Virtual Sprouts: Web-based Gardening Games to Teach Nutrition and Combat Obesity

A. Research Education Program Plan

A.1: Novel approaches to educate the community on NIH-funded research and health-related
issues, and prevent obesity in minority youth through meaningful play - Changing the field:
We propose to develop, disseminate and evaluate Virtual Sprouts: Web-Based Gardening Games
in a game- based environment that has the potential for broad dissemination. In the virtual
garden, children learn to select what to plant, plant their garden, watch it grow, tend,
water and weed, and harvest the crops. It has been argued convincingly that games and game
technologies enhance children's learning and achievement (Gee, J., 2003) through fun (Kim,
B., Park, H., & Baek, Y., 2009), meaningful play (Gee, J., 2003), enhanced involvement in
the subject matter (Kim, B., Park, H., & Baek, Y., 2009), increased motivation and more
independent engagement (Tüzün H, YIlmaz-Soylu M, Karakus T, Inal Y, KIzIlkaya G., 2009).
Therefore, a virtual gardening environment has great promise as a learning tool.
Furthermore, the realm of gardening provides rich and deep curriculum opportunities, while
at the same time increasing access to fresh, low-cost produce, improving nutrition, building
bridges between neighbors and creating stronger communities (Twiss, J., Dickinson, J., Duma,
S., Kleinman, T., Paulsen, H., & Rilveria, L., 2003). Because gardening covers a broad
spectrum of knowledge, from basic sciences through nutrition to human behavior, gardening
education offers an excellent opportunity to address Science, Technology, Engineering, and
Mathematics (STEM) education. Although access to fresh produce remains extremely poor across
great swaths of Los Angeles (Kipke, M.D., Iverson, E., Moore, D., et al., 2007), the city is
not a stranger to gardening efforts. For example, Los Angeles has been home to the Common
Ground gardening program since 1978. This program is dedicated to excellence in Master
Gardening, developing school gardens, and teaching families how to garden, grow their own
food and prepare it in a healthful manner. Virtual Sprouts Web-Based Gardening Games is
poised to have broad reach and high impact because a) internet technology is widely
accessible and extremely popular among youth and b) games and virtual environments enhance
learning and improve achievement.

B. Innovation

B.1: A Virtual guide in a virtual garden: Innovative approaches to learning, teaching and
changing behavior: When designed in accordance with proven instructional principles, the use
of games and game technologies have been shown to enhance children's learning and
achievement (Gee, J., 2003). Emerging literatures suggest that narratively rich, (Tanenbaum,
K., & Tanenbaum, J., 2009) socially interactive and developmentally appropriate (Barab, S.,
Thomas, M., Dodge, T., Carteaux, R., & Tuzun, H., 2005) games that incorporate interactive
storytelling and believable agents (Riedl, M., & Stern, A., 2006) have the potential to
strengthen learning and motivation. Research on effective communication shows that the
perceived characteristics of an agent or guide (i.e., gender, social status, believability,
integrity) will influence behavioral choices in real and virtual worlds (Frankenhuis, W.E.,
Dostsch, R., Karremans, J.C., & Wigboldus, D.I.H.J., 2010), and that children are quite
sensitive to the believability and status of the agent (Robinson, E.J., & Mitchell, P.,
1992). Diffusion of innovation delineates the importance of the characteristics of opinion
leaders to the speed of adaptation of new ideas (Rogers, E., 1995). Therefore using
state-of-the-art, award winning technologies available at USC'a Interactive Media Division
(IMD) of the School of Cinematic Arts (SCA) and the Institute for Creative Technologies
(ICT), we will develop a highly credible, personable and attractive pedagogical agent to
play the role of Virtual Guide through the Virtual Sprouts learning experience. The guide
will simultaneously assume the roles of coach, mentor, collaborator, and friend. This will
include orienting the player to the game, providing explanations of game events, making
suggestions about ideal actions to take, participating in the game activities, and conveying
elements of the underlying narrative.

