Separate and combined effects of operant ABA renewal mitigation strategies

Authors: Carlos Henrique Santos Silva, Valdeep Saini

Abstract

Due to the undesirable effects of operant renewal for behavioral interventions, recent research has advocated for the advancement of renewal mitigation strategies. One strategy includes the use of extinction cues, which are stimuli used to establish discriminative control over responding in the second context that are subsequently transferred to the initial context. A second strategy involves context fading, which refers to progressively increasing the similarity between the second context and the initial context. The current study evaluated the separate and combined effects of these techniques using a preclinical human laboratory arrangement. Participants were exposed to the extinction cue strategy, the context fading strategy, both strategies, or neither strategy during a three‐phase ABA renewal procedure using differential reinforcement of an alternative response combined with extinction. The results indicated that context fading or combining context fading with an extinction cue was effective at mitigating renewal. The use of an extinction cue alone reduced renewal relative to the control group, but this difference was not statistically significant. The results are discussed in terms of methodological and theoretical differences across strategies as well as implications for future research on renewal mitigation strategies.

Operant renewal refers to a process by which an operant response reduced in one context reemerges because of a context change (Bouton et al., 2012). The term “context” is used to describe a wide array of stimulus conditions that could come to control behavior including physical stimuli, the passage of time, and states of deprivation or satiation (Bouton, 2002, 2019). In nonhuman studies, contexts can be delineated by different conditioning chambers, smells, colors, or the presence and absence of visual stimuli, among others (Bouton, 2019). Alternatively, in preclinical human studies that rely on laboratory computer tasks, contexts could be defined based on different task materials, colors, sounds, or task scenarios to distinguish between contexts (Kimball et al., 2020; Saini & Mitteer, 2020; Sullivan et al., 2018). In applied research, the context is usually described with respect to different physical settings (e.g., home, clinic; Saini et al., 2018) but in some cases can also include different mediators and the presence or absence of objects (Ibañez et al., 2019; Kelley et al., 2018).

“ABA renewal” refers to a specific type of renewal that begins with the acquisition of some response in one context (A), which is subsequently reduced in a second context (B). Renewal is observed if the response reemerges with reexposure to the initial context (A), without any change to the operant contingency across those contexts (Kimball & Kranak, 2022; Pritchard et al., 2016). Although the specific prevalence of ABA renewal outside of laboratory settings is unknown, it is the most commonly studied model of renewal with humans (Saini & Mitteer, 2020). Moreover, although extinction has historically been used alone to reduce the target response in the second context during the ABA model, recent studies have advocated for the use of extinction combined with differential reinforcement of an alternative response (DRA) because this approach better resembles the conditions under which renewal is likely to be observed in nonlaboratory settings (Kimball et al., 2023; Saini & Mitteer, 2020)

Understanding operant renewal is significant given that it is often viewed as a framework for relapse, which is defined as the failure of behavior change effects to maintain when environmental conditions vary (Podlesnik et al., 2017). In this sense, renewal is problematic because it affects intervention durability and generality, which are crucial for the success of behavior‐analytic practices (Mitteer et al., 2022; Saini et al., 2024). Notably, successful interventions delivered for a variety of clinical problems may have their effectiveness reduced in posttreatment conditions, such as when the context changes (Podlesnik & Kelley, 2015). Indeed, a number of recent applied studies have suggested that the prevalence of renewal is significant enough to warrant the attention of basic and applied researchers (Falligant et al., 2021, 2022; Muething et al., 2020).

Given the disruptive effects that renewal could have on intervention success, numerous studies have called for targeted renewal mitigation strategies (Haney et al., 2021; Kelley et al., 2018; Kimball et al., 2023; Randall et al., 2024; Saini & Mitteer, 2020). The advancement of such strategies could directly promote the success of behavioral interventions across settings, which is a primary goal of clinical behavior analysis (Stokes & Osnes, 1989). Fortunately, a number of renewal mitigation techniques have been proposed and require empirical investigation to test their validity and generality.

