Unmasking social functions: Outcomes from a retrospective consecutive case series of 19 applications

Abstract

Prior research has identified the potential dangers associated with challenging behavior, specifically self‐injurious behavior, and has thus highlighted the need to employ protective procedures to ensure the safety of individuals with self‐injurious behavior. Although protective procedures can sometimes suppress responding, some small‐n studies suggest they can also elucidate or unmask social functions when the initial functional analysis indicates that challenging behavior is only automatically maintained. However, large‐scale studies of functional analysis outcomes indicate that co‐occurring automatically and socially maintained challenging behavior is relatively uncommon. We conducted a retrospective consecutive case series study to describe a set of procedures to unmask social functions when the initial functional analysis indicated an automatic function. Results suggest that protective procedures unmasked social functions in 26.32% of cases.

Individuals with intellectual and developmental disabilities often engage in challenging behavior, including self‐injurious behavior (SIB; Bowring et al., ²⁰¹⁷; Emerson et al., ²⁰⁰¹; Steenfeldt‐Kristensen et al., ²⁰²⁰). SIB can jeopardize the individual's safety—injuries incurred by individuals with histories of SIB are widely documented (e.g., Hyman et al., ¹⁹⁹⁰; Rooker et al., ^{2018, 2020}). For example, Hyman et al. (1990) conducted a records review of 97 individuals who engaged in challenging behavior, including SIB, while admitted to an inpatient hospital. The researchers found that 77% of the participants had previously sustained injuries ranging from minor afflictions (e.g., scratches and bruises) to severe damage (e.g., retinal detachment, cauliflower ear).

Functional analyses (Iwata et al., ^1982/1994) identify variables maintaining challenging behavior and provide a foundation for individualized treatment approaches. Notably, large‐scale studies of functional analysis outcomes indicate that SIB is often maintained by a single social function (Melanson & Fahmie, 2023). Specifically, Melanson and Fahmie (2023) reported on all functional analysis outcomes from the literature over the last 40 years, including 425 outcomes for SIB. A single function was identified in 79.29% of functional analyses of SIB, whereas multiple control was determined in only 11.06% (the rest were undifferentiated). Furthermore, 56.51% of the differentiated functional analysis outcomes suggested SIB was maintained by a single social function (i.e., escape, attention, or tangible), whereas 31.25% suggested it was solely automatically maintained. Interestingly, the results of Melanson and Fahmie suggested that multiple control with automatically maintained and socially maintained behavior is relatively uncommon, reported for only 4.2% of cases. Unfortunately, the researchers did not further break down the outcomes of multiply controlled challenging behavior; thus, to what extent automatically maintained SIB is multiply controlled is unclear, nor is it clear what other functions co‐occur with automatically maintained SIB when multiple control is indicated.

Disentangling the different types of multiply controlled SIB may be particularly challenging because of the unique patterns of responding observed in functional analyses of automatically maintained SIB. In the foundational study by Iwata et al. (1982/1994), two patterns of responding observed in functional analyses were considered indicative of automatic reinforcement: (a) highest responding in the no‐interaction condition and low levels of responding in the control condition (Child 4) and (b) high and variable levels of responding across all conditions (Child 2). However, in some cases, a combination of these two patterns may be observed. For example, Scheithauer et al. (2017) reported outcomes from functional analyses of two individuals who engaged in SIB, both of whom emitted higher rates of responding in all test conditions relative to the control condition. Generally, when SIB occurs at differentially higher levels in test conditions relative to the control, the variables maintaining SIB are attributed to the contingencies present in those test conditions. However, when responding is differentiated in the no‐interaction condition and occurs at a similar level in one or more of the social test conditions, it may not always be clear whether SIB is solely automatically maintained (and perhaps the stimuli in the control condition compete with SIB) or if SIB is multiply controlled by automatic and social reinforcers. Interpreting results of functional analyses can sometimes be further complicated by discrepant outcomes across different functional behavior assessment formats. For example, SIB can occur at high levels across all conditions (test and control)—suggesting it is automatically maintained—but, information garnered from other assessments, such as indirect assessments (e.g., Paclawskyj et al., ²⁰⁰¹), may suggest that it is also maintained by one or more social consequences.

Identifying co‐occurring automatic and social functions of SIB has implications for treatment. However, social functions may be difficult to identify when SIB is also automatically maintained. When functional analysis outcomes are ambiguous (e.g., high levels of SIB in the no‐interaction and one or more test conditions relative to the control) or do not corroborate those of other assessments (e.g., high levels of SIB in all test and control conditions when an indirect assessment suggests a social function), additional analyses may be necessary. For example, Scheithauer et al. (2017) conducted a subsequent functional analysis in which they systematically applied protective equipment across test conditions in accordance with a multiple‐baseline design. The researchers hypothesized that the protective equipment may diminish or potentially eliminate the sensory consequences of SIB, which may then allow identification of social functions. The baseline conditions mirrored the initial functional analysis test conditions, and the sensory extinction conditions included the same test conditions with the addition of specific protective procedures (helmet with face shield and Posey mitts for Greg and Posey mitts with response blocking of body‐to‐surface SIB for Carl). Carl's SIB decreased to near‐zero levels across all test conditions during sensory extinction, whereas Greg's SIB persisted in the modified attention condition. Thus, although the initial functional analysis suggested SIB for both participants was possibly multiply controlled, the use of protective equipment helped clarify the outcomes by confirming that SIB was solely automatically maintained for Carl and unmasking a social function for Greg.

A masked social function, for the purpose of this article, is a function that is not clearly identified in the initial functional analysis but that emerges or is further elucidated in a subsequent phase through various manipulations, such as the application of protective procedures. That is, rates of SIB are differentially higher in the social test condition relative to the control condition when protective procedures are applied. In addition to the work of Scheithauer et al. (2017), there are a few other examples of masked social functions in the literature (e.g., Kuhn & Triggs, ²⁰⁰⁹; McKerchar et al., ²⁰⁰¹). For example, Kuhn and Triggs (2009) reported on outcomes of a multielement functional analysis for an individual, Heather, who exhibited automatically maintained SIB. Like Carl and Greg in Scheithauer et al., Heather engaged in high rates of SIB across the alone, attention, and escape test conditions and low to zero rates in the control. In the subsequent assessment, the researchers used a padded helmet to attenuate reinforcement presumed to result directly from the occurrence of SIB. SIB persisted in the attention condition and decreased in the alone and escape conditions, suggesting that her SIB was socially maintained (by positive reinforcement in the form of attention) and automatically maintained. McKerchar et al. (2001) also found that protective equipment helped to unmask a social attention function.