C. Approach

C.1: Research Education Plan

C.1.a: Overview of the Plan: This project will proceed in three iterative phases:
Development, Dissemination and Evaluation. Development of a) curriculum, b) virtual garden
and c) pedagogical agent will proceed in parallel, and be integrated during testing and
evaluation. A participatory research group comprised of key community members, teachers,
clinicians, education experts, 8-11 year old youth and their parents will be formed to
provide iterative rounds of input and evaluation throughout the development of Virtual
Sprouts. This participatory research group will provide input through focus groups, beta
testing, idea-building and advisement throughout the development of Virtual Sprouts. We have
successfully used this model in previous health and intervention studies. The participatory
research group is composed of about 20 diverse members so as not to tax individual members
with study needs and to ensure diverse input. Sub-groups of the larger participatory
research group are formed for specific tasks, depending on research needs and group
interest. The Virtual Sprouts Game will be developed around a virtual pedagogical agent, who
will guide the player through the game using cognitive and affective learning principles.
The Virtual Environment will be a game system developed using deep narrative and dramatic
elements. Both the pedagogical agent and game environment will be extensively and
iteratively tested with members of our participatory research group during development. The
Virtual Sprouts game will then be introduced into three of the USC Family of Schools
Elementary schools, three of the CTSI clinics and the California Science Center (see letters
of support). Dissemination at the schools will use school internet capabilities.
Dissemination at the CTSI clinics and the Science Center will use Virtual Sprouts
interactive kiosks that will be placed by the study. Children will be able to access their
Virtual Sprouts accounts from any computer. Prior to dissemination and kiosk placement,
teachers, clinicians and Science Center staff will receive a series of workshops and
seminars on childhood obesity in general, and the Virtual Sprouts game in particular.
Teachers, Science Center staff and clinicians will also receive the Virtual Sprouts Game
Manual, which will be developed by the Virtual Sprouts team as an on-paper and virtual web-
based interactive module-by-module teacher's aide.

C.1.b: Curriculum Development: During the first year of this grant, we will work on
developing and focus group testing the nutrition and gardening curriculum. Curriculum will
be tested iterative through focus groups and beta testing with members of the participatory
research group. The gardening and nutrition program will not only teach children about the
health benefits of fruits and vegetables but children will learn additional skills in math,
science, and history. The program will include lessons relating to culture, including
growing crops that have cultural significance such as nopales, beans, corn and squash (the
latter three crops are historically referred to as "the three sisters"). Gardening provides
an ideal vehicle for teaching Science, Technology, Engineering, and Mathematics (STEM)
topics such as how plants grow and how gardens can be managed in an environmentally sound
manner (science including biology, earth sciences, agriculture, and the study of weather and
climate), how to make good nutrition choices with the produce they harvest (nutrition
sciences, health behavior). Furthermore, thinking about planting, harvesting, and climate
involves a great deal of math and engineering. The web-based interactive nature of the
Virtual Sprouts game will introduce children to new uses for advanced technologies.

C.1.c: Game, Pedagogical Agent and Virtual Environment Integration: The game will contain
the Pedagogical Agent and Virtual Environment and will provide the interaction framework
through play mechanics. Rather than a serial task-driven game, we envision a persistent
virtual environment in which the player will explore Virtual Sprouts and be challenged and
nurtured by the agent in joint social play activities. Play experience will be episodic,
requiring a few minutes of play per day over several weeks. Certain interactions in the
virtual environment will be joint activities (such as watering or harvesting vegetables),
and challenges issued by the agent will involve the agent personally. For example, the agent
may challenge the learner to plant, grow, and harvest ingredients for a certain recipe, then
prepare it. Based on the quality of the learners work, the agent may like or dislike the
resulting meal. Use of this kind of intrinsic feedback blurs the line between learning and
gameplay, and makes the agent an integral and thus indispensable part of the experience.
During the first year, we will determine the breadth (variety) and depth (challenge) of the
game experience in relation to gardening and cooking tasks appropriate for the target
population through paper prototyping and iterative play testing (Fullerton T, Swain C,
Hoffman S., 2008). Iterative Play testing Methods. Iterative play testing process, involves
the following steps:

1. Self-testing of paper prototypes with project team in order to come up with core
mechanics. This will occur during regularly scheduled face-to-face meetings and
prototyped concepts will concentrate on subject- specific behavioral role-play and
technical feasibility;

2. Testing of digital prototypes with colleagues, friends or family in order to gain
third-party perspective. This will occur in phases based on the software development
progress and will involve direct observation through informal usability studies.

3. Testing of partially complete digital game prototype with random strangers (N=15) who
partially fit the target population criteria (recruited through advertising) in order
to gain objectivity. Testers will play for a few hours, report bugs and answer custom
pre/post-questionnaires on play mechanics and usability.

4. Testing completed digital game prototype with sample (N=20) of target audience (obese
youth and their families). Testers will be required to use the game for several days at
a time, report bugs and answer custom questionnaires about game mechanics, reward
mechanisms and other game-specific measures.

C.1.d: Pedagogical Agent Components: The pedagogical agent will consist of several key
components that, when combined, provide a knowledgeable, friendly, and entertaining
companion for learning in the Virtual Sprouts environment through cognitive-affective
approaches. These components include:

1. Expert knowledge: The character will possess of cognitive model of effective gardening
and cooking that will allow it to support the activities of the Virtual Sprouts game.
This will enable the agent to explain how to create and maintain a healthy garden,
harvest and cook items from the garden, and provide explanations of the health benefits
of eating organically prepared foods. The expert model will also form the basis for
assessment of learner activities to give them credit when they take a correct action
(e.g., water at the correct time and amount) and to know when they've made a mistake
(e.g., planted two seeds too close to one another). These assessments will feed into
the pedagogical decision system.

