Feltz et al.

BMC Psychology (2016) 4:54

DOI 10.1186/s40359-016-0165-9

STUDY PROTOCOL Open Access

Simulated Partners and Collaborative

Exercise (SPACE) to boost motivation for
astronauts: study protocol
Deborah L. Feltz1*, Lori Ploutz-Snyder4, Brian Winn3, Norbert L. Kerr2, James M. Pivarnik1, Alison Ede5,
Christopher Hill1, Stephen Samendinger1 and William Jeffery3

Abstract
Background: Astronauts may have difficulty adhering to exercise regimens at vigorous intensity levels during long
space missions. Vigorous exercise is important for aerobic and musculoskeletal health during space missions and
afterwards. A key impediment to maintaining vigorous exercise is motivation. Finding ways to motivate astronauts
to exercise at levels necessary to mitigate reductions in musculoskeletal health and aerobic capacity have not been
explored. The focus of Simulated Partners and Collaborative Exercise (SPACE) is to use recently documented
motivation gains in task groups to heighten the exercise experience for participants, similar in age and fitness to
astronauts, for vigorous exercise over a 6-month exercise regimen. A secondary focus is to determine the most
effective features in simulated exercise partners for enhancing enjoyment, self-efficacy, and social connectedness.
The aims of the project are to (1) Create software-generated (SG) exercise partners and interface software with a
cycle ergometer; (2) Pilot test design features of SG partners within a video exercise game (exergame), and (3) Test
whether exercising with an SG partner over 24-week time period, compared to exercising alone, leads to greater
work effort, aerobic capacity, muscle strength, exercise adherence, and enhanced psychological parameters.
Methods/Design: This study was approved by the Institutional Review Board (IRB). Chronic exercisers, between the
ages 30 and 62, were asked to exercise on a cycle ergometer 6 days per week for 24 weeks using a routine
consisting of alternating between moderate-intensity continuous and high-intensity interval sessions. Participants
were assigned to one of three conditions: no partner (control), always faster SG partner, or SG partner who was not
always faster. Participants were told they could vary cycle ergometer output to increase or decrease intensity during
the sessions. Mean change in cycle ergometer power (watts) from the initial continuous and 4 min. interval sessions
was the primary dependent variable reflecting work effort. Measures of physiological, strength, and psychological
parameters were also taken.
Discussion: This paper describes the rationale, development, and methods of the SPACE exergame. We believe this
will be a viable intervention that can be disseminated for astronaut use and adapted for use by other populations.
Keywords: Behavioral health, Exercise, Exergames, Fitness, Köhler effect, Motivation, Relational agent, Software-
generated partner, Virtual reality

* Correspondence: [email protected]
1
Department of Kinesiology, Michigan State University, East Lansing, MI
48824, USA
Full list of author information is available at the end of the article

© The Author(s). 2016 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Feltz et al. BMC Psychology (2016) 4:54 Page 2 of 11