Establishing discriminative control over target responding using unique intervention‐correlated stimuli and subsequently transferring those stimuli across contexts has been one strategy suggested to mitigate renewal (Kimball et al., 2023; Saini et al., 2024). This strategy has been referred to as “extinction cues,” “treatment cues,” and “retrieval cues” and is hypothesized to be effective because stimuli associated with the intervention could exert discriminative control over responding in new situations (Bernal‐Gamboa et al., 2022; Nieto et al., 2017). For example, Bernal‐Gamboa et al. (2022) reinforced the lever pressing of rats in Context A (an operant chamber), extinguished lever pressing in Context B (a distinct second chamber) in the presence of different types of cues (e.g., intermittent or continuous light; more or less intense light), and then tested the magnitude of renewal by reintroducing Context A with or without the cue. A greater magnitude of renewal was observed when the cue was absent in the renewal test phase than when the cue used in Context B was transferred into Context A. These results have been replicated elsewhere (Trask & Bouton, 2016; Willcocks & McNally, 2014); however, the generality of this technique to DRA arrangements requires further investigation.

An alternative method for renewal mitigation could involve systematically increasing the similarity between contexts through a process known as context fading, which involves progressively increasing the similarity between the intervention setting (Context B) and the original context (Context A; Saini & Mitteer, 2020). For example, in a study involving young children who engaged in inappropriate mealtime behavior, Haney et al. (2021) examined treatment durability within a renewal framework when using context fading. In Context A (caregivers), children were allowed to escape from eating nonpreferred foods. In Context B (therapists), escape extinction was implemented. For a subset of participants, a caregiver was trained to progressively implement the intervention while the therapist systematically faded themselves out. In the last phase (Context A), the therapist was completely removed from the setting and the caregiver continued to implement the intervention. All three participants who were exposed to the fading procedure showed less renewal than those whose caregivers were not faded in (see also Kelley et al., 2018). Similar to the extinction cue strategy, the extent to which context fading serves as a viable renewal mitigation strategy when extinction is combined with DRA requires further investigation.

There have been no laboratory or applied studies that have directly compared extinction cues and context fading as renewal mitigation strategies for operant responding. Furthermore, there have been no studies that have examined the combined effects of more than one mitigation strategy on renewal. This is particularly noteworthy given that renewal preparations using respondent conditioning suggest that renewal is more effectively attenuated when combining multiple techniques (Laborda & Miller, 2013; Thomas et al., 2009). Finally, the generality of either technique to operant extinction with DRA has been largely understudied.

In light of the findings from the extant literature on renewal, a preclinical study was designed to further explore renewal mitigation. Using a computer simulation of operant ABA renewal, the purpose of this study was to evaluate the separate and combined effects of extinction cues and context fading as renewal mitigation techniques.

METHOD

Participants

Forty adults aged 18 years or older (M = 28.5, 18–46 years; 23 male, 17 female) were recruited by using posters that were displayed around a university campus containing a call for participation in a behavioral study. Demographic information beyond age and gender was not collected. All participants had an equal opportunity to earn a $100 Amazon gift card through a raffle regardless of their performance in the study. A single participant was selected at random as the winner at the end of the study.

Exclusion criteria included (a) individuals under the age of 18, (b) difficulty distinguishing between colors (e.g., color blindness), (c) difficulty sitting at a desk for prolonged periods (i.e., greater than 1 hr), (d) difficulty looking at a computer screen for prolonged periods, and (e) difficulty using a computer mouse. No participants were excluded based on these criteria.

Setting

The experiment occurred in a research laboratory room equipped with a table, a chair, a laptop, and a mouse, all of which were used during the study. Participants were seated facing the computer. An additional chair was placed in the corner of the room for the experimenter to sit and attend to any technical issues that may have arisen during the course of the study. The duration of the experiment was 45 min per participant.

Apparatus

Participation occurred by interacting with experimental software developed using an object‐oriented programming language known as Visual Basic. During the task, the context was represented by different‐colored screen backgrounds (yellow and blue) that changed according to each experimental phase. Stimulus colors were not counterbalanced across participants.

Across all contexts and phases, a black rectangle was visible on the left side of the screen and two circles (aligned vertically) were placed on the right side of the screen. Different responses were required according to each phase (i.e., either clicking the rectangle in baseline or dragging one circle onto the other during DRA). The shapes and background colors were chosen arbitrarily; however, clicking and dragging responses were selected due to their topographical difference and likelihood of being easily distinguishable.

