VidCoach: A Mobile Video Modeling System for Youth with Special Needs

Abstract Autism Spectrum Disorder (ASD) is often characterized by a deficit in social skills, including the ability to interview successfully for jobs, a key hurdle to be overcome for independent living. One way that individuals with ASD can learn and retain valuable life skills is through the use of video modeling. In this paper, we describe the design of VidCoach, a mobile application built to support both peer modeling and self-modeling for individuals with ASD. VidCoach can aid adolescents in work transition programs with learning and retaining job interview skills, and we present a scenario that highlights how a use might interact with VidCoach in this particular context.

Keywords Autism Spectrum Disorder, education, assistive technology, children, teens, video modeling

ACM Classification Keywords H.5.2. Information Interfaces and Presentation (e.g., HCI), User Interfaces; K.3 Computers and Education

Introduction and Background According to Social Learning Theory, the acquisition of new skills through observational learning and modeling can account for natural acquisitions in behavior [1]. For individuals with ASD, research has shown that this theory, applied through video modeling, can assist in

Permission to make digital or hard copies of all or part of this work for

personal or classroom use is granted without fee provided that copies are

not made or distributed for profit or commercial advantage and that

copies bear this notice and the full citation on the first page. To copy

otherwise, or republish, to post on servers or to redistribute to lists,

requires prior specific permission and/or a fee.

CHI’13, April 27 – May 2, 2013, Paris, France.

Copyright 2012 ACM 978-1-XXXX-XXXX-X/XX/XX...$10.00.

Rachel Rose Ulgado University of California Irvine Irvine, CA 92612 rulgado@uci.edu Katherine Nguyen University of California Irvine Irvine, CA 92612 katheran@uci.edu Van Erick Custodio University of California Irvine Irvine, CA 92612 vcustodi@uci.edu

Aaron Waterhouse University of California Irvine Irvine, CA 92612 waterhoa@uci.edu Rachel Weiner University of California Irvine Irvine, CA 92612 rweiner@uci.edu Gillian R. Hayes University of California Irvine Irvine, CA 92612 hayesg@uci.edu

learning and retaining positive behaviors. Advantages in video modeling include the ability to give positive feedback and present concepts and instruction in engaging ways [2]. Video modeling is effective, at least in part, because requiring someone to look at a small spatial area and listen to necessary language forces them to direct their focus to relevant stimuli [1]. However, to date, video modeling has mostly been limited to classroom applications.

Video modeling typically involves watching recorded videos of others modeling positive behavior. Video modeling can be split up into two types, peer modeling and self-modeling. Peer modeling focuses on imitation from models similar to one’s self (in physical characteristics, age, group, ethnicity, etc.). Self-modeling allows people to see themselves successfully completing a task. Self-modeling offers opportunities for self-evaluation and visual progression of personal growth. Current methods of video modeling typically involve caregivers presenting workshops and displaying the videos through television sets. These workshops are removed in both time and place from the actual activity being modeled.

The ubiquitous nature of mobile technologies provides an opportunity to make video modeling more connected to the activities users are trying to accomplish. Thus, in this work, we designed and developed a mobile iOS application, VidCoach, which allows individuals to watch peer model videos as well as record their own videos for self-evaluation and improvement. Designing VidCoach We used an iterative process for designing and developing VidCoach. We first conducted observations as part of a three-year collaborative action research

project [3] with six school districts in the Southern California area and two agencies from Orange County, California. This program includes a series of workshops aimed at teaching students and staff in a work transition program to use mobile off the shelf devices to improve student independent living skills. Each year 61 students (approximately 3/4 with ASD) ranging in age from 12 to 22 participated in the program alongside staff members from across the districts and agencies [4]. The students involved in the program had varying disabilities, including ASD.

We then conducted semi-structured interviews with eight professionals and caregivers of individuals with ASD who specialize in supporting students with ASD in overcoming their barriers to employment. The interviews were conducted in person at their workplace and lasted approximately one hour each. The interviewers took notes and audio recordings of the interviews. During these interviews, we asked if they were familiar or had previously used video modeling as a tool in their work. We also gathered their suggestions and recommendations for using video modeling as an assistive tool for individuals with ASD.

Our empirical work highlighted some of the biggest challenges to using and creating technologies to support ASD, such as the high amount of variation in abilities that must be supported. In particular, interviewees stressed the need for a portable way to conduct video modeling sessions, noting that current methods were far removed from daily practices as the literature also indicated. Thus, we were particularly focused in this work on the creation of a robust, low-cost, mobile solution. Likewise, they noted the importance of variation in representations of particular

Figure 1: A user interacting with VidCoach by self-recording.

Figure 2: Menu structure for VidCoach.

skills to prevent students from merely memorizing a script, an issue we observed repeatedly during the workshops. This finding implies that mobile tools must balance variation and extensive content with issues around available resources on a mobile device.