Perhaps because of the difficulties associated with disentangling different types of multiply controlled SIB, it appears that only a few studies with small sample sizes have examined this phenomenon of masked social functions. Therefore, the purpose of our study was to expand this work by conducting a retrospective consecutive case series study (Hagopian, 2020)¹ to (a) describe a set of procedures to unmask social functions when the initial functional analysis indicated an automatic function and (b) determine the most commonly unmasked social functions within our sample.²

METHOD

Participants and setting

All data were collected during routine clinical care while the participants were admitted to an inpatient hospital unit for the assessment and treatment of challenging behavior between the years 2010 and 2020. For the initial screening, we identified all cases for which results from the initial phase of the functional analysis yielded an automatic function for any topography of SIB as determined through structured criteria (for specific criteria applied, see Hagopian et al., ²⁰²³). We then determined whether the clinicians overseeing cases that met these criteria for inclusion conducted a subsequent functional analysis that included at least one social test condition evaluated in the initial functional analysis where the only modification was the addition of protective procedures. The functional analysis with added protections may have been a subsequent phase within the same assessment or conducted as a separate functional analysis. In the latter case, we initially required that the analyses be no more than 4 weeks apart to limit the likelihood that any unmasked functions were a result of some other variable. However, the time between assessments did not exceed 4 weeks.

We defined protective procedures based on the categories of protections outlined in Frank‐Crawford et al. (2024), including protective equipment and mechanical restraint. Protective equipment referred to stimuli worn to mitigate the risk of injury without imposing restrictions on movement (e.g., arm guards, gloves). Mechanical restraint, on the other hand, included stimuli applied to mitigate the risk of injury by restricting movement such that the individual could not fully complete the response (e.g., arm splints). Supporting Information A provides illustrative examples of protective procedures. We further specified that protective procedures had to correspond with the target response. For example, if the target response was head‐directed SIB, a helmet would be considered a protective procedure but shin guards would not. Finally, we examined whether interobserver agreement was evaluated for at least 20% of sessions.

For cases that passed the initial screening, the following inclusion criteria were then applied: (a) no medications were started or discontinued during the assessments, (b) no protective procedure used in the initial functional analysis was removed in the subsequent functional analysis, and (c) no antecedents or consequences in the social test conditions were modified in the subsequent functional analysis test condition where protective procedures were applied. Any analysis for which protective procedures were used contingently was excluded.

Figure 1 depicts a flowchart illustrating the screening of participants for inclusion. We reviewed records for 361 cases. We excluded 242 cases for which the functional analysis did not indicate an automatic function for any topography of SIB. An additional 19 were excluded because the automatic function was for a topography other than SIB. Thus, 100 cases (27.70%) included an initial functional analysis that yielded an automatic function for at least one topography of SIB. In total, 86 of these cases were excluded from further analysis for various reasons. Most excluded cases (n = 73) did not include a functional analysis with added protections. Multiple protective procedure modifications (including the removal of protections applied in the initial functional analysis) led to the exclusion of four cases. For three cases, protective procedures were added during the no‐interaction and control conditions but were not evaluated for social test conditions, leading to their exclusion. The remaining cases were excluded because protective procedures were applied contingently (n = 2), the protective procedures did not target the automatic function (e.g., a singlet was added to address rectal digging, but the target response was SIB; n = 2), or medication changes occurred across assessments (n = 2).

FIGURE 1.

Flowchart of participant screening and inclusion.

Thus, 14 cases met all criteria for inclusion, hereafter referred to as “participants.” Two topographies of SIB were evaluated separately for five participants, for a total of 19 applications of the initial functional analysis. The initial functional analysis was defined as the preliminary assessment that included a no‐interaction condition indicating an automatic function and at least one social test condition (attention, escape, or leisure‐ or edible‐tangible). The functional analysis with added protections was subsequently conducted with at least one of the same social test conditions evaluated in the initial functional analysis. Protective procedures were applied noncontingently across all conditions in this subsequent functional analysis. For some participants, different combinations of protective procedures were evaluated or modified in a subsequent phase of the functional analysis with added protections. These were evaluated separately as a new functional analysis with added protections. In total, 24 functional analyses with added protections were evaluated across the 19 applications. To summarize, cases refer to patient charts that were reviewed during the screening process (N = 361); participants refer to individual cases that met all inclusion criteria (n = 14); applications refer to the number of initial functional analyses included in the study (with the number of initial functional analyses determined by the number of separate topographies of SIB assessed; n = 19).

Table 1 includes participant demographics and diagnoses. Participants included 14 individuals (11 male, 3 female) aged 9 to 21 years. Most participants identified as White and not Hispanic or Latinx. All participants had a diagnosis of autism spectrum disorder and at least one psychiatric or developmental disorder. Intellectual disability levels ranged from none to severe. Diagnoses were made prior to their inpatient hospitalization. All participants had a history of severe, treatment‐resistant SIB that resulted in numerous injuries prior to admission and thus warranted inpatient hospitalization.

TABLE 1.

Participant demographic information.

Participant	Gender	Age	Race and ethnicity	Psychiatric and medical diagnosis	ID	SIB topography	Auto subtype
P1	M	17	White & Not Hispanic or Latinx	SMD with self‐injury, DICD − Other, ASD	Severe	H‐SIB	A‐SIB 2
P2	M	10	White & Not Hispanic or Latinx	SMD with self‐injury, DBD, Chronic Motor Tic or Vocal Tic Disorder, Mixed Expressive/Receptive Language Disorder, ASD, ADHD, Anxiety Disorder	Moderate	H‐SIB	A‐SIB 3
P3	F	20	White & Not Hispanic or Latinx	SMD with self‐injury, ASD, Mood Disorder‐ Unspecified, Episodic	Moderate	H‐SIB, B‐SIB	A‐SIB 2
P4	M	12	White & Unspecified	SMD with self‐injury, ASD, Unspecified Disturbance of Conduct, Anxiety Disorder	Unspecified	H‐SIB	A‐SIB 2
P5	M	9	Unspecified	SMD with self‐injury, ASD	None	H‐SIB, S‐SIB	A‐SIB 2
P6	M	11	White & Not Hispanic or Latinx	SMD with self‐injury, DBD, ASD	Moderate	H‐SIB, S‐SIB	A‐SIB 3
P7	F	11	White & Not Hispanic or Latinx	SMD with self‐injury, DBD, ASD, and abnormality on Chromosome 15	Severe	H‐SIB, B‐SIB	A‐SIB 2
P8	M	13	White & Not Hispanic or Latinx	SMD with self‐injury, ASD, Pica, Tuberous Sclerosis complex	Severe	H‐SIB, S‐SIB	A‐SIB 3
P9	M	14	Unspecified	SMD with self‐injury, ASD	Severe	H‐SIB	A‐SIB 3
P10	M	14	White & Not Hispanic or Latinx	SMD with self‐injury, Conduct Disorder, Anxiety Disorder, ASD with Catatonia, Bipolar and Related Disorder	Severe	H‐SIB	A‐SIB 2
P11	F	10	White & Not Hispanic or Latinx	Mood Disorder‐ NOS, ASD	Unspecified	H‐SIB	A‐SIB 3
P12	M	17	White & Unspecified	SMD with self‐injury, ASD, Unspecified Disturbance of Conduct	Severe	H‐SIB	A‐SIB 2
P13	M	10	Other & Not Hispanic or Latinx	SMD with self‐injury, DICD, ASD, ADHD, Disruptive Mood Dysregulation Disorder	Moderate	H‐SIB	A‐SIB 3
P14	M	17	White & Unspecified	ASD	Mild	H‐SIB	A‐SIB 2

Note: ID = intellectual disability; SIB = self‐injurious behavior; M = male; F = female; SMD = stereotypic movement disorder with self‐injury; DICD = disruptive, impulse–control, and conduct disorder; ASD = autism spectrum disorder; DBD = disruptive behavior disorder; ADHD = attention deficit/hyperactivity disorder; NOS = not otherwise specified; H‐SIB = head‐directed self‐injurious behavior; B‐SIB = body‐directed self‐injurious behavior; S‐SIB = skin‐directed self‐injurious behavior; A‐SIB 2 = automatically maintained self‐injurious behavior Subtype 2; A‐SIB 3 = automatically maintained self‐injurious behavior Subtype 3.