Background software agents (also referred to as relational or social

Astronauts need to adhere to high intensity exercise regi- agents) and apply social rules much as if they were hu-
mens to mitigate reductions in muscle strength and en- man. There is a strong research base suggesting humans
durance, bone density, and reduced aerobic capacity that can establish significant social relationships (e.g., keep
occur during long space missions. Exercise is also consid- promises and perceive virtual characters as teammates)
ered a key psychological countermeasure to risks of with SG partners [5–8].
adverse behavioral health although much less is known Traditional group or partner exercise leads to higher
about the dose-response relationship between exercise in- adherence than individualized exercise programs [9, 10],
tensity and behavioral outcomes. While most astronauts but structured group exercise is not possible for astro-
are able to sustain high intensity exercise programs over a nauts during space missions due to limited space and
4–6 month period (the average International Space exercise equipment. In addition, exercising in pairs may
Station mission duration), there is concern that for longer be difficult to coordinate. Exercising with an SG partner
duration missions, such as Mars, a key impediment to offers several advantages (e.g., availability, flexibility, au-
maintaining intense exercise levels is motivation. Identify- tonomy) over a live human partner. An SG partner’s
ing motivational strategies and technologies to support abilities can be adjusted automatically over time to attain
high intensity exercise over long durations has not been a level that may be the most motivating to the player,
explored. Exercise video games (exergames) have been thereby hypothetically, keeping the player engaged and
marketed as a way to increase motivation and enjoyment active. Additionally, exercising with an SG partner has
of exercise by being entertaining, engaging and providing the potential to make workout sessions more enjoyable,
a means to interact with other players. Although many improve self-efficacy regarding physical performance
exergames involve competition among players, there has capability and adherence to the regimen, and create a
been little attempt to analyze what game features and sense of social connectedness with the virtual character.
interpersonal interactions would best motivate users to The effects of an SG partner may be even stronger when
continue exercising with these games. used in socially isolated environments where there is lit-
National Aeronautics and Space Administration’s tle human interaction.
(NASA) research is evaluating a new high intensity inte- Active video games (i.e., exergames) have become in-
grated resistance and aerobic training program (SPRINT) creasingly popular and have been marketed as a fun way
during 6 months of spaceflight on the International Space to increase people’s motivation to exercise [11]. Several
Station (ISS) (ongoing) and during 70 days of bed rest studies have found that people are motivated to exercise
(which simulates a reduced gravity environment) [1]. Pre- with active games that are entertaining, engaging, and
liminary evidence from bed rest research studies suggests interactive [12]. However, even exergames can become
that a vigorous intensity exercise program during bed rest boring within a short period of time if played in isolation
is very effective. However, bed rest studies involve [13]. Few exergames take advantage of the potential of
SPRINT training for ~100 days during bed rest and the group dynamics to motivate physically active play, and
ISS study requires training for ~180 days. Eventual space there has been little attempt to analyze what interper-
exploration missions will require compliance with an exer- sonal interactions would best motivate people to use and
cise program for 2–3 years. Motivation and adherence to continue exercising with these games [11].
high intensity exercise, coupled with a socially isolating Recent research has shown that an SG partner, who
environment with atypical access to social support may was moderately more capable than the participant in an
compromise compliance, especially if astronauts’ regimens exergame, increased the player’s physical activity persist-
become monotonous. Exercise programs that enhance en- ence more than playing the game alone [6, 7]. This re-
joyment, self-efficacy, and a sense of social connectedness search is based on group motivation dynamics principles
may mitigate decrements in mood and feelings of social that stress upward social comparison and a sense of in-
isolation [2, 3]. dispensability of one’s efforts to their more capable part-
Group dynamics, using social psychological mecha- ner under conjunctive task conditions [14]. Under such
nisms such as social comparison and indispensability to conditions, the dyad team can persist no longer than its
group achievement, may be a useful means to address weaker partner–when the weaker member stops, it was
lack of motivation (i.e., the level of effort) for vigorous impossible for the stronger partner to continue. Thus,
physical exercise [4]. Our research is designed to deter- motivation is likely to be enhanced when one sees his/
mine whether astronauts’ motivation to exercise at in- her efforts as being highly instrumental in achieving
tense levels repeatedly over long durations can be team success [14]. However, to date, the motivating ben-
improved using a virtual, software-generated (SG) part- efits of an SG partner have not been explored with phys-
ner–one that is anthropomorphic but clearly artificial ical exertion tasks over an extended period of time and
and synthetic. People can respond socially to computers/ at high exercise intensities required of astronauts.
Feltz et al. BMC Psychology (2016) 4:54 Page 3 of 11

Further, whether the motivating benefits of exercising greater workout effort, physiological parameters (viz.,
with an SG partner who is continually superior will at- aerobic capacity, ventilatory threshold, and musculoskel-
tenuate over long-term intense exercise has not been ex- etal fitness), self-efficacy, enjoyment, interest in continu-
plored. This gap in the literature is important because ing the game, and perceptions of social connectedness
the full promise of an SG partner in exercise games for compared to exercising alone (See Fig. 1 for an overview
long-term exercise (e.g., simplicity of manipulating rela- of the different aims and flowchart through phases). The
tive ability over time; and avoiding scheduling conflicts following hypotheses were tested in Aim 3:
with a live partner) hinges on this question. Given previ-
ous research suggesting that humans will establish sig- H1: Exercising with a conjunctive SG partner over 24-
nificant social relationships with SG partners [5, 15], we week time period, compared to exercising alone, leads
sought to explore related questions in the context of ex- to greater workout effort and adherence.
ercising with an SG partner over a period of 24 weeks. H2: Exercising with a conjunctive SG partner over 24-
week time period, compared to exercising alone, leads
Specific aims and hypotheses to better aerobic capacity, higher ventilatory threshold,
The primary aim of Simulated Partners and Collabora- and greater thigh muscle strength.
tive Exercise (SPACE) was to determine whether re- H3: Exercising with a conjunctive SG partner over 24-
cently documented motivation gains in task groups week time period, compared to exercising alone, leads
(dyads in particular) can be harnessed to improve motiv- to greater enjoyment in the activity, self-efficacy, inter-
ation in an interactive exergame using SG partners. Aim est in continuing the exergame, and sense of social
1 involved the development of software to create SG ex- connectedness.
ercise partners and interface with cycle ergometers. The
SG partner features were tested with focus groups of as- Method
tronauts and NASA physical trainers and piloted with a Study design
convenience sample of physically active kinesiology stu- SPACE is a 6-month randomized control trial design. To
dents. After the SG exercise partners and games were better simulate actual astronauts, participants were
developed, the design features of the SG partners, the middle-aged adults who are competitive athletes or highly
exergame, and the questionnaires were pilot tested (Aim physically active exercisers, recruited from mid-Michigan.
2), using a short duration training study with master’s Participants were assigned to one of three conditions: no
athletes and fitness club members who are similar in age partner individual control, an always superior SG partner,
and fitness to experienced astronauts. Aim 3 tested the or an SG partner who was not always superior. Partici-
long-term (i.e., 24 weeks) effectiveness of the SG partner pants were asked to exercise on a cycle ergometer 6 days
in maintaining participants’ prescribed fitness goals, per week for 24 weeks using a routine consisting of