At the top of the screen, a counter displayed the total number of points accumulated during the course of the experiment. When the required response produced points, the shapes disappeared, the schedule of reinforcement paused, and a textbox with the message “collect point +1” appeared at the center of the screen. This represented the reinforcer consumption response, and this period was included in the total session duration (45 min). Following the consumption response, the point was placed in the counter, the shapes reemerged at the same place, the schedule of reinforcement resumed, and participants were able to continue the task.

General procedure

A group design consisting of single‐case data sets was used. In total, there were three experimental groups and one control group, with participants randomly assigned to groups until there were 10 participants per group. Each participant within a group was exposed to either a different renewal mitigation strategy (experimental groups) or no mitigation strategy (control group).

Before the session began, each participant received general instructions about the study and how to use the computer mouse. The following vocal instructions were Welcome to our study on reward learning. Your task today will be completed on a laptop, and your goal is to obtain points during the task. You will use the mouse to respond and to figure out how to earn points. How you respond is completely up to you, and you may stop responding at any time. No credit is assigned to how well you play, but you are encouraged to score as many points as possible. When the screen displays “end,” you will call me over. Please hand the laptop back only when the screen says “end,” in the event of an emergency, or if you wish to withdraw from the study. Watches and cellular phones are not allowed in the experimental room. They must be safely stored away during session time. Do you have any questions?

All participants were exposed to all three phases, and each phase lasted 15 min. Each phase represented one context in ABA renewal. Phases changed automatically following 15 min, and the task ended automatically after 45 min (i.e., 15 min per phase).

In the first phase, Context A involved acquisition of the target response, wherein clicking on the black rectangle produced points according to a variable‐interval (VI) 10‐s schedule of reinforcement. This schedule was selected because Berry et al. (2014) indicated that higher rates of reinforcement (e.g., dense ratio schedules) in baseline are associated with a lower degree of renewal. We selected a relatively leaner VI schedule given that observing the renewal effect in the control group was necessary for comparative purposes and statistical analysis. Furthermore, similar preclinical studies of renewal (e.g., Sullivan et al., 2018) have used this schedule.

In the second phase, Context B simulated extinction plus DRA wherein clicking on the black rectangle was placed on extinction and dragging the lower circle onto the circle above it produced points on a VI 10‐s schedule. After each response, the lower circle reset back to its original position.

In the third phase, extinction plus DRA remained in effect while Context A was reintroduced (i.e., the operant contingencies in place during Phase 2 remained in effect during Phase 3 despite the reintroduction of the original context from Phase 1). However, depending on the group to which participants were assigned, the manner in which Context A was reintroduced differed. Across all phases, clicks or other responses outside the shapes had no programmed consequences.

Control group

The purpose of this group was to determine the degree of renewal in the absence of any mitigation strategies.

Phase 1 (Context A)

The screen background color was blue. In this phase, participants clicked on the rectangle to produce points in a manner consistent with the general procedure described above. Engaging in the alternative response produced no programmed consequences.

Phase 2 (Context B)

After 15 min, Phase 2 began. The screen background color switched from blue to yellow, and engaging in the target response (i.e., clicking on the rectangle) was placed on extinction. Engaging in the alternative response (i.e., dragging the lower circle onto the circle above it) produced points in a manner consistent with the general procedure.

Phase 3 (Context A)

After 15 min, or 30 min cumulatively, the third phase initiated. The screen background color switched from yellow to blue. The contingencies from Phase 2 remained in effect.

Extinction cue group

The purpose of this group was to evaluate the single effect of using an extinction cue to mitigate renewal.

Phase 1 (Context A)

Phase 1 was identical to the control group.

Phase 2 (Context B)

Phase 2 was identical to the control group, with the following A red triangle (cue) was placed inside the rectangle shape (the shape that previously produced points in Phase 1). The purpose of the red triangle was to establish discriminative control over clicking by serving as an S‐delta (S^Δ^) in a manner consistent with prior research.

Phase 3 (Context A)

Phase 3 was identical to the control group, with the following The red triangle from Phase 2 was also present and visible during Phase 3.

Context fading group

The purpose of this group was to evaluate the single effect of context fading as a renewal mitigation strategy.

Phase 1 (Context A)

Phase 1 was identical to the control group.