Therapists received training in safely managing challenging behavior and conducting functional analyses. They conducted sessions under the supervision of doctoral‐level Board Certified Behavior Analysts. Session locations varied and included padded treatment rooms with a one‐way observation window, shared living areas, activity rooms with a one‐way observation window, and bedrooms. When necessary or if specified by the clinical team, padded mats were available for therapists to block or otherwise safely manage SIB. Staff protective equipment was available at the therapist's discretion unless required by the clinical team (in which case, it was mandatory that the therapist used the equipment for their safety).

Response and protective procedure definitions

Response definitions

Data collected on target responses included attempts to engage in and successful completions of SIB. Although target SIB responses were individually defined for each participant by their clinical team, the following definitions capture an inclusive description of the response topographies. Target responses included head‐directed SIB, such as hitting oneself in the head or banging head against a surface; body‐directed SIB, such as hitting or kicking the torso, arms, or legs; and skin‐directed SIB, which included behavior that damaged the skin barrier, such as scratching, pinching, or digging nails into skin. Five participants engaged in multiple topographies of SIB, and for four of these participants, the functional analysis with added protections revealed different response patterns across topographies. Thus, we report on each as a separate application.

In addition to SIB, observers collected data on other co‐occurring target responses, such as aggression and disruption. Topographies of these categories of behavior were also individually defined. Aggression generally included attempts or successes at hitting, punching, kicking, head‐butting, and hair‐pulling, and disruption typically included throwing items, ripping paper, and swiping materials. For all participants except P4, P10, and P14, consequences were programmed for multiple response topographies of challenging behavior during the functional analyses (e.g., for P1, the functional analysis assessed SIB, aggression, and disruption); however, SIB was the primary response that occurred in the functional analyses for all participants. That is, these non‐SIB responses either did not occur (e.g., P1), occurred at low levels (e.g., P2), or occurred at a variable rate across sessions (e.g., P6) such that their function could not be determined. In no case was a social function identified in the initial functional analysis based on outcomes for these non‐SIB responses. Some participants also engaged in other topographies of SIB for which the initial functional analysis was either undifferentiated (i.e., body‐directed SIB for P1, P9, P10, and P12) or suggested a single social function (body‐directed SIB was found to be escape maintained for P2). Supporting Information B includes the mean response rate for these topographies across all conditions of the initial functional analysis and functional analyses with added protections. For P11, data were also collected on self‐restraint, for whom data collection began approximately halfway through the initial functional analysis and continued into the functional analysis with added protections. The definition of self‐restraint included sitting on hands, crossing hands across the chest, placing hands behind the back, and holding them within or under clothing (e.g., shirt or pants).

Protective procedures

The clinical team responsible for each participant's care determined which specific protective procedures, or combination, to use during the assessments. All protective procedures were reviewed with and agreed on by the medical team and the participant's caregivers. In addition to those applied during all functional analyses with added protections, protective procedures were also applied in the initial functional analysis for P1, P4, P5, P7, and P10, which was held constant across both analyses. Table 2 includes the protective procedures held constant across both functional analyses as well as the added protections in the subsequent functional analyses.

TABLE 2.

Participant protective procedures information.

Participant	Topography	Protections in Initial FA	FA with AP	FA 2 with AP
P1	H‐SIB	Helmet w/o FS, 1 arm splint (right arm)	1 arm splint (left arm)	—
P2	H‐SIB	—	Canvas sleeves	—
P3	H‐SIB	—	Canvas sleeves	—
	B‐SIB	—
P4	H‐SIB	Helmet w/o FS	Wrist wrap	—
P5	H‐SIB	Helmet (unspecified)	Modified helmet (unspecified), gloves	Arm guards, double gloves, knee guards
	S‐SIB
P6	H‐SIB	—	Canvas sleeves	Helmet (unspecified)
	S‐SIB	—
P7	H‐SIB	Helmet (unspecified)	Knee pad, arm splints	—
	B‐SIB
P8	H‐SIB	—	Canvas sleeves, neck roll	—
	S‐SIB	—		—
P9	H‐SIB	—	Canvas sleeves	—
P10	H‐SIB	Helmet w/o FS	Helmet w/ FS, padded shirt, padded pants	Helmet w/ FS, stomach wrap, thigh wrap
P11	H‐SIB	—	Helmet (unspecified)	—
P12	H‐SIB	—	Helmet (unspecified)	—
P13	H‐SIB	—	Helmet (unspecified)	—
P14	H‐SIB	—	Arm splints	—

Note: FA = functional analysis; AP = added protections; FS = face shield; Unspecified = face shield information on helmet not provided; w/o = without; w/ = with; H‐SIB = head‐directed self‐injurious behavior; B‐SIB = body‐directed self‐injurious behavior; S‐SIB = skin‐directed self‐injurious behavior.

In addition to the above noted specific participant protections, the inpatient unit where the participants were admitted at the time of this study also routinely employed a number of other protections for all patients. These included routine monitoring by medical personnel, individualized session termination criteria, and the use of padded session rooms.

Data collection, interobserver agreement, and interrater agreement

Data collection

Trained observers collected data on target responses using a specialized computer software program (BDataPro; Bullock et al., ²⁰¹⁷). Throughout sessions, they collected data on the frequency and duration of all target responses and self‐restraint (P11), respectively. BDataPro converted frequency data to rate (responses per minute; RPM) and duration data to a percentage of session.

Interobserver agreement

Interobserver agreement for each target response was assessed by two independent observers using the mean‐count‐per‐interval method. Each session was divided into 10‐s intervals. The smaller number of responses recorded by one observer was divided by the larger number of responses scored by the other observer. Intervals in which both observers recorded 0 instances of the target response were assigned a value of 1. The values derived in each interval were then summed, divided by the total number of intervals, and multiplied by 100. A second observer collected data in at least 20% of sessions in each functional analysis for each participant. On average, interobserver agreement was assessed for 51.57% of sessions for the initial functional analysis and 49.73% of sessions for the functional analysis with added protections. The mean agreement coefficients were 95.09% for SIB (range: 75.52%–100%) in the initial functional analysis and 95.77% (range: 72.71%–100%) in the functional analysis with added protections. Agreement for P11's self‐restraint was 94.77% (range: 58.08%–100%).