•Present features of SG
partner with astronauts
and trainers from JSC
(N = 7) and physically
Aim 1: Develop SG active adults over age
partner and 35 years (N = 4)
exergame •Pilot game design with
convenient sample of
kinesiology students (N
= 8)

•Pilot design features and exergame with

physical active middle-aged adults ages
Aim 2: Pilot design 30 - 62 years (N = 82)
features of SG •Length of participation: 6 days
partner and
exergame

Aim 3: Test
effects of SG •Design: 3 (Exercise condition:
Individual, Always Superior SG
partner over partner, Not always superior SG
longer-term using partner) x 2 (Exercise gender) x
best SG mode 4 (Training regimens:
Continuous, Interval sprint,
from Aim 2 Interval 2-min, Interval 4-min.) x
24 (Weeks); N =41

Fig. 1 Overview of the aims and flowchart through phases of SPACE study
Feltz et al. BMC Psychology (2016) 4:54 Page 4 of 11

alternating between days of moderate-intensity continu- exercise experience and availability for interviews. Astro-
ous (at or above 75% of their maximum heart rate) and naut trainers at the Johnson Space Center were recruited
three types of high-intensity interval sessions: (a) long, based on availability in their schedules. No incentives
4 × 4 min intervals at or above 90% HRmax with 3 min ac- were provided for participation. We prioritized for equal
tive rest (i.e., cycling at a recovery rate), (b) medium 6 × 2 numbers of men and women.
min intervals at 70, 80, 90, 100, 90, 80% of HRmax, re- A separate focus group of highly active middle-aged
spectively with 2 min active rest, and (c) short, 30 s max adults was recruited from the local community through
sprint intervals with 20 s active rest. The design consists the first author’s personal knowledge of local master’s
of a 3 (Exercise condition) × 2 (Participant gender) × 4 athletes. Participants were required to be between 35
(Training regimens: Continuous, Interval sprint, Interval and 60 years of age and exercise at least four times per
2-min, Interval 4-min.) × 24 (Weeks) mixed design with week at moderate to high intensity. One-half of the
repeated measures on the last two factors. Participants group was required to be female. No incentives were
were able to vary cycle ergometer wattage to increase or provided for participation. Two co-investigators (1 fe-
decrease intensity only during the continuous and 4-min. male), with focus group experience, conducted all focus
interval sessions. We did not want the participants to group interviews.
overexert on the first half of the ladder on the 2-min inter-
vals, and the sprint intervals were already set for max- Pilot testing recruitment
imum intensity. Table 1 contains details of the exercise A convenience sample of university kinesiology majors,
regimen. who were physically active and at least 18 years of age,
Prior to conducting the intervention, we conducted were recruited to test the game and mechanics of the
focus groups of astronauts, astronaut trainers, and highly cycle ergometer-game interface for six sessions within a
physically active middle-aged adults. In addition, pilot 2-week period. Flyers were posted throughout the kinesi-
tests were carried out to refine the exergame and assess- ology building and announcements were made in kinesi-
ments. In the 6-month study, assessments were made on ology classes. We attempted to recruit for equal
numerous physiological, performance, and psychological numbers of men and women. Participants were told that
variables. This protocol paper adheres to the SPIRIT they would be given a “Training Like an Astronaut” t-
guidelines. shirt for their participation. Testing was conducted by
members of the research team.
Participants Solicitation of participants for the 6-day pilot (Aim 2)
Recruitment and eligibility criteria was for those who were 30–62 years of age who would
There were three phases of recruitment: focus groups, like to improve fitness and who exercise at least 30 min.
piloting testing, and the 24-week intervention study. per day, three times per week at moderate to high inten-
Each involved separate criteria. sity. We strived for a mean age of 48 years for the sam-
ple, the mean age of an experienced astronaut.
Focus group recruitment Participants were recruited from the local community
Recruitment of astronauts and astronaut trainers took through flyers posted at races, fitness centers, and ath-
place at the Johnson Space Center in Houston, TX. Eligi- letic shops; through emails to running, triathlete, and
bility for astronauts included long-duration flight and cycling clubs; and through employee fitness programs.
Table 1 Weekly exercise regimen for 24-week study
Day Workout Description
1 Continuous 5 min. warm-up starting at 50% of HRmax and progressing in intensity until Ss reach an effort that will
elicit 75% max HR, followed by 30 min. of continuous cycling at or above 75% max HR. Ss are allowed
to increase or decrease their work intensity.
2 Long intervals 5 min. warm-up starting at 50% of HR max. and progressing in intensity until Ss reach an effort that will
elicit 90% max HR, followed by 4 × 4 min intervals at or above 90% max HR with 3 min. active rest
(at 50% HR max). Ss are allowed to increase or decrease their work intensity.
3 Continuous
4 Medium intervals 5 min. warm-upstarting at 50% of HR max. and progressing in intensity until Ss reach an effort that will
elicit 70% max HR, followed by 6 × 2 min intervals at 70, 80, 90, 100, 90, 80% max HR with 2 min. active
rest (at 50% HR max). Ss are not allowed to increase or decrease their work intensity.
5 Continuous
6 Short, sprint intervals After 10 min. warm-up, progressing in intensity until Ss reach an effort that will elicit max HR (100%),
Ss pedal at that same workload for 8 × 30 s sprint intervals with 20 s active rest. Ss are not allowed
to increase or decrease their work intensity.
Feltz et al. BMC Psychology (2016) 4:54 Page 5 of 11