Phase 2 (Context B)

Phase 2 was identical to the control group, with the following Over the course of the 15‐min phase, the background color gradually and progressively changed from yellow to blue such that the yellow background from Phase 1 became identical to the blue background forthcoming in Phase 3. In other words, Phase 2 began with the yellow background from Phase 1 and ended with the blue background in Phase 3. Thirty transition shades were used to achieve this effect, with each shade appearing for 30 s before the next transition shade appeared. The transition between shades was not easily detectable by the human eye, as the entire phase represented a gradual shift in the background color.

Phase 3 (Context A)

Phase 3 was identical to the control group, with the following The background color did not abruptly switch from yellow to blue because it was already the correct shade of blue at the beginning of the phase.

Combination group

The purpose of this group was to evaluate the additive effects of using an extinction cue concurrently with context fading as a combined mitigation strategy.

Phase 1 (Context A)

Phase 1 was identical to the control group.

Phase 2 (Context B)

Phase 2 was identical to the control group, with the following First, the red triangle used for participants in the “extinction cue group” was also in place for the combination group. Second, the screen‐color‐fading technique used for the “context fading group” was also in place for the combination group.

Phase 3 (Context A)

Phase 3 was identical to the control group, with the following First, the red triangle from Phase 2 was also present and visible during Phase 3. Second, the background color did not abruptly switch from yellow to blue because it was already the correct shade of blue at the beginning of the phase.

Data analysis

Two primary measures were taken in all experimental phases for all (a) responses per minute of mouse clicks in the rectangle (i.e., target response) and (b) responses per minute of dragging one circle onto the other (i.e., alternative response). Responding across these two topographies provided a general indication of response acquisition, trends, and variability throughout the experiment. However, renewal was defined as the response rate of the target response in any of the first 5 min in Phase 3 exceeding the mean rate of the same response for the last 5 min of Phase 2 (Finch et al., 2022). Results that produced this pattern of responding were indicative of ABA renewal, which was assessed via differences in group means of single‐case data.

Proportion of baseline target responding was calculated to account for differences in the mean response rate across groups in Phase 1. In this respect, each minute of the mean target response in Phase 2 (Context B) and Phase 3 (Context A) was compared with the mean rate obtained in Phase 1 (Context A). Calculations involved the rate of the target response in each minute of Phase 2 and Phase 3 divided by the average rate of the target response observed across all baseline sessions (Fisher et al., 2019; Mace et al., 2010).

In addition to response rates, we calculated reinforcer rates for each participant across each phase by counting the total amount of reinforcers earned in a given phase and dividing that value by 15 (i.e., the duration of each phase).

The mean target response in each of the first 5 min in Phase 3 was compared using the Kruskal–Wallis nonparametric test, with a rejection criterion for Type I error of ⍺ = .05. During this analysis, the test variable H was calculated. This test was selected because the assumptions for a one‐way analysis of variance (ANOVA) were not met. In such cases, the Kruskal–Wallis test is a more appropriate and robust statistical test to determine whether there are statistically significant differences between more than two groups (Nwobi & Akanno, 2021). We decided to forego mixed‐effect analyses (e.g., Martinez‐Perez et al., 2022) due to our sample size (DeHart & Kaplan, 2019). If H was significant, the effect size was estimated using the formula r = z /√N, where r is the effect size, z the z score, and N the total number of participants in the comparison. This formula is an appropriate effect size index used for the Kruskal–Wallis test (Field, 2018), and this analysis allowed us to determine the magnitude of effects if there were any between‐group differences in the degree of renewal. Effect sizes greater than 0.10 were considered small, effect sizes greater than 0.30 were considered medium, and effect sizes greater than 0.50 were considered large (Cohen, 2016).

RESULTS

Table 1 displays the mean target and alternative responses per minute for each group in each phase, whereas Figure 1 displays mean trend levels and variability across groups for target responses (individual participant data are available in Appendix A of the Supplementary Information). In the first phase (Context A), participants in all groups acquired the target response; however, participants in the control (top left panel) and extinction cue (top right panel) groups displayed higher levels of the target response relative to the context fading (bottom left panel) and combination (bottom right panel) groups. An analysis was conducted to ensure that differences in responding in the third phase could not be attributed to differences in baseline rates of responding across groups. It was determined that there was no significant difference between groups with respect to Phase 1 response rates, H(3) = 1.301, p = .729. Pairwise comparisons with adjusted p values showed that there were no significant differences between the control and extinction cue groups (p = 1.000, r = .55), the control and context fading groups (p = 1.000, r = .21), the control and combination groups (p = 1.000, r = .12), the extinction cue and context fading groups (p = 1.000, r = .15), the extinction cue and combination groups (p = 1.000, r = .68), and the context fading and combination groups (p = 1.000, r = .22).