Data extraction interrater agreement

Interrater agreement for screening and inclusion criteria was evaluated for 22.16% (n = 80) of the records reviewed; records for review were selected using a random number generator (i.e., random.org). An agreement was scored if the first and second authors included or excluded the case in the study. Interrater agreement was calculated by dividing the total number of agreements by the total number of agreements and disagreements and multiplying the resulting quotient by 100. The agreement for participant inclusion was 100%. Because the records reviewed by the second data collector were selected at random, we also determined the interrater agreement separately for the 14 included participants because only two (14.29%) had been randomly selected for secondary review in the initial randomization. The first and second authors reviewed each of the remaining 12 participants together and verified that they met all inclusion criteria.

General procedures

For each participant, functional analyses were completed during routine clinical care using procedures similar to those outlined by Iwata et al. (1982/1994). Stimuli (e.g., specific types of instructions, attention, and toys) included in test and control conditions were individualized for each participant based on indirect and descriptive assessments. In the functional analysis with added protections, the only modification to the social test conditions was the addition of protective procedures.

Initial functional analysis

The functional analysis consisted of three test conditions (no‐interaction, attention, and escape) for all participants. During the no‐interaction condition, there were no programmed antecedents or consequences; however, the therapist remained in the room with the participant for safety (to block head‐directed SIB as indicated in their protocol for P1, P5, and P10). During the attention condition, the therapist directed their attention toward an activity (e.g., reading a magazine) while the participant had access to moderately preferred toys. Contingent on the target response, the therapist delivered brief (i.e., 3–5 s) statements of attention (e.g., “You will hurt yourself, don't do that.”). All other behavior produced no programmed consequences. In the escape condition, instructions were presented using a three‐step prompting procedure that included vocal instruction, modeling, and physical guidance. When the target response occurred, the therapist delivered 30 s of escape, during which all instructional stimuli were removed, and the therapist turned away from the participant. In addition, some participants experienced a tangible (leisure or edible) test condition based on caregiver report or observation suggesting that challenging behavior may be related to accessing these stimuli: leisure‐tangible (P4 and P14) or edible‐tangible (P5 and P8). In the leisure‐tangible condition, the participant had 2‐min presession noncontingent access to the item. The session was initiated with the therapist removing the item and stating, “Time to put this away,” while restricting access to the item. The therapist returned the item for 30 s contingent on the target response. In the edible‐tangible condition, the participant sampled the item prior to the session. The therapist initiated the session by removing the snacks and delivering one piece of snack contingent on the target response. Across all test conditions (except the edible‐tangible test), the reinforcement interval ended after 30 s regardless of target responses (i.e., challenging behavior that occurred during reinforcement did not reset the reinforcement interval).

The test conditions were compared with a control condition using a multielement design for all assessments except an edible‐tangible assessment (P5) in which a pairwise design was used. During the control condition, the therapist remained in the room with the participant, who had continuous access to at least one preferred item. The therapist provided attention at least every 30 s (and more frequently if the participant initiated interaction), issued no instructions, and provided no programmed consequences for the target response. For P8, the control condition also included the noncontingent delivery of small pieces of preferred food placed on a plate, which the therapist replenished upon consumption. Similarly, during the edible‐tangible control for P5, the therapist delivered one small piece of food on a plate and replaced it after consumption. No other items were available during the edible‐tangible control for P5.

Functional analysis with added protections

The attention condition was evaluated in the functional analysis with added protections for P1, and the attention and escape conditions were evaluated for P2. For the remaining participants, all conditions evaluated in the initial functional analysis were repeated in the functional analysis with added protections. A multielement design was used with P4, P7, P8, P11, P12, P13, and P14. A pairwise design was used with P1, P2, and P6. For P9 and P10, a pairwise was combined with an extended no‐interaction phase. A combination of multielement and pairwise designs (i.e., different phases used different designs) was used with the remaining participants, P3 and P5. The aforementioned functional analysis procedures were also used in the functional analyses with added protections, with two modifications. For P1, the control condition included continuous attention rather than attention delivered on a fixed‐time schedule of 30 s. For P10, the control condition was replaced with an alone with toys condition due to elevated rates of aggression toward therapists.

For all participants, the protective procedures were applied noncontingently across all conditions and for the entirety of each session. Two separate functional analyses with added protections were completed for P5, P6, and P10, the outcomes of which were analyzed separately.

Process for functional analysis interpretation

Functional analysis interpretation consisted of several steps. First, we visually interpreted all functional analyses during the initial screen. Second, we applied structured criteria (for procedural details, see Hagopian et al., ²⁰²³) for all functional analyses. Third, we examined within‐session patterns of responding (see Roane et al., ¹⁹⁹⁹) for conditions associated with extended access to the hypothesized reinforcers (i.e., escape and leisure‐tangible condition) to determine whether SIB occurred more often in the presence than in the absence of the hypothesized establishing operation (EO). These steps were then repeated for the functional analysis with added protections.

Applying structured criteria

We plotted an upper criterion line between the second and third highest data points in the control condition and a lower criterion line between the second and third lowest control‐condition data points. For each test condition, we subtracted the number of data points below the lower criterion line from the number of data points above the upper criterion line (data points falling on the upper criterion line were counted as below the upper criterion line). The difference was then divided by the total number of sessions from the given test condition, yielding the quotient score. The process of applying structured criteria is described in greater detail by Hagopian et al. (2023).

Interpreting functional analysis results

For the purposes of this study, we sought to evaluate the extent to which social functions were present when the initial functional analysis outcome indicated that SIB was automatically maintained. Results for the functional analysis were interpreted as indicating SIB was automatically maintained based on one of the three following criteria: (a) the quotient score for the no‐interaction condition was equal to or greater than 0.50 (i.e., SIB was consistently higher in the no‐interaction condition than in the control condition), (b) the overall mean rate of SIB was higher across conditions with less external stimulation (e.g., no‐interaction, attention, and tangible) than in those with more external stimulation (e.g., escape and control), or (c) rates of SIB were high (e.g., greater than or equal to 1.5 RPM with fewer than 10% of sessions with zero responses; Hagopian et al., ²⁰²³) and occurred at similar levels in all conditions.

A masked social function was defined as a function that was not clearly identified in the initial functional analysis but emerged in a subsequent phase through the application of protective procedures. A social function was classified as not clearly identified if (a) the quotient score for the given social test condition was less than 0.50, (b) SIB was equally likely in the presence and absence of the EO in the escape or leisure‐tangible condition, or (c) there was a high degree of overlap between the no‐interaction and attention or edible‐tangible conditions.