2. Pedagogical knowledge: The provision of guidance requires a delicate balance of
providing timely (and wanted) support while allowing the learner to feel a sense of
freedom and remain stay in control to feel a sense of ownership of the garden. Care
must therefore be taken in decisions of how and when to provide guidance. Too little
guidance can lead to frustration for many users (i.e., the agent should be helpful),
while too much may threaten users' feelings of autonomy and freedom in the game
environment (i.e., the agent should not be perceived as controlling). Thus, through a
combination of unsolicited guidance triggered by evidence of confusion and/or lack of
domain knowledge and solicited guidance, when the learner requests help, the agent will
seek to be a productive helper while ensuring that the learner is acquiring the needed
knowledge and skills to succeed at Virtual Sprouts, and in their regular lives as the
relate to eating and exercise.

3. Emotion/affect knowledge: In addition to supporting game play and providing
explanations for game events, the agent will also support the learner's affective
state. Cultivating and nurturing interest, motivation, and positive feelings about the
domain are critical in the pursuit of behavior change, and expressions of empathy from
the agent can play into the learner's willingness to remain engaged (McQuiggan, S.W.,
Rowe, J.P., & Lester, J.C., 2008). These intervention techniques will be implemented in
concert with the game- and cognitive-focused tactics described above. For example, a
user who is just beginning and has completed his or her first vegetable will receive
high praise, which will diminish with time for that specific accomplishment. Similarly,
for learners who express a negative perception of gardening and vegetables, the agent
will work harder to motivate and entertain.

4. Animation and sound: As discussed above, the appearance of the agent will have an
impact on the learner's affective state, both during and after the interaction. A
well-known and respected character, such as one modeled after the First Lady (if
possible), will increase the chances of establishing familiarity and credibility from
the outset. Use of appealing animations and joint activity, such as watering the garden
simultaneously, will bolster feelings of co-presence and teamwork. The character's
voice, which will be delivered along with subtitles for English as a second language
learners, will be chosen based on the preferences of users from the target audience in
early testing of the system.

C.1.e: Game, Pedagogical Agent and Virtual Environment Development: The phases of
development are outlined below:

1. Domain knowledge and learning objective analysis: During year 1, we will gather and
formalize the domain knowledge in a way that can be represented in the game for use by
the agent. This will feed into the expert modeling and provide the overarching
structure for tracking the learner's understanding of the domain. We will also play
test new game types and existing games for cooking and gardening through one-on-one
observational studies and interviews to determine fun and engaging play mechanics and
design principles using GAP Methodology (Desurvire, H., & Wiberg, C., 2008).

2. Expert modeling and feedback: Using the identified learning objectives and game
mechanics defined during year 1, in year 2 we will build computational models of the
skills needed to succeed at the Virtual Sprouts game. These models will mirror skills
needed to tend to an actual garden and cook actual food. As discussed, these models
will play a key role in building the assessment and pedagogical models to drive the
agent's interactions with the learner. Feedback content will be authored to correspond
to knowledge components represented in the expert model.

3. Pedagogical agent design & prototyping: In spiral development with the expert model,
the pedagogical and affect decision modules will be developed. Functionality for
solicited and unsolicited feedback will be implemented. By the end of year 2, the agent
will be providing support for learning and affect.

4. Agent and game integration: In this stage, the agent will be fully integrated with the
Virtual Sprouts practice environment (during year 3). This will include completing
animations that involve the agent working in the virtual world, as well as resolving
any underlying discrepancies between the game design and the models driving the agent's
behaviors. Play testing of game while agent is development will occur through role-play
using real people, a technique commonly used at USC's Game Innovation Usability Lab.
For example, early design phases of the Wellness Partners study, play testing of the
game system (which was meant to be operated by a virtual character that did not yet
exist) was done through email and text messaging via a human in real-time. This was
highly engaging and captured positive and negative affective states between agent and
player.

5. Agent visual and content revisions: The agent and underlying representations will be
modified and extended based on user and subject-matter expert feedback. By the end of
year 3, the agent will be ready for full pilot testing and summative evaluation