Compensation for the study included a “Training Like Study samples

an Astronaut” t-shirt and $6.00 per session to cover The study sample for Aim 1 included two focus groups.
mileage and parking to be received at the end of the The first group comprised four experienced astronauts
study. Payment was not contingent on exercise (2 female) and three astronaut personal trainers (1 fe-
performance. male). Participants reviewed the prototype design of the
Potential participants were screened using the Phys- male and female SG partners and provided feedback on
ical Activity Readiness Questionnaire (i.e., and ex- facial features and expressions, somatotype, voice, and
cluded if they answered ‘yes’ to anyone of the verbal interactions. They also provided feedback on the
following: have a heart condition that precludes par- features of the game, including the variety of exercise
ticipating in moderate to vigorous physical activity; terrain, workout summary (average RPM, distance trav-
feel chest pain during physical activity; feel chest pain eled, etc.), and the virtual trainer who provided game in-
while resting; lose balance because of dizziness, lose structions. Based on feedback from the first focus group,
consciousness; have joint or bone problems that could a second focus group of four highly active male and fe-
be made worse by physical activity; if a doctor is cur- male athletes/exercisers, over 35 years of age, reviewed a
rently prescribing medication for blood pressure or second version of the SG partner (more muscular, more
heart condition; or if there are any known reasons expressive) and game (more varied terrain) that had
why the participant should not do physical activity). been developed to further refine the appearance of the
Also, all men over 44 years. had to obtain physician SG partners, exergame interface, and the nature and
consent prior to participating. quality of interactions between participants and their SG
Potential participants were also screened by self- partners (e.g., detail of introductions, greetings). After
reported physical activity levels. Eligibility required at conducting focus groups, the game was pilot tested on a
least 30 min. of physical activity per day, three times per convenience sample of six highly active kinesiology stu-
week at moderate to high intensity. Those who met the dents (2 female) at the university. They rode a stationary
initial qualifications were given an incremental exercise cycle on a simulated bike path for 30 min. for 6 days
test (cycle ergometer) in order to estimate their maximal within 2 weeks to test game mechanics and protocol.
oxygen consumption (VO2 max). Participants were re- The study sample for Aim 2 consisted of 82 highly
quired to reach an estimated VO2 max value of 35 ml/ physically active adults, ages 30 – 62 years. These were
kg/min or achieve the 150-watt stage of the test. Any participants who would be of similar age and fitness
potential participants who did not meet the aforemen- levels to those who would be recruited for the full inter-
tioned requirements were excluded from the study. A vention study.
total of six participants either self-selected out of the For the long-term intervention (Aim 3), a total of 419
study after the incremental exercise test or did not participants expressed interest in enrolling in the study.
qualify. Of those, 221 completed a screening survey on Qual-
trics. A final sample of 41 highly physically active adults,
(18 female, 44%; one Hispanic) enrolled in the study in
Twenty-four week intervention study recruitment two cohorts. The first cohort of participants included 11
Solicitation of participants for the 24-week intervention women, 12 men (M age = 46.74 ± 6.98). The second co-
study (Aim 3) was similar to the 6-day pilot study. How- hort included 7 women, 11 men (M age = 44.17 ± 9.31).
ever, recruitment has been conducted in two separate co- We have strived for equal numbers of males and fe-
horts across 2 years because of space limitations to conduct males, but in any case, insured proportional numbers of
the intervention. Compensation for the study included the males and females in each condition.
same “Training Like an Astronaut” t-shirt and $6.95 per Power analyses were performed using G*Power soft-
session. Participants received payment only for the number ware (see gpower.hhu.de). To examine effects within and
of days they complete for a potential total of $1000. Pay- between treatment groups for the primary effort
ment was provided on a monthly basis but was not contin- dependent variable, a repeated-measures ANOVA of
gent on performance. Instead of a graded exercise test, three measurement blocks (for averaged exercise session
potential participants performed the same test as previously data), with a moderate correlation among repeated mea-
described for the six-day pilot study, but expired respiratory sures (ρ = 0.3), suggested we would detect a moderate ef-
gases were collected during this test so VO2 was measured, fect (f = .35) with the sample size of 13 individuals per
rather than estimated. As was the case with the 6-day pilot group (total N = 39) with a probability of .95. These pro-
study, participants had to reach at least 35 ml/kg/min or jections were consistent with previous conjunctive-
make it through 150 Watt stage of max test to qualify for partnered studies conducted by the research group that
this phase of the study. The astronaut average VO2 max is have shown large treatment effects (e.g., d = .99 [4]; d =
~42 ml/kg/min. and we strived for that as a sample mean. 1.38 [16]).
Feltz et al. BMC Psychology (2016) 4:54 Page 6 of 11