The mean target response rate concomitantly decreased for all groups at a similar rate. During Phase 2, the mean target response decreased from 72.4 to 7.7 responses per minute in the control group (89.3% decrease), 66.7 to 8.1 responses per minute in the extinction cue group (87.8% decrease), 40.3 to 9.9 responses per minute in the context fading group (75.4% decrease), and 51.2 to 8.5 responses per minute in the combination group (83.3% decrease).

When Phase 3 was introduced (i.e., reexposure to Context A), the target response patterns obtained from the control and extinction cue groups met the definition of renewal, with increases in the target response immediately following a context change and at a rate greater than the mean of the final 5 min of Phase 2. The largest magnitude of renewal was observed with the control group, which also generally produced more persistent responding across the entirety of Phase 3.

Figure 2 displays mean trend levels and variability across groups for alternative responses. In Phase 2 (Context B), the mean alternative response rate increased for all groups; however, there was variability in the acquisition rate and overall level across groups. For example, the alternative response occurred at a mean of 33.2 responses per minute and 21.5 responses per minute in the control (top left panel) and context fading (bottom left panel) groups, respectively. Despite this variability, the alternative response rate remained relatively stable on average across the duration of Phase 2. Interestingly, the three experimental groups produced slight increases in the mean alternative response rate in Phase 3 relative to Phase 2, whereas the control group produced a slight reduction in the mean alternative response. However, these differences in alternative response rate across Phases 2 and 3 were not statistically significant.

Figure 3 displays the mean target response rate in the last 5 min of Phase 2 with the means in the first 5 min of Phase 3. When comparing these rates, there was a significant difference between groups, H(3) = 30.00, p = .001. Pairwise comparisons with adjusted p values showed that there was a significant difference between the control (top left panel) and context fading (bottom left panel) groups (p = .000, r = .95), the control and combination (bottom right panel) groups (p = .000, r = 1.0), the extinction cue (top right panel) and context fading cue groups (p = .014, r = .68), and the extinction cue and combination groups (p = .006, r = .73). There was no significant difference between the control and extinction cue groups (p = 1.000, r = .27) or between context fading and combination groups (p = 1.000, r = .04).

Figure 4 displays the mean proportion of baseline for each group in Phase 2 and Phase 3, which compares the mean target response per minute in Phase 2 and Phase 3 as a proportion of the overall mean rate obtained in Phase 1. This analysis allows for the comparison of renewal magnitude between groups by only considering the rate of responding in Phase 2 and Phase 3 relative to baseline responding in Phase 1. The control (top left panel) group showed a higher degree of renewal for at least the first 10 min of Phase 3 when compared with the experimental groups. The extinction cue (top right panel) group displayed a lower degree of renewal in the first 5 min of Phase 3 than the control group. There was no evidence of renewal for the context fading (bottom left panel) and combination (bottom right panel) groups.

Table 2 displays the mean number of reinforcers acquired in each phase per group. All groups except for the control group showed a similar a slight increase in reinforcers obtained across phases, with no substantial differences across groups. Reinforcers obtained in the control group were somewhat equivalent. Taken together, the differences in responding across phases for each group are unlikely to be the result of reinforcement rates, which has previously been described as a contributing variable toward the occurrence of renewal (Podlesnik & Shahan, 2009).

DISCUSSION

In a preclinical human laboratory study, the present experiment was used to evaluate the separate and combined effects of extinction cues and context fading on the mitigation of ABA renewal. Taken together, all three experimental groups (extinction cue, context fading, combination) produced lower levels of renewal than the control group. However, only the context fading and combination groups reduced renewal to a significant level relative to the control group, suggesting that extinction cues used in isolation may be less effective at mitigating renewal than context fading used in isolation or combining extinction cues with context fading.

