). Autism and pervasive developmental disorders. Journal of Child Psychology and Psychiatry, 45, 135–170. Windsor, J., Piche, L. M., & Locke, P. A. (1994). Preference testing: A comparison of two presentation methods. Research in Developmental Disabilities, 15, 439–455.