C.1.f: Dissemination to USC Family of Schools, CTSI/CTSA community clinics and California
Science Center: In years 4 and 5, Virtual Sprouts will be rolled out to 7 elementary schools
that are members of the USC Family of Schools, 3 CTSI/CTSA clinics and the California
Science Center. These activities are extensively detailed below in sections C.1.h:
Partnerships and Collaborations, C.1.j.3. Program Faculty/ Staff (under Consultation and
Dissemination Experts), C.2. Evaluation Plan, and C.3. Dissemination Plan. Briefly, we will
conduct a Randomized Clinical Trial (RCT) for the school-based implementation of the game.
In our dissemination venues (science center, museum and community clinics) no control
conditions will be applied. For these venues, descriptive and regression approaches will be
used in the venues to test intervention and dissemination evaluation. A major element of
implementation and dissemination will be the Virtual Sprouts Website, which will be
developed by Co-I Marientina Gotsis and maintained by Co-PI Spruijt- Metz's team. Each
participant (including children, teachers, parents and other interested community members)
will join the Virtual Sprouts community and study via the website. Consent and assent will
be available virtually. Each player will have their own account, so that they can play from
any computer with a web interface, or from a mobile phone with web access. All evaluative
surveys and logs will be available and administered through the Virtual Sprouts website.
Paper and pencil versions will be made available for the school environment as requested.
All participants will receive a Personal Sprout Identifier (PSI). Their PSI will be
confidential and they will use their PSI to log in to the Virtual Sprouts Website, continue
game play, receive rewards, and complete evaluations.

C.1.g: Teacher Professional Development: The vehicle for training teachers to use the
Virtual Sprout game is a hybrid approach and includes a Summer Teacher Academy (STA) serving
as a kick off program with face to case and USC Distance Education Network (DEN) mediated,
school site follow-up professional development in elementary school classrooms using a
lesson study approach. Teacher academies have been recognized in the teacher education
literature as a powerful and impactful way of comprehensively training and supporting
teachers (Crowther, D.T., Vilá J.S., Fathman A.K., 2006).

The Summer Teacher Academy: A summer teacher academy (STA) will occur in August each
implementation year immediately after year one of the program's development phase and in
years 2-5. Through the STA, teachers will be trained annually. The academies will be three
days long at USC's campus as each of the target schools are in relative close proximity with
USC. Mornings in the academy will be spent learning the nutrition conceptual knowledge,
exploring the application and practicing using a guided experiential learning approach to
facilitating the game use. The content for the training will be aligned with grades 6-12
California health content standards focused on nutrition and related health. Lesson study,
beginning with observational demos and progressing to teacher created role-play will be of
focus of the professional development. The teacher participants will spend a portion of the
morning observing and practicing Virtual Sprouts game, discussing the application and use of
the game in their classrooms. As academic year follow-up, teacher participants will practice
using the Virtual Sprouts game application and engage in peer review of game facilitation
with fellow teachers (lesson study) in tandem with the USC teacher education experts in
their classrooms.

Stigler's Lesson Study Model of Teacher Development Model: Throughout the Project's teacher
training, participant teachers will use a research-based lesson study approach to build and
study implementation of the nutrition and facilitation of Virtual Sprouts intervention with
their students. Lesson study, according to Stigler (McQuiggan, S.W., Rowe, J.P., Lester,
J.C., 2008), refers to a process whereby teachers closely examine their lesson facilitation
with a focus on addressing student need via data-driven decision-making. In the lesson study
model, teachers learn together. Participants plan, observe, and refine instruction and
learning facilitation (Desurvire, H., & Wiberg, C., 2008). Lesson study compels teachers to
examine their own practice in depth in the context of student learning, connects teachers
with their students and their professional community, and inspires them to continue to
improve. This model of teacher professional development has been applied widely and
successfully in Japan and has recently been initiated by teachers across the US as a
scientifically-based best practice. For the purpose of the summer academy, participant
teachers will "study" videotaped lesson exemplars using the facilitations cycle with the
Virtual Sprouts game.

Teacher Academy Follow-up - The Professional Learning Communities (PLCs): Follow-up meetings
for teachers with the teacher training lead (co-PI Ragusa) will occur during the ensuing
academic years of the Virtual Sprouts. Each participant teacher will meet quarterly with
school site administrative members, and the project PIs. Additionally, participant teachers
will enroll in an online computer mediated virtual classroom community of practice using
USC's DEN courseware platform aligned to the teacher resource web portal to share ideas with
teacher participant peers and the PIs. Teachers will engage in "virtual" lesson study
(McQuiggan, S.W., & Rowe, J.P., 2008) and post their Virtual Sprouts activities and
implementation plan for the intervention specific to their classroom. Continued lesson study
will continue via this web portal, as teacher participants will be required to "attend" on
line at least four times monthly. This allows for maximum PLC participation with minimum
teacher burden and enables a collaborative focus in the PLCs. Additionally, this allows for
the PLCs to continue across schools beyond the SEPA funding period as a sustainability
strategy. The PLC structure is designed to create a sustained community of practice with all
participant teachers (Desurvire, H., & Wiberg, C., 2008).

Inclusion Criteria:

- Child in 3rd, 4th, or 5th grade

Exclusion Criteria:

- Child in grades earlier than 3rd or later than 5th