Discussion
Research consistently indicates that active learning strategies outperform informational strategies during safety-response training (e.g., Gatheridge et al., 2004). BST has also been demonstrated many times in the abduction prevention literature to be an effective method for training safety responses to lures presented in person (e.g., Gunby et al., 2010). Therefore, because BST was used successfully in previous abduction-prevention studies, we hypothesized BST would likewise be an effective methodology when employed to teach youth with ASD safety responses to lures encountered in a virtual context (i.e., while playing video games online). During BST, the participant’s safety scores increased from those observed during baseline and the participant reached the performance criterion on the 10th trial. These results support this hypothesis and are consistent with the findings of past abduction-prevention research.
Although the participant reached the performance criterion during training, follow-up assessments revealed his responding did not maintain. These results are similar to other BST studies where some participants demonstrated skill acquisition during roleplay assessments, but their performance did not maintain during follow-up assessments until an IST component was included (e.g., Himle et al., 2004).
For example, Johnson et al. (2006) compared the effectiveness of combining BST with an IST component, to BST alone, when teaching preschool children abductionprevention skills. The results of the study showed the BST with IST group earned higher safety scores during 2-week, 1-month, and 3-month follow-up assessments than the scores of the BST alone group. Based on the findings of Johnson et al. (2006), IST was provided to the participant after he did not perform the four-step safety response sequence correctly during the fourth posttest probe of the posttest assessment. After the inclusion of IST, the participant successfully performed the four-step safety response sequence correctly. These results support combining BST with IST for participants whose responding does not maintain with BST alone.
However, in light of these results, there were limitations to this study. Most notably, experimental control was not sufficiently demonstrated because only one participant was recruited for the study. The experimental design that was proposed prior to the onset of this study was a multiple-baseline design evaluated across participants; however, recruitment for the study was hindered by COVID-19 precautions. For example, in order to reduce exposure, the first author was confined to interacting with a small number of children at the behavioranalytic practicum site where he was employed at the time of the study. Only one child met the inclusionary criteria for the study from the small group of children the first author was permitted to have contact with. Similarly, access to other clinics in the same corporate network were restricted also due to COVID-19 preventative measures. Thus, experimental control was not adequately demonstrated as BST was not evaluated across multiple participants. Yet, in lieu of this significant limitation, the present study provides preliminary evidence that safety behaviors for online interactions can be trained for youth with ASD using BST. However, questions remain regarding the persistence of the new learning. Future research should evaluate the effectiveness of implementing BST to train online-safety responses to youths with ASD with multiple subjects recruited for the study.
Another limitation of the study was the clinic where the study was conducted had a firewall which blocked online access to Plants vs Zombies: Battle for Neighborville. The first author connected the PC computer each session to a personal hotspot that was broadcast from his mobile device. As a result, the internet connection was poor during some trials and the voice chat did not function (i.e., the confederate was unable to make the IRPI). In response to the absence of in-game communication during an affected trial, the researcher covertly called the confederate using another cellphone, enabled speakerphone, and placed the mobile device out of the participant’s view prior to conducting the IRPI probe. Using the cellphone to simulate the voice-chat communications is another limitation of the study. It is unknown whether that participant’s performance was influenced during trials that the mock threat was presented to the participant using the cellphone instead of using voice-chat.
Moreover, some assessment and training sessions were interrupted when the connection to the Origin server was lost. Reconnecting to the server resulted in lengthy delays (e.g., 5-10 min). In order to avoid delays associated with disconnecting and reconnecting to the Origin server, the third step in the safety response sequence was adjusted from “exiting” the game, to “pausing” the game at the onset of the study. These technical issues may have threatened the procedural integrity of the study. Future researchers should conduct IRPI probes and training sessions in a location with a strong internet connection and without a firewall that blocks access to video game websites.
The study was also limited because it was relatively labor intensive to implement. Two individuals were required to train the safety responses to one child. Future researchers should consider conducting the training in a group format with individualized probes to make the training more cost effective.
In addition, only the participant’s responses to the presentation of the mock threat were assessed during the study. Future research may consider evaluating participant’s responses both to nonthreat conversation overtures, as well as responses to the mock threat.
Lastly, the study was limited because the researchers were unable to prevent unknown individuals from interacting with the participant during assessments and training. The video game Plants vs Zombies: Battle for Neighborville was selected for this study because it is a multiplayer-online video game that was age-appropriate, allowed for voice chat and in-game messaging, and the participant expressed his preference for the game prior to the study. In the multiplayer environment of Plants vs Zombies: Battle for Neighborville, when the voice-chat function is enabled, all vocal communication from players in the multiplayer environment who also have voice chat enabled are audibly broadcasted through the computer’s speakers. Textual communication between players is also displayed on the computer screen. The game settings only allow for a maximum of four players to be muted at one time. Therefore, players extraneous to the study were able to contact the participant during trials. Following an occurrence of extraneous-player communication, the first author interrupted the participant’s gameplay and muted the extraneous player. Subsequently, some trials were interrupted when this occurred. There were also trials when more than four players had their voice chat enabled and some communication with the participant could not be controlled (i.e., muted) by the first trainer.
For instance, during numerous trials there was one specific player (unknown to the participant or experimenter) that frequently contacted the participant using voice chat. In addition, he sent the participant invitations to join his private group throughout IRPI probes and training. Following these engagements, the game was paused, the player was muted, and any invitation to private games were also declined. Future research should select games to use in the study that give the experimenters better control over which communications are broadcast to participants. Although it was a limitation, it should be noted that communication among unknown players was the rule and not the exception during this study which highlights the need for this specific type of safety training.
In light of these limitations, the present study extends the BST literature because it is the first study to teach safety responses for threats encountered while playing online video games. This is an important line of research because video games are massively popular and frequently played by youth with ASD. These games can present a serious risk to this vulnerable population because they provide access to the internet and youth with ASD, due to the social deficits associated with the disorder, are typically more susceptible to dangers (e.g., solicitations from internet predators) that could be encountered online.
Finally, this study also broadens the BST research because it represents the first attempt to utilize BST to address online safety. There are numerous safety threats that reach beyond the scope of this study (e.g., cyberbullying, digital literacy, sextortion). Technology with access to the internet (e.g., smartphones) is ubiquitous in modern society, especially with teens. Thus, online safety certainly warrants attention from the field of behavior analysis. The results of this study support future research to evaluate the efficacy of BST to teach online-safety skills for a host of malevolent threats present on the internet.