The importance of self-modeling, in which people watch themselves successfully performing the desired actions or behaviors, emerged as another key consideration. Although the research literature describes the efficacy of self-modeling [5], in practice, staff members and teachers struggle to implement these ideas. Challenges include everything from finding a domain problem that is amenable to self-modeling to understanding how to record and edit the videos themselves. We chose interviewing as a domain problem in part because of the challenges students we observed have with this skill but also because it is particularly appropriate for self-modeling with the videos largely being “head shots” with clearly defined communication exchanges. These videos can easily be recorded in snippets using the forward-facing camera on a mobile device, such as an iPod or smartphone, with prompting to indicate when a new snippet should be recorded and what it should include.

Finally, the importance of tracking progress and adjusting the technology in light of progress was discussed frequently by interviewees and mentioned repeatedly in the literature [6]. These findings indicate that any mobile system for video modeling should have the ability to log and analyze progress as well as to adjust the amount of prompting and other supports provided by the application depending on skill level.

Our prototype application, VidCoach, supports peer

modeling by allows users to watch videos of job interviews, which involve video snippets of interviewers asking questions and an interviewee responding to those questions. After an interviewer asks a question, a prompt pops up, telling the user to complete a task. Following this prompt, the interviewee answers the question. Afterward, a second type of prompt appears, asking the user to answer either a true/false or multiple-choice question to reflect on the task they just watched. As users improve their capabilities, these prompts can be turned off.

Once a user feels comfortable enough to practice a job interview video, they are able to do so with their front-facing camera. Instead of the interviewee video snippets, VidCoach prompts the user to record snippets of their own responses to the interviewer’s questions. They are able to re-record their responses as part of their practice until they feel satisfied with their work.

After a user has completed recording his or her responses for one of the job interview videos, they are able to watch their entire interview in sequence, or can select an individual response to review. Users are able to see themselves responding successfully to the interview questions and can monitor their own progress.

Conclusion and Future Work VidCoach extends the benefits of video modeling to aid individuals in the learning and retention of valuable life skills to a mobile space in which the video modeling practices are tightly tied to the actual activities. In our future work, we will conduct a study to examine the feasibility of use of VidCoach in support of individuals in work transition programs preparing for job interviews. Likewise, we will measure efficacy in terms of how

Figure 3: An video snippet of a peer model – in this case, an interviewee.

many missteps students make in realistic job interviews before and after using our tool over an extended period of time. Finally, we will check for generalization of the skills by comparing performance in job interviews to the script used in the model videos and prompts embedded within VidCoach.

VidCoach currently supports interview skills, but its extensible architecture allows for the inclusion of a variety of other videos, particularly those involving social skills and communication. Thus, in the future, we hope to include a wider variety of videos and possibly develop a mechanism by which end users can upload new videos to be shared with the network of VidCoach users.

Finally, it is clear from our empirical work that additional support for monitoring progress, particularly at a distance would be helpful. Thus, we plan to develop features for parents, teachers, and other people who act in support of the students with ASD. For example, we would like to implement graphs demonstrating automatically detected social missteps or more simply use. Likewise, the ability for caregivers to view and comment on videos at a distance can support greater independence in use without sacrificing guidance. Finally, allowing peers to view and comment on each other’s videos could replicate at a distance the commonly used practice in schools of using social skills groups to critique and improve upon individual performance.

Acknowledgements Special thanks to the Undergraduate Research Opportunities Program (UROP) at the University of California, Irvine for partially funding this work as well

as to the NSF Award #0846063. Also, many thanks to Linda O’Neal, Steve Hosaflook, Kerri McCanna, and to the STAR group and everyone who volunteered to be models in our videos.

References

[1] Shipley-Benamou, R., Lutzker, J., and Taubman, M. “Teaching Daily Living Skills to Children with Autism Through Instructional Video Modeling.” Journal of Positive Behavior Interventions, Volume 4, Number 3, Summer 2002, pp. 166-177. [2] GoldSmith, T. and LeBlanc, L. “Use of Technology in Interventions for Children with Autism.” Journal of Early and Intensive Behavior Intervention, Volume 1, Number 2, 2004.

[3] Hayes, G.R. 2011. “The Relationship of Action Research to Human-Computer Interaction. ACM Transactions on Computer-Human Interaction.” 2011. 18 (3): 15. [4] Hayes, G.R., Yeganyan, M.T., Brubaker, J.R., O’Neal, L., and Hosaflook, S.W. 2012. “Using Mobile Technologies to Support Students in Work Transition Programs.” Twenty-First Century Skills for Students with Autism. Katharina Boser and Matthew Goodwin, Eds. Brooke’s Publishing. In Press. [5] Bellini S. and Akullian, J. 2007. “A Meta-Analysis of Video Modeling and Video Self-Modeling Interventions for Children and Adolescents With Autism Spectrum Disorders.” Council for Exceptional Children, Vol. 73, No. 3, pp 264-287, 2007.

[6] Haring, T., Kennedy, C., Adams, M., and Pitts-Conway, V. “Teaching Generalization of Purchasing Skills Across Community Settings to Autistic Youth Using Video Modeling.” 1987. Journal of Applied Behavior Analysis, 20, 89-96

Figure 4: An example of a prompt that helps users reflect on the task they just watched.