Description of SPACE exergame, cycle ergometer might be in the study. Although experimenters were not
interface, and testing facility blind to conditions, they were unaware of the experi-
The exergame, developed specifically for this study, in- mental hypotheses.
corporated the exercise regimen of continuous and
interval training and included a series of different bike Control
paths for each of the 6 days. SPACE includes a same-sex In the individual control condition, participants cycled
SG trainer who provides instructions for all of the work- under the same conditions and instructions as in the
outs. In the partnered conditions, the game includes a previous week. They must try to cycle at their target
same-sex SG partner who is introduced by the SG watts or higher for the Continuous 30-min. and 4-min
trainer as a teammate intervals sessions; however, if they feel they cannot cycle
Participants can view workout information on the at that intensity, they can lower their watts. They re-
screen, such as current intensity level, (measured in ceived feedback, prior to each session, on their previous
watts), RPMs, the distance cycled, whether they are workout performance for the type of workout they have
above or below their target watts, and a clock that that day (i.e., continuous 30-min, 4-min interval, etc.).
counts down from 30 min for the continuous workout
protocol or for the specific interval time they are work- Always superior partner (AWS)
ing on for interval days. Participants can change inten- Participants were told that they will be cycling for the
sity or bike speed (+/- 5, 10, or 20 watts) by selecting rest of the study with an SG partner who will be their
the appropriate buttons on their keypad. Because of the teammate. They were told that their teammate, named
design of the cycle ergometer, participants cannot Chris (always same sex as participant), is programmed
change their intensity by pedaling faster or slower —for to be slightly more fit than they are, but that he/she is
example, pedaling more slowly results in an increase in designed to respond to a workout as any person would,
resistance, keeping the overall intensity constant. and can experience fatigue at some point during the ex-
SPACE is interfaced with the Monark LC4 cycle erg- ercise session. This was manifested in two ways: (a) false
ometer (with adjustable seat and handle bar) and is used feedback on potential initial, baseline performance from
in conjunction with a PC, monitor, and numeric keypad. Week 2 (the SG partner was alleged to have been pro-
Participants pedal the cycle ergometer at a fixed wattage grammed to cycle about 1.15 times faster than the par-
based on a prescribed percentage of their target HRmax ticipant) and (b) during the exercise session, the image
while viewing their gameplay via the computer monitor. of the SG partner was shown on the video monitor start-
The testing facility has six exercise cubicles and can ac- ing out at the same pace as the participant but quickly
commodate up to six participants at a time. There is a moving into a faster pace and then always be shown to
separate room to perform pre- and post-session mea- be outperforming the participant. The 15% discrepancy
sures (e.g., blood pressure). Additionally, another lab in between participant and SG partner was determined
the building houses all physiological equipment relevant from feedback in focus-group testing.
to the current investigation, including an isokinetic ma- The SG trainer explained the nature of the conjunctive
chine (Biodex 3 dynamometer), body composition (Bod- task (i.e., that they are working together toward a team
pod), and metabolic carts (Parvo). workout score, which is determined by whoever bikes the
shorter distance). Further, they were told that they and
Intervention their teammate are linked together, so that if one of them
All qualified participants had 6 days of baseline cycling cycles too far ahead AND the other is below their target
in Week 1, using the 6-day exercise regimen, to adjust watts, the team member who is ahead will be slowed
their work intensity (as set in watts) without the SPACE down, until the gap has lessened. When this happens, they
game. In Week 2, all participants were introduced to the will also see a red hue on the bottom of the screen, indi-
SPACE game, known to them as “Training Like and cating that their teammate has had to slow down.
Astronaut” and went through the no-partner Control In order to build rapport with the SG partner, partici-
condition of the 6-day exercise regimen at their target pants were introduced to him/her, through a guided
watts intensity, where an SG same-sex trainer appeared dialogue-tree interface which allows participants to re-
on the monitor and provided instructions for the work- spond to questions posed by their SG partner by selecting
outs. After Week 2, the project manager randomized from several pre-programmed responses on screen. Exam-
participants, balanced for gender, to one of three condi- ples include asking if the participant is from Michigan
tions: no-partner Control, Always superior partner with response choices of “Yeah, I’m a Michigander” or
(AWS), or Not always superior partner (NAS). Partici- “Actually, I’m not from around here.” Depending on the
pants were blinded to the conditions they were in and choice option chosen, the SG partner responds with ap-
were told not to discuss their exercise with anyone who propriate follow-up dialogue and additional questions.
Feltz et al. BMC Psychology (2016) 4:54 Page 7 of 11