To determine whether SIB was multiply controlled by one or more socially mediated reinforcers and automatically maintained, we began by examining the results of the initial functional analysis. Specifically, if the quotient score for a social test condition was equal to or greater than 0.50, suggesting that the response was both sensitive to one or more socially mediated reinforcers and automatically maintained, additional steps were taken to clarify whether SIB was solely automatically maintained or likely multiply controlled. First, we evaluated within‐session patterns of responding for conditions with more access to the hypothesized reinforcers (i.e., escape and leisure‐tangible) to determine if SIB persisted when the programmed EO for the reinforcer was absent (EO absent) versus when it was present (EO present; e.g., Roane et al., ¹⁹⁹⁹). If SIB persisted in the absence of the EO at a level similar to that observed when the EO was present, then we interpreted the outcome as solely automatic. For example, if SIB persisted after the removal of the instructional material in an escape condition, then we would not have concluded that SIB was maintained by escape, and further assessments may have been warranted to elucidate the functional analysis results. For those conditions where reinforcer access was brief such that within‐session analyses could not be conducted (i.e., attention and edible‐tangible), we determined whether SIB was multiply controlled by evaluating the level of overlap between the no‐interaction and social test condition. If the level of SIB was consistently higher in the attention or edible‐tangible condition than in the no‐interaction condition and the quotient score was at least 0.50 for the social test condition, then this would provide relatively clear evidence that SIB may be both automatically maintained and sensitive to social consequences. However, if there was considerable overlap such that SIB in the social test condition was not differentially higher than in the no‐interaction condition, then we could not clearly determine whether SIB was sensitive to the contingencies arranged in the social test condition, the reinforcement associated with engaging in the response itself, or both. Thus, in these cases, we felt it was prudent to hypothesize that SIB was automatically maintained and required further assessment to clarify the results of the social test conditions.

RESULTS

We grouped participants based on the conclusions drawn from both functional analyses into four distinct patterns. Table 3 provides a breakdown of these four general patterns of outcomes across participants.

TABLE 3.

Functional analysis outcomes across participants.

P ID	Functional analysis outcomes			Overall conclusions
	Initial	AP (1)	AP (2)
Pattern 1: Social Unmasked
P1	Auto	Attn		Auto & Social
P2	Auto	Attn & Escape	‐	Auto & Social
P3 a	Auto	Attn	‐	Auto & Social
P4	Auto	Auto & Attn	‐	Auto & Social
P5 b	Auto	Auto	Edible‐Tang	Auto & Social
Pattern 2: Social Elucidated
P6 b	Auto & Attn	Attn	Attn	Auto & Social
P6 c	Auto & Attn	Attn	Auto & Attn	Auto & Social
P7 c	Auto & Attn	Attn	‐	Auto & Social
P8 c	Auto & Edible‐Tang‐	Edible‐Tang	‐	Auto & Social
Pattern 3: Social Ambiguous
P7 a	Auto & Attn	Undifferentiated	‐	Auto & Social
P8 b	Auto, Attn, & Edible‐Tang	Undifferentiated	‐	Auto & Social
P9	Auto & Attn	Undifferentiated	‐	Auto & Social
Pattern 4: Automatically Maintained Only
P3 c	Auto	Undifferentiated	‐	Auto
P5 c	Auto	Undifferentiated	Undifferentiated	Auto
P10	Auto	Undifferentiated	Undifferentiated	Auto
P11	Auto	Auto	‐	Auto
P12	Auto	Auto	‐	Auto
P13	Auto	Undifferentiated	‐	Auto
P14	Auto	Undifferentiated	‐	Auto

Note: P ID = participant identification; AP = added protections; Auto = automatic; Attn = attention; Edible‐Tang = edible‐tangible.

^a Body‐directed self‐injurious behavior.

^b Skin‐directed self‐injurious behavior.

^c Head‐directed self‐injurious behavior.

Individual participant outcomes

Pattern 1: Social unmasked

We classified outcomes as Pattern 1 (social unmasked) when the initial functional analysis indicated that SIB was solely automatically maintained and that the addition of protective procedures unmasked a social function. This pattern was observed in five of 19 applications (26.32%). For P1 through P5, structured criteria for the initial functional analysis indicated that SIB was solely automatically maintained, and the functional analysis with added protections unmasked a social function.

Figure 2 depicts outcomes for P2 as an exemplar of Pattern 1. The initial functional analysis indicated that SIB was automatically maintained, as SIB occurred at high and similar levels in all conditions. In the functional analysis with added protections, canvas sleeves were added, and the attention and escape conditions were each evaluated using a pairwise design. Differentiated rates of SIB were observed in the attention condition compared to the control (quotient score = 0.93), indicating that SIB was also likely sensitive to attention. Additionally, SIB was elevated in the escape condition and the quotient score was 1.00, suggesting that the addition of protections unmasked an escape function.³

FIGURE 2.

Illustrative example of Pattern 1: Social unmasked. Data are shown for P2. PP = protective procedures; SIB = self‐injurious behavior. The solid phase line shows the change in protective procedures from the initial functional analysis to the functional analysis with added protections. The dashed phase line depicts the different phases of the functional analysis with protections (i.e., attention pairwise and escape pairwise).

Table 3 and Supporting Information D through G depict outcomes for P1, P3 (body‐directed SIB), P4, and P5 (skin‐directed SIB). For each case, the initial functional analysis indicated automatically maintained SIB and that the addition of protective procedures unmasked social functions. However, for P3 and P5, different response patterns emerged across topographies of SIB. Specifically, body‐directed SIB for P3 and skin‐directed SIB for P5 were determined to be multiply controlled; that is, they were both automatically maintained and maintained by attention (P3) or access to edibles (P5). However, head‐directed SIB for P3 and P5 were both found to only be automatically maintained (see Pattern 4 outcomes).

Pattern 2: Social elucidated

We classified outcomes as Pattern 2 (social elucidated) when the initial functional analysis indicated SIB was automatically maintained and likely sensitive to at least one social reinforcer and that the addition of protective procedures confirmed that social function. This pattern was observed for four applications (21.05%). For P6 (skin‐ and head‐directed SIB), P7 (head‐directed SIB), and P8 (head‐directed SIB), the initial functional analysis indicated SIB was automatically maintained and likely sensitive to at least one social reinforcer. The addition of protective procedures confirmed a likely social function, and overall outcomes suggested that SIB was multiply controlled by automatic and social reinforcement. Recall that for multiply controlled SIB, we evaluated the level of overlap between the no‐interaction and social test conditions and confirmed a social function if the level of SIB was consistently higher in the social condition relative to the no‐interaction condition.

Figure 3 includes the outcomes for head‐directed SIB for P7 as an exemplar of Pattern 2. In the initial functional analysis, P7 wore a helmet, and high rates of SIB were observed across all test conditions and the control condition. Differentially higher rates of SIB were observed in the attention condition (M = 8.42 RPM; quotient score = 1.00) and the no‐interaction condition (M = 6.20 RPM; quotient score = 0.56) relative to the control condition. In addition, SIB occurred at consistently higher rates in the attention condition relative to the no‐interaction condition. These data suggested that P7's head‐directed SIB was likely sensitive to attention and automatically maintained. The subsequent functional analysis included a knee mat and arm splints. The addition of protective procedures resulted in an overall decrease in the total level of SIB, and only the attention condition was differentiated relative to the control condition (quotient score = 1.00). Thus, the overall conclusions from both functional analyses for P7 were that head‐directed SIB was multiply controlled.

FIGURE 3.

Illustrative example of Pattern 2: Social elucidated. Data are shown for P7^c. PP = protective procedures; SIB = self‐injurious behavior. The solid phase line shows the change in protective procedures from the initial functional analysis to the functional analysis with added protections. P7^c = Head‐directed SIB.