As an isolated renewal mitigation technique, context fading was effective at reducing renewal to near‐zero levels in Phase 3, suggesting that this may be a particularly robust procedure despite there being limited basic or translational research on this strategy. Haney et al. (2021) hypothesized that context fading provides a greater number of opportunities for the target context to be associated with nonavailability of reinforcement for the target response, thereby decreasing the likelihood of generalization failures. It may also be the case that this technique takes advantage of stimulus shaping (sometimes referred to as “errorless” stimulus control) wherein the topographical features of context stimuli are gradually changed over trials so that discriminated responding is more readily evoked across successive stimulus changes (Schilmoeller & Etzel, 1977; Stoddard & Sidman, 1967). As a result, errors in responding (i.e., renewal) are unlikely to occur given that stimulus control has been established gradually. It is possible that the present study produced a more reliable reduction in renewal when using context fading than was observed by Haney et al. (2021) given that the present study likely provided participants with a more subtle change in successive stimulus discriminations than was provided in their study. Moreover, it's possible that the present study was more effective at controlling extraneous variables given that Haney et al. (2021) conducted their study in an applied setting. Consequently, and given the relative simplicity in the contexts used in the present study (i.e., colored computer screen backgrounds), the generality and practicality of context fading in applied settings should be explored in future translational research.

The successive stimulus shaping strategy employed during context fading may have also contributed to the lower mean response rate and slower acquisition of the alternative response relative to the control group. The changing stimulus conditions across Phase 2 created by the reintroduction of Context A may have influenced the alternative response to some extent and induced competition between the target and alternative responses. As a result, the effects of context fading on alternative responses should be explored further in future research.

The extinction cue strategy failing to reduce renewal to a significant level appears to be at odds with prior research on this technique. For example, Willcocks and McNally (2014) used a 60‐s tone presented contingently (Experiments 4 and 6) or noncontingently (Experiments 1 and 6) as an extinction cue during ABA renewal and found that in all experiments the extinguished response was attenuated when the cue was used in Phase 3 but reoccurred when it was not used. In the present study, it is possible that the failure of the extinction cue to be effective could be due to selective stimulus control acquired by compound stimuli. It is possible that in Phase 1, both the blue background and the rectangle became a compound discriminative stimulus for the target response, whereas in Phase 2, both the yellow background and the red triangle inside the rectangle served as an S^Δ^ for the target response. When the participants were reexposed to Context A, in which the target response was established, the red triangle alone may have been initially insufficient to suppress the target response or compete with the previously established compound stimulus. This notion is consistent with previous research on stimulus control that has demonstrated that discrimination training with compound stimuli may lead to selective stimulus control when each stimulus is presented separately (Johnson et al., 1969; Perez et al., 2015). Ultimately, this may have led to the red triangle not effectively establishing control over target responding as expected. It is possible that more discernable extinction cues could produce more reliable mitigation effects (Dinsmoor, 1995; Halbur et al., 2021). The durability of an extinction cue to protect against renewal should be explored in future research, especially given that this strategy has been recommended in applied research (e.g., Greer et al., 2019; Kimball et al., 2023).

The combination group used both the extinction cue and context fading as strategies to mitigate renewal. Using visual analysis, the mean rates of responding in Phase 3 were not indicative of renewal. Using statistical analysis, there was a significant difference in the first 5 min of Phase 3 responding between combination and control groups. Furthermore, the overall pattern of responding in Phase 3 across these two groups was comparable. This suggests that context fading as a renewal mitigation technique is likely effective with or without an extinction cue and that the addition of an extinction cue does not significantly decrease renewal (near‐zero rates were produced for both groups). Considering the outcomes obtained when the two mitigation strategies (extinction cue and context fading) were used in isolation, it is likely that context fading played an important role in the success of mitigating renewal in the combination group given that each strategy produced different results when investigated separately.

Notably, one extension of this study was the application of renewal mitigation procedures to DRA as opposed to extinction alone. Incorporating an alternative response into the renewal framework may partly explain why some results in the present study differed from those of prior research. For example, the present study showed a less robust effect of an extinction cue than was observed in prior research that used extinction alone and showed a more robust effect of context fading than was shown in prior research that used extinction alone. It is conceivable that the extent to which different renewal mitigation techniques are effective is somewhat dependent on the availability of an alternative response, correlating contingencies for each response, and the contexts in which such responses are learned, as suggested by prior research in which renewal was studied using DRA and its variations (e.g., Kimball et al., 2020; Saini et al., 2018) and conceptual analyses of renewal during DRA (e.g., Kimball & Kranak, 2022; Saini & Mitteer, 2020). The role of DRA in both producing renewal and influencing renewal mitigation strategies should be explored in future research and directly compared with extinction alone.