There are five question and response interactions. The times during each of the interval sessions. During these
type of personal information exchanged was developed occasions, the NAS partner says things like, “Sorry, I just
through focus group testing as part of the SG develop- couldn’t keep up today” or “Those sprints were tough!”.
ment the SG partner.
In terms of continued rapport-building, the SG partner
comments periodically throughout the 24-week inter- Procedures
vention, either before or after the workout. This occurs Prior to the start of the study, participants were invited to
a total of 32 times with comments such as, “Ready”; an information and orientation session in the facility
“Good work”; and “I’m looking forward to our workout.” where the study would be conducted. The session pro-
In addition to the introductory dialogue, there are three vided information on the pretest measures that would be
other occasions where a dialogue-tree format occurs: (a) conducted (VO2 max test, muscular strength test of quad-
in Week 13 at the halfway point of the study, (b) in riceps and hamstrings, body composition), compensation
Week 21 with 1 month to go, and (c) on the last day, for participation, and the schedule. Participants could
where the SG partner thanks the participant for working choose either a Sunday through Friday or a Monday
out with him/her and says goodbye and the participant through Saturday time schedule. Participants were then
is then given the option to reply. scheduled for their pretest assessments. Descriptions of
As a part of designing the SG partner to have human these pretest measures are detailed under Measures.
qualities but still be recognized as a computer, there are
two separate weeks when he/she is not available. In Pretest assessments
Week 9, the SG partner is sick with a software “virus” Pretest assessments were scheduled 1 to 3 weeks prior
for 4 days, and in Week 19, he/she has an injury and is to the first week of the study. Participants came to the
out for the week. exercise testing lab and were first administered the in-
Just before participants begin the game after meeting formed consent to participate in the study. If partici-
their SG partner for the first time, the SG trainer ex- pants consented, they then were given the instructions
plains that they can see their average watts and the dis- and were evaluated for body composition and performed
tance they biked in their first session (Week 2 baseline) the VO2 max test. After the completion of the VO2 max
compared with their SG partner’s average watts and dis- test, researchers indicated whether participants met the
tance. The SG partner’s performance also shows a 15% eligibility criteria for VO2 max. If participants met cri-
better score. They are then reminded that their workout teria, they were scheduled for a second lab visit where
score for this session is determined by whoever cycles their lower body muscular strength was measured. After
the shorter distance in 30 min. continuous workout (or the second testing visit was completed participants were
4-min interval) and, that on the screen, they will be able then ready to begin participating with the exergame.
to see both their own score and the team score. They
were instructed that if they finished ahead of the SG
partner, the partner’s score would be the team score. If Week 1
the SG partner finished ahead of them, their own score During the first week of the study, participants came to
would be the team score. At any time that the partici- the exercise facility to get accustomed to the 6-day exer-
pant is ahead of the SG partner, the partner is visible in cise regimen, outlined in Table 1. Exercise was initially
a profile view in the corner of the screen. As in the Con- prescribed based on percentages of VO2max and was
trol condition, participants receive feedback, prior to monitored by HR. For example, the HR and workload
each session, on their own previous workout perform- corresponding to 90% of VO2max during the initial max
ance for the type of workout they have that day. test was recorded as the training target for the 4 × 4 min
intervals. Participants performed all of the workouts
Not always superior partner (NAS) under the supervision of research staff with an exercise
All information provided to the NAS participants is the physiology background and the research staff adjusted
same as that provided to those in the AWS condition. the participant’s work intensity if necessary based on the
The only difference in the two conditions is that the par- HR. If adjustments to participants’ cycle power output
ticipant was sometimes be able to surpass the SG part- were made by the research staff, the adjusted watts were
ner. This happened on 17 of the possible 117 occasions used for all subsequent sessions. Participants were not
(15%) where participants cycled with a partner. The 15% exposed to the game in this adjustment week. Prior to
was a best guess on how to both maintain the impres- the start of each session (during this week and through-
sion of partner superiority and minimize participant out the entire study), participants were fitted with a
discouragement based on focus group responses. There Polar HR monitor and a blood pressure cuff for pre-
are eight times during continuous sessions and three exercise cardiovascular measurements.
Feltz et al. BMC Psychology (2016) 4:54 Page 8 of 11