Table 3 and Supporting Information H–I depict overall outcomes and outcomes for P6 and P8. Similar outcomes were obtained for P6 and P8. That is, in the initial functional analysis for P6, skin‐ and head‐directed SIB were sensitive to attention and automatically maintained, and for P8, head‐directed SIB was sensitive to edible‐tangible reinforcement and automatically maintained. The addition of protective procedures confirmed these social functions for both participants.

Pattern 3: Social ambiguous

We classified outcomes as Pattern 3 (social ambiguous) when the initial functional analysis indicated that SIB was automatically maintained and likely sensitive to at least one social reinforcer and that the addition of protective procedures did not clarify the social function. This pattern was observed for three applications (15.79%): P7 (body‐directed SIB), P8 (skin‐directed SIB), and P9.

Figure 4 illustrates the outcomes for P9 as an exemplar of Pattern 3. During the initial functional analysis, P9 engaged in elevated rates of SIB across all test conditions relative to the control condition. The quotient score was 1.00 for all test conditions. Additionally, SIB occurred at a consistently higher rate in the attention condition (M = 29.56 RPM) than in the no‐interaction condition (M = 13.70 RPM). The higher rate of SIB in the attention condition versus the no‐interaction condition suggested that SIB was likely sensitive to attention and automatically maintained. In the functional analysis with added protections, canvas sleeves were added. The addition of canvas sleeves resulted in an overall decrease in the total level of SIB, and the quotient score for each test condition was less than 0.50. These results did not clarify a social function. Supporting Information J and K included similar outcomes for P7's body‐directed SIB and P8's skin‐directed SIB, respectively.

FIGURE 4.

Illustrative example of Pattern 3: Social ambiguous. Data are shown for P9. PP = protective procedures; SIB = self‐injurious behavior. The solid phase line shows the change in protective procedures from the initial functional analysis to the functional analysis with added protections. The dotted phase line depicts the different phases of the same functional analysis with added protections (i.e., extended no‐interaction, attention pairwise, and escape pairwise).

Pattern 4: Automatically maintained only

We classified outcomes as Pattern 4 (automatically maintained only) when the initial functional analysis indicated that SIB was solely automatically maintained and that the addition of protective procedures confirmed that SIB was not sensitive to the social contingencies that were tested. This pattern was observed for seven applications (36.84%): P3 and P5 head‐directed SIB as well as P10 through P14.

Figure 5, 6 illustrates the outcomes for P11 as an exemplar of Pattern 4. No protective procedures were applied for P11 in the initial functional analysis, and this participant engaged in low but consistent levels of responding across all conditions. The quotient scores were below 0.50 for all test conditions. Her SIB may have been suppressed because she engaged in self‐restraint. Data on self‐restraint are depicted by the gray bars and were recorded for the last two series of the initial functional analysis. The mean percentage of self‐restraint was 30.20% in the no‐interaction condition and 38.74% in the control condition. Therefore, based on the levels of self‐restraint, P11 was classified as likely having Subtype 3 automatically maintained SIB (for more information on subtypes, see Hagopian et al., ^{2015, 2023}). In the functional analysis with added protections, a helmet was added. Similar to the initial functional analysis, the quotient scores for all test conditions were below 0.50, and P11 engaged in moderate levels of SIB and self‐restraint. These results did not indicate a social function, and we concluded that P11's SIB was likely solely automatically maintained. See Table 3 for overall outcomes and Supporting Information L through Q for similar outcomes obtained for head‐directed SIB for P3 and P5, as well as for P10, P12, P13, and P14.

FIGURE 5.

Illustrative example of Pattern 4: Automatic reinforcement only. Data are shown for P11. PP = protective procedures; SIB = self‐injurious behavior. The solid phase line shows the change in protective procedures from the initial functional analysis to the functional analysis with added protections. Self‐restraint data are recorded for control and no‐interaction sessions beginning at Session 19.

FIGURE 6.

Overall differentiated and undifferentiated outcomes for social test conditions.

Overall outcome

Functional analysis outcomes

Table 3 and Figure 6 depict the outcomes across both functional analyses. Outcomes across the 19 initial functional analysis applications indicated that SIB was automatically maintained (see Table 3, Column 2). Furthermore, a likely social function was also identified in seven applications: attention for P6 (skin‐ and head‐directed SIB), P7 (body‐ and head‐directed SIB), P8 (skin‐directed SIB), and P9, and an edible‐tangible function for P8 (skin‐ and head‐directed SIB). For the remaining applications, no clear social function was identified in the initial functional analysis.

The functional analysis with added protections unmasked social functions for five participants (see Columns 3 and 4 of Table 3). Across those five participants, six social functions were unmasked: attention for four participants (P1, P2, P3 [body‐directed SIB], and P4), escape for one participant (P2), and edible‐tangible for one participant (P5 [skin‐directed SIB]). Further, the social function identified in the initial functional analysis was confirmed for P6 (skin‐ and head‐directed SIB; attention), P7 (head‐directed SIB; attention), and P8 (head‐directed SIB; edible‐tangible). For P7 (body‐directed SIB), P8 (skin‐directed SIB), and P9, the protective procedures appeared to suppress responding; thus, the social functions identified in the initial functional analysis could not be confirmed. For P3 (head‐directed SIB), P5 (head‐directed SIB), P10, P11, P12, P13, and P14, no clear social function was identified in either the initial functional analysis or the functional analysis with added protections.

Table 6 (and the last column of Table 3) depict the overall outcomes across participants, specifically for the social test conditions of both functional analyses. Of the six unmasked functions, attention was the most prevalent (n = 4; 66.67%), followed by escape (n = 1; 16.67%) and edible‐tangible (n = 1; 16.67%). Attention was also the most confirmed social function (elucidated in three of the four functional analyses with added protections; 75%), followed by edible‐tangible (elucidated in one of the four functional analyses with added protections; 25%). Finally, attention was also the most suspected social function that could not be confirmed (suspected in all three functional analyses in Pattern 3, 100%), followed by edible‐tangible (suspected in one of the three functional analyses in Pattern 3, 33.33%).

Protective procedures

Table 2 displays the information for protective procedures. During the initial functional analysis, a helmet was used in seven of 19 applications (36.84%) and included an arm splint in one application. For the remaining applications, no protective procedures were used in the initial functional analysis. In total, 24 functional analyses with added protections were conducted. In 14 (58.33%), a single item was added. For the remaining 10 (41.67%) functional analyses with added protections, a combination of protective procedures was introduced (e.g., for P8, arm splints and a neck roll were added). The most commonly used items were helmets and canvas sleeves.