Relatedly, using DRA in studies of relapse better resembles the clinical applications commonly designed in applied settings (e.g., functional communication training). Therefore, in the present study, the DRA contingency not only simulated traditional applied interventions but also provided information on alternative response acquisition and how alternative responses are affected by different target response mitigation strategies. Although there was some variability across groups, on average the alternative response was acquired at comparable levels and maintained for all groups in Phase 2. This suggests that the addition of the extinction cue, context fading, or both did not meaningfully interfere with alternative response acquisition. Finally, alternative responding was not disrupted during Phase 3 and increased when the initial context was reintroduced (although these increases were not statistically significant). Taken together, changes to stimulus conditions to mitigate renewal did not affect the acquisition or maintenance of alternative responding across contexts during this study. However, this could have been affected by the experimental arrangement, which restricted responding to two operants. It is possible that the maintenance of alternative responses could be more successful in applied settings (e.g., Kelley et al., 2018; Saini et al., 2018).

In addition to using DRA to evaluate the effects of renewal mitigation procedures, this study differed from other human‐laboratory tasks in several other ways. First, we used topographically distinct target and alternative responses, which is not common in renewal studies using DRA (e.g., Kimball et al., 2023). Second, this study included a consummatory response, which could increase the generality of the results to studies of renewal using food reinforcers in nonhuman animal experiments. Third, the experimenters sat in the room with the participant, which could have produced reactivity or sensitivity to the research environment. Fourth, participant inclusion did not require Phase 1 or Phase 2 response stability criteria, and as a result, experimental control was weak for several participants (e.g., see P19 and P22 in Appendix A). Therefore, the findings of this study should be viewed through the lens of the experimental modifications made during the study compared with those of other human‐laboratory studies.

Interestingly, this study did not experience any attrition, nor did it require any participants be excluded or removed. In contrast, Finch et al. (2022) reported that only 13 of 25 recruited participants met the inclusion criteria. This may be because the experiment in the current study was only 45 min long and was a continuous computer task, whereas the Finch et al. study included 1‐ or 3‐hr sessions. It may also be the case that participant demographics contributed to this difference in attrition. However, one limitation of this study is that we did not collect demographic information beyond asking participants their age and gender.

A final limitation is that the primary data analysis in this study relied on comparisons of single‐subject group means. Such an analysis (i.e., collapsing individual subject variability into group means) has inherent shortcomings including how individual differences contribute to the overall efficacy of renewal mitigation, obscuring underlying trends or patterns that may be important for understanding the conditions under which renewal mitigation strategies are or are not effective, and failing to acknowledge that renewal is an individual rather than group relapse phenomenon. As a result, future studies should explore the generality of these results when subjects are analyzed at the individual level.

In conclusion, the present study supports the notion that renewal is a robust phenomenon when left unmitigated and, in the context of relapse, can be problematic for the durability and maintenance of behavioral interventions. As a result, efforts have been made to further understand the phenomenon and to consider strategies for its prevention. The present study provides preliminary evidence that the use of context fading could effectively mitigate renewal. In contrast, the use of an extinction cue reduced the degree of renewal relative to the absence of a renewal mitigation strategy but may not be as effective at reducing renewal completely.

The maintenance of behavior change across contexts, including those in which there are preestablished reinforcement contingencies, is ubiquitous to learning. Such transitions could produce the reemergence of previously learned behavior due to contextual factors. In some cases, the renewal of such behavior could be undesirable. Future studies should continue to explore strategies for the mitigation of such events and to promote the generalization and durability of newly learned responding across contexts. Doing so may promote viable technologies for response maintenance.

AUTHOR CONTRIBUTIONS

Carlos Henrique Santos da Silva: Conceptualization, writing original draft, investigation, methodology, validation, visualization, software, formal analysis, data curation.

Valdeep Saini: Conceptualization, methodology, visualization, review and editing, project administration, supervision, resources.

CONFLICT OF INTEREST STATEMENT

On behalf of all authors, the corresponding author states there are no conflicts of interest to declare.

ETHICS APPROVAL

The research described herein received ethics clearance from Brock University's Research Ethics Board [FILE #22‐170].

Supporting information