Week 2: baseline Measures

As described previously, during Week 2, all participants Mean change in cycle ergometer workload (Watts) from
cycled in the Control condition, which includes having the initial continuous and 4 min. interval session is the
an SG trainer to provide instructions. Participants were primary dependent variable reflecting motivational ef-
instructed to wear the headphones provided or to bring fort. Measures of physiological, strength, and psycho-
their own so that they could hear the instructions and logical parameters (perceived effort, enjoyment, self-
not overhear participants in adjacent exercise stations. efficacy, and social connectedness) were also obtained. A
HR, blood pressure, and rating of perceived exertion schedule of the frequency of measures administered is
(RPE) measures were taken during each session of base- contained in Table 2.
line and continued throughout the study. During the
continuous workout, HR was collected at 10, 20, and Primary measures
30 min. RPE was collected at the end of the workout Effort Mean change in cycle ergometer workload (watts)
(30 min. mark). HR and RPE were taken at the end of from each participant’s initial targeted workload (indi-
each 4 × 4 min. interval (4, 11, 17, and 25 min.). During vidually determined during max testing), as well as mean
the sprint workout, RPE was collected at the end of the change in workload over 24 weeks from baseline are the
10 min. warm up and at the end of the 8th sprint. HR primary dependent variables reflecting work effort for
was collected at the end of the warm-up and after each both the continuous and 4-min. interval sessions. With-
of the 8 intervals in the sprint workout. HR was col- out missing data, there are a total of 66 Continuous, 22
lected at the end of each of the 6 × 2 min. intervals and 4 × 4 min interval, 22 6 × 2 min interval, and 22 sprint
RPE was collected at the end of the 4th and 6th intervals. interval effort measures. Effort to persist in the 2-min
At the end of the workouts, participants were required interval workouts was measured in the number of sec-
to cool-down for 3 min. or until HR was below 130 onds completed without decreasing intensity. Effort to
beats per min. Blood pressure was taken after the work-
out and the participant was given a survey if there was Table 2 Frequency of measures
one for the day they just completed. Measure Applicable workouts Frequency
Fitness variables: Not specific to Weeks 1, 12, and 24
Week 3 through 24 VO2 max a workout
Ventilatory
Starting in Week 3, participants were randomly assigned threshold
to one of the three conditions. All procedures remained Thigh strength
the same. During Week 13, participants completed mid- Body composition
point VO2 max, body composition, and leg strength test- Free-living physical Not specific to Weekly
ing using the same protocol from the beginning of the activity a workout
study. Using the fitness data from the VO2 max test, ad- Rating of perceived Continuous Daily during sessions
justments could be made to prescribed watts for partici- exertion Sprint intervals
Short intervals
pants if training effects had occurred. If participants Long intervals
improved their VO2 max, prescribed watts were adjusted Heart rate Continuous Daily during sessions
to the new regression line that predicts their perform- Sprint intervals
ance at a variety of intensities. Short intervals
Long intervals
Strategies to improve adherence to the intervention in-
cluded follow-up phone calls if participants missed a ses- Blood pressure Not specific to Before and after each
a workout session
sion to try to minimize any drop out or adherence
problems. Occasional nonparticipation was anticipated Self-efficacy Continuous Weeks 2, 4, 8, 12, 17, 20, 24
Long intervals
as inevitable and managed afterward in missing data
Enjoyment Continuous Weeks 1, 2, 3, 9, 24
analysis. The project manager also established a secure
Sprint intervals
and direct online communication with all participants to Short intervals
encourage proactive notification and management of any Long intervals
potential issues with scheduled session appointments. Social connectedness Not specific to Weeks 12, 24
This deterred missed appointments that occur due to a workout
multiple inherent, non-study personal schedule conflicts. Team perceptions Not specific to Weeks 3, 12, 24
Adherence to the protocol was monitored in the lab by a workout
staff. Participation would be discontinued for any indi- Alternative Godspeed Not specific to Weeks 3, 12, 24
Indices a workout
vidual who incurred an injury, within or outside of the
exercise sessions that could be aggravated by continuing Game interest Not specific to Week 24
a workout
the exercise program.
Feltz et al. BMC Psychology (2016) 4:54 Page 9 of 11

persist was also measured in the sprint workouts by the Enjoyment, social connectedness, and interest in the
number of intervals completed (up to a maximum of 8). game Enjoyment was assessed with a 5-item version of
Participant-controlled changes in watts above the target the Physical Activity Enjoyment Scale [19] regarding
were not allowed in the 2-min interval workout, though “how you feel at the moment about the physical activity
they could decrease their watts if the workout at their you have been doing.” The Social Connectedness Scale
prescribed watts was too difficult. In the sprint interval, [20] was adapted to include six items to measure partici-
participants could not decrease watts, but they could pants’ sense, in general, of companionship, togetherness,
stop if unable to continue. and relatedness during the exercise sessions. Game
interest was assessed on the final day with a single item
Rating of perceived exertion (RPE) Along with an that asked participants to rate their interest in playing
objective measure of effort, participants reported their an exercise video.
subjective evaluation of effort levels throughout exercise,
using the 15-point version of the Borg RPE scale (minimal Self-efficacy Self-efficacy beliefs were measured pre-
effort = 6; maximum effort = 20, which when multiplied by and post session for the continuous and 4-min interval
10 corresponds to an estimate of heart rate). Ratings were workouts only. Pre-session efficacy was rated in terms of
then averaged for each experimental session. We assessed the up-coming workout and post-session efficacy was
RPE because perceptions of exertion also may influence rated regarding beliefs about the next time the partici-
one’s motivation to persist at a taxing task [17]. pant engaged in that workout. Participants rated their
confidence on an 11-point probability scale that they
VO2max. and ventilatory threshold The test was con- could cycle for 30 min (continuous workout) at six dif-
ducted using an electronic cycle ergometer, and expired ferent intensities, starting at 75% of the participant’s
respiratory gases were collected using a Parvo metabolic VO2max and increasing by 5% for each item up to 100%.
cart. The test ended when participants reached volitional For the 4-min interval workout, participants rated their
exhaustion and stopped pedaling or were instructed to confidence on the same scale that they could complete
stop when they reached two out of three criteria indicat- all four, 4-min intervals (4-min interval workout) start-
ing that they had achieved VO2max (i.e., plateau in VO2 ing at 90% of their VO2 max and increasing by 5% up to
occurred, heart rate was higher than 95% of predicted, 100% (i.e., 3 items).
and respiratory exchange rate was over 1.05.) Ventilatory
threshold values were calculated from the VO2max test Additional perceptual measures To check for percep-
data by identifying the breakpoint of pulmonary ventila- tions of the SG partner’s humanness, participants in
tion (VE) from VO2 as described by Amann et al. These partnered conditions completed a questionnaire that
tests were measured at the time of screening, at Week used the Alternative Godspeed Indices [21]. To check
13, and at posttest [18]. for participants’ perceptions of their partner as a team-
mate, partnered participants completed a 5-item ques-
Quadriceps and hamstring strength Isometric muscu- tionnaire that assessed their perceptions of the working
lar strength was measured at the right knee (45 deg, 5 s relationship with their partner (e.g., I felt I was part of a
contraction) including peak torque extension and peak team) [22, 23].
torque flexion using the Biodex 3 dynamometer. Partici-
pants warmed-up with four repetitions of extension and Monitored measures Blood pressure and HR were
flexion, followed by 3 repetitions each for the actual test. monitored for safety. Blood pressure was measured
These tests were performed at the time of screening, at using a GE Dinamap automatic blood pressure monitor.
Week 12, and at posttest. HR was measured using the Polar HR monitor during
each exercise session. Additionally, participants were
Body composition Each participant’s body density was asked to recall any outside activity and record the activ-
measured via air displacement plethysmography (Bod- ity and the duration they participated in the activity to
Pod), and %fat and fat free mass were calculated accord- account for any outside influences on performance and
ing to standard equations. fitness measures.