Subtypes of automatically maintained SIB

SIB was subtyped based on the level of differentiation between the no‐interaction condition and the control condition using procedures outlined by Hagopian et al. (2023).⁴ Subtypes of SIB are characterized by the degree of differentiation between the no‐interaction and control conditions and the presence of self‐restraint. Subtype 1 is characterized by clear differentiation between the no‐interaction and control conditions. Subtype 2 is characterized by a lower degree of differentiation (e.g., overlap of no‐interaction and control condition data points), and the presence of self‐restraint classifies Subtype 3. The level of differentiation was derived from the receiver operating characteristic curve, which identified the optimal cutoff point that maximized sensitivity and specificity for identification of Subtypes 1 and 2 automatically maintained SIB (Hagopian et al., 2023). If the level of differentiation was greater than 62.50% and the individual did not display self‐restraint during at least 25% of the session in the no‐interaction condition, then SIB was categorized as Subtype 1; if level of differentiation was equal to or less than 62.50% and they did not display self‐restraint during at least 25% of the session in the no‐interaction condition, then SIB was classified as Subtype 2. If SIB was already identified to be automatically maintained and the participant exhibited self‐restraint during at least 25% of the session in the no‐interaction condition, then SIB was classified as Subtype 3 (regardless of the level of differentiation). Based on these criteria, no SIB for any participant in our study met classification for Subtype 1. SIB was classified as Subtype 2 for eight participants and as Subtype 3 for six participants.

In the current sample, a social function (Patterns 1, 2, and 3) was equally likely to be observed for those with Subtype 2 (n = 6) and Subtype 3 (n = 6). However, slight differences were observed when considering the subtypes within each of these three patterns. Four of the five applications with Pattern 1 (unmasked social function) had SIB classified as Subtype 2 and one had SIB classified as Subtype 3. For the sake of this comparison, we collapsed data for Patterns 2 and 3, as overall outcomes from the initial functional analysis for both resulted in a hypothesis of multiply controlled SIB. For five of these seven applications, SIB was classified as Subtype 3, and SIB was classified as Subtype 2 for the remaining two. Thus, for this sample, 80% of the participants for whom a social function was later unmasked with protective procedures had SIB classified as Subtype 2. In contrast, 71.43% of participants for whom a social function (in addition to an automatic function) was identified in the initial functional analysis had SIB classified as Subtype 3. The sample was too small to make any substantive comparisons related to the category of unmasked function (e.g., attention vs. escape).

DISCUSSION

Our findings indicate that when an initial functional analysis identifies an automatic function, the addition of protective procedures in a subsequent functional analysis unmasked social functions (n = 5 out of 19 applications; 26.32%; Pattern 1), confirmed a hypothesized social function (n = 4; 21.05%; Pattern 2), did not further elucidate a suspected function (n = 3; 15.79%; Pattern 3⁵), or did not identify a social function (n = 7; 36.84%; Pattern 4). Six social functions were unmasked, with attention being the most common (66.67%), followed by escape (16.67%) and tangible (16.67%). Although protective procedures are used to maintain safety during the assessment (Frank‐Crawford et al., 2024) and treatment (Sullivan et al., 2025) process, the use of protective procedures may have other benefits. For example, modifying protections during the functional analysis may identify or help elucidate social functions of SIB for some individuals, perhaps via sensory extinction. Ultimately, at least one topography of SIB was found to be multiply controlled (Patterns 1, 2, and 3) for nine of the 14 participants (64.29%). SIB was found to be solely automatically maintained for only five participants (35.71%). These results align with the literature on masked functions (e.g., Kuhn & Triggs, 2009; McKerchar et al., ²⁰⁰¹; Scheithauer et al., ²⁰¹⁷), indicating that SIB may be both automatically maintained and sensitive to social reinforcement (often in the form of attention).

Although this study did not evaluate treatment outcomes, prior research suggests that treating multiply controlled behavior, specifically when automatic and social functions are involved, is complex (Boyle et al., 2024). Reinforcement‐based interventions, such as noncontingent reinforcement informed by competing stimulus assessments, have been most efficacious in addressing automatically maintained SIB (Rooker et al., 2018). However, for individuals with treatment‐resistant subtypes of automatically maintained SIB (Subtypes 2 and 3; Hagopian et al., 2015), interventions that rely on reinforcement alone are not always efficacious, perhaps because responding is predominantly allocated to SIB rather than item engagement (Hagopian et al., 2020). Research has found that augmenting tactics, such as response promotion and disruption, are more efficacious (Frank‐Crawford et al., 2023; Hagopian et al., ²⁰²⁰) and reallocation of responding to options associated with contingent reinforcement can disrupt treatment‐resistant subtypes of SIB (Rooker et al., 2022). Although effective, these strategies may reduce motivation for social reinforcers. Additionally, combining treatment components to target different functions (Boyle et al., 2024) is also advisable. For example, Scheithauer et al. (2017) reported treatment outcomes to validate the results of their sensory extinction analysis and observed reductions in SIB for both participants. The treatment package consisted of differential reinforcement with extinction for both participants—functional communication for Greg (social and automatic) and reinforcement for toy interaction and response interruption and redirection for Carl (automatically maintained only).

Interestingly, SIB for four of the five applications in which a social function was unmasked (Pattern 1) was classified as Subtype 2, whereas SIB for five of the seven in which a hypothesized social function was confirmed (Pattern 2) or failed to elucidate a social function (Pattern 3) was classified as Subtype 3. These findings, from a small sample of individuals with severe, treatment‐resistant SIB, should be interpreted with caution, but they are interesting and merit discussion. With Subtype 2, SIB occurs at high rates across conditions regardless of the level of stimulation present (Hagopian & Frank‐Crawford, 2018), potentially obscuring social functions. The application of protective procedures may help attenuate the consequences associated with automatically maintained SIB, without altering establishing operations for socially mediated consequences, thereby unmasking social functions. Subtype 3 SIB is characterized by the presence of self‐restraint in the functional analysis (Hagopian et al., 2015, 2023); furthermore, it is hypothesized that self‐restraint may occur to avoid aversive consequences produced by SIB (Fisher & Iwata, 1996). Thus, it is possible that self‐restraint may attenuate the consequences associated with automatically maintained SIB somewhat while having little effect on socially maintained consequences. This dynamic could result in an increased likelihood of identifying social functions without protective procedures in the initial functional analysis, as seen in Patterns 2 and 3. These hypotheses represent topics in need of further research.

Additionally, the functional analyses with added protections complemented, but did not replicate the initial functional analysis results. Of the 24 functional analyses with added protections, 20 included a no‐interaction condition. When considering these 20 functional analyses with added protections, nine resulted in undifferentiated outcomes, six indicated SIB was socially mediated, three continued to indicate SIB was solely automatically maintained, and two indicated SIB was multiply controlled. Had the functional analysis with added protections been conducted on its own, without the information garnered from the initial functional analysis, the conclusions would have been incomplete. Specifically, although the functional analysis with added protections unmasked or confirmed social functions in eight of these applications, it masked that SIB was automatically maintained in 15 of them.

The percentage of outcomes suggesting multiply controlled SIB in our study exceeded that reported in prior research (e.g., Melanson & Fahmie, ²⁰²³; Weber et al., ²⁰²⁴). For example, Weber et al. (2024) found that multiply controlled challenging behavior occurred for 52.10% of their sample, but it occurred with 64.29% of our participants. In addition, attention was identified as the most common social function in our study, whereas escape and access to tangibles were more common in Weber et al. One reason for these differences is that the data from the present study were gathered from individuals with automatically maintained SIB, whereas the previous literature reports on multiple control data among individuals with challenging behavior of all types and functions. In addition, prior studies have not always reported outcomes across topographies; perhaps automatic and attention functions are more commonly observed to co‐occur for SIB relative to other topographies. Given our small sample size, differences between our findings and those of Weber et al. may also reflect intersubject variability. Therefore, future research may include consecutive controlled case series studies (Hagopian, 2020) evaluating the prevalence of multiply controlled challenging behavior across functional classes.

Several additional limitations stem from the retrospective nature of this study and are relevant to guiding future research. First, the functional analysis design changed from a multielement to pairwise for several participants and the no‐interaction condition was not always included in the functional analysis with added protections, which was potentially problematic because we used the comparison between the no‐interaction and attention condition to, in part, determine multiply controlled SIB. These analyses were included because we were interested in the unmasking of social functions rather than additional confirmation of the automatic function. Thus, the inclusion of the no‐interaction condition in the subsequent functional analysis was ideal but not necessary for inclusion. Although the design changes align with previous research on masked social functions (e.g., Kuhn & Triggs, ²⁰⁰⁹), they can be problematic because Hagopian et al. (2013) found that design changes, such as moving from a multielement design to a pairwise design, were sometimes associated with clarifying ambiguous functional analysis outcomes. It is possible that for some individuals in the current study, the high and undifferentiated responding during the initial functional analysis was not indicative of automatically maintained SIB but rather a result of their failure to discriminate between the test conditions. For these individuals, design changes in the functional analysis with added protections may have enhanced discriminability and clarified outcomes. Although this possibility cannot be ruled out based on the current data alone, it is worth noting that outcomes from indirect and descriptive assessments supported that each of these participant's SIB was, at least in part, automatically maintained. Furthermore, the patterns we associated with automatically maintained SIB in the current study are empirically supported (e.g., Hagopian et al., ²⁰¹⁵; Iwata et al., ^1982/1994; Scheithauer et al., ²⁰¹⁷).

Second, the study is limited by its reliance on clinically derived data, which introduces the potential for procedural variations due to biases and preferences. All assessment‐related decisions, including where to conduct sessions, when to end analyses, and whether and what type of protective procedures to use, were made at the discretion of the clinical team in consultation with the participant's caregivers. There were no program‐specific criteria for conducting functional analyses with added protections; rather, decisions regarding when to conduct the functional analysis with added protections and the type of protections to use were based on clinical judgment. Thus, members of the clinical team may have chosen not to conduct functional analyses with added protections with some individuals with automatically maintained SIB. Moreover, if this occurred, there may have been systematic differences between individuals with whom the clinical team did or did not conduct a functional analysis with added protections. The possibility also remains that some forms of protective procedures (e.g., arm splints) may have inhibited SIB. Further, some of the protective procedures evaluated may not have adequately addressed the reinforcing properties of SIB, which may have precluded the identification of a social function for some individuals (e.g., those with Pattern 4 outcomes). Third, procedural fidelity data were unavailable. Thus, each of the aforementioned limitations sets the stage for improving future prospective research.

Fourth, most functional analyses in the current study targeted multiple topographies of challenging behavior. Beavers and Iwata (2011) suggested that the inclusion of multiple response topographies may increase the likelihood of outcomes indicating multiple control. Supporting Information B depicts the mean rate of topographies associated with undifferentiated outcomes. An evaluation of the function of these individual non‐SIB topographies in our study was inconclusive, as these topographies either rarely occurred or occurred variably such that a function could not be identified. In one notable case (P10), the initial functional analysis of body‐directed SIB yielded undifferentiated results; however, the functional analyses with protections unmasked attention and escape functions for this topography. We did not include body‐directed SIB in our analysis because it did not meet our inclusion criteria (the initial functional analysis did not suggest it was automatically maintained). Nevertheless, findings highlight areas for future research, such as evaluating the effects of protective procedures on other, nonautomatically maintained, topographies of challenging behavior.

Finally, we sought to corroborate the outcomes of the functional analyses by reviewing the participants' clinical records to determine whether the interventions included components that were designed to address the social functions identified for those with outcomes consistent with Patterns 1, 2, and 3 compared to Pattern 4, for whom only an automatic function was identified. However, most treatments included either contingent or noncontingent access to attention and tangible stimuli as well as tokens, breaks, or other stimuli arranged in instructional contexts, and it was not possible to determine why certain treatment components were added via the record review. Thus, it is unclear when components were added based on functional analysis outcomes or for other reasons. For example, applied and translational research on the prevention of challenging behavior suggests that differential reinforcement and—perhaps to a lesser degree—noncontingent reinforcement may be efficacious strategies for preventing the worsening or development of socially maintained challenging behavior (e.g., Fahmie et al., ^{2016, 2018}; for a discussion, see Richman, ²⁰⁰⁸). Thus, it is possible that some components were added to prevent the development of new social functions.

These findings are encouraging and further highlight that the functional analysis is an individualized, iterative hypothesis‐testing process. Protections are often arranged to reduce risk of injury when working with individuals who engage in SIB (e.g., Frank‐Crawford et al., ²⁰²⁴; Sullivan et al., ²⁰²⁵). Employing varying levels of protections during functional analyses and comparing outcomes across the assessments can also reveal different, and important, information. Prior research has suggested that using protective procedures during functional analyses may suppress responding and hinder the identification of a behavioral function (e.g., Borrero et al., ²⁰⁰²). However, when used strategically and in conjunction with evaluations of within‐session patterns of responding, these protections may help to unmask functions or clarify ambiguous outcomes (e.g., Kuhn & Triggs, ²⁰⁰⁹; McKerchar, 2001; Scheithauer et al., ²⁰¹⁷). For those individuals who engage in some of the most treatment‐resistant subtypes of SIB, any steps that may inform our interventions should be considered when designing assessments to help us better understand this complex behavior. Adopting such measures will likely facilitate the development of the most effective, comprehensive interventions for individuals with SIB.

AUTHOR CONTRIBUTIONS

The first author contributed to conceptualization, investigation, writing original draft, reviewing and editing manuscript, methodology, visualization, and data curation. The second author contributed to conceptualization, writing original draft, reviewing and editing manuscript, methodology, project administration, and supervision. The third author contributed to writing original draft, reviewing and editing manuscript, and conceptualization.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare.

ETHICS APPROVAL

These data were obtained during the course of providing clinical services that were sought by the individuals and their guardians to address the clinical targets described. The relevant party provided ongoing consent for treatment and approved all the procedures described. Institutional review board approval was obtained to disseminate these data.

Supporting information

Data S1 Supporting Information

Kaur, J. , Frank‐Crawford, M. A. , & Borrero, J. C. (2026). Unmasking social functions: Outcomes from a retrospective consecutive case series of 19 applications. Journal of Applied Behavior Analysis, 59(1), e70045. 10.1002/jaba.70045

DATA AVAILABILITY STATEMENT

All data are available on request from the corresponding author. Supporting Information A includes examples of protective procedures. Supporting Information B includes the mean rate of topographies associated with undifferentiated outcomes. Supporting Information C includes within‐session data. Supporting Information D through Q includes individual participant graphs.