Secondary measures Statistical analyses

A number of psychological measures were taken to as- For the continuous and 4-min. sessions, a blocked ses-
sess participants’ enjoyment, self-efficacy, social connect- sion repeated-measures analyses of variance (ANOVA)
edness, and interest in the game. In addition, measures will be utilized to examine the effect of treatment on the
of participants’ perceptions of the SG partner were primary motivational effort measure (mean change in
assessed for those in the partnered conditions. power output) across conditions and across the study.
Feltz et al. BMC Psychology (2016) 4:54 Page 10 of 11

Group and individual linear trend analyses and growth The study also has some limitations. The exercise pro-
curve modeling will be applied to assess potential inter- gram uses only the aerobic segment of NASA’s SPRINT
versus intra-individual variability trajectories and pat- exercise regimen. Including the weight training compo-
terns of change over the 24 weeks. ANOVA will be used nent would provide a more thorough test of the effect-
to analyze the mean differences for percent of cycling 2- iveness of an SG partner to sustain motivation in a
min. intervals and 30-s. sprints at or above targeted level vigorous exercise program. Additionally, participants did
of effort. In all analyses of these measures, baseline per- not live in an isolated environment, similar to what as-
formance measures will also be covaried to control for tronauts experience at the ISS or on deep space mis-
pre-program levels of fitness. Missing data will be evalu- sions. The effects of an SG partner may be even stronger
ated for randomness and a within-subjects, same-session when used in an environment where inhabitants are so-
type linear interpolation will be used to impute the data cially isolated, especially in testing its social connections
set. Interpolation of missing data was chosen due to the effects. Further, the interactions with the SG partner
nature of the effort measure (i.e., longitudinal objective were relatively superficial and limited by the software.
measure of exercise effort). Having an SG partner with some artificial intelligence
Hypotheses related to the effect of treatment on ob- capabilities could enrich the social interaction such that
jective measures of fitness (e.g., VO2max measures; ven- the SG partner is more than just a cycling partner but
tilatory threshold; isokinetic strength) will be tested also a relational agent.
using ANOVA at the pre, mid, and post-study time The authors plan to communicate results to partici-
points. pants and through a local community presentation. We
Given that prior research has found little evidence that plan to communicate scientific results through presenta-
enhanced effort achieved with conjunctive exercise part- tions at professional conferences and through publica-
ners is accompanied by any aversion to the exercise task, tions in scientific journals.
loss in self-efficacy, or rise in subjectively experienced
Abbreviations
exertion, we will check to see if these patterns replicate AWS: Always superior condition; HRmax: Maximum heart rate; IRB: Institutional
in this study. We will also examine self-reports of social Review Board; ISS: International Space Station; ml/kg/min: Milliliters per
connectedness both generally and toward the SG part- kilogram per minute; NAS: Not always superior condition; NASA: National
Aeronautics and Space Administration; NSBRI: National Space Biomedical
ner, in terms of feeling like a teammate. For each vari- Research Institute; PAES: Physical Activity Enjoyment Scale; RPE: Rating of
able, we will use a general linear model to analyze perceived exertion; SG: Software-generated; SPACE: Simulated Partners and
means across the multiple time points administered and Collaborative Exercise; SPRINT: NASA’s high intensity integrated resistance
and aerobic training program; VE: Volume of expired respiratory gases; VO2
for the entire study, as well as correlations to blocked max: Maximum rate of oxygen consumption
means for effort measures. We also will check for sex
and age differences. Funding
Space is funded by NASA/National Space Biomedical Research Institute,
MA03401. The funding source has had no role in the design of the study
and collection, analysis, and interpretation of data, nor in writing the
Discussion manuscript.
This paper describes the rationale, development, and
Availability of data and materials
methods of the SPACE exergame program. This inter- The dataset that will be analyzed during the current study will be available
vention was designed to maintain intense levels of exer- from the corresponding author on reasonable request. There will be no
cise for astronauts and other adults who need to personal identification of participants in the data set.

exercise at vigorous levels for health and performance Authors’ contributions

purposes. We believe this will be a viable sustainable DF drafted the manuscript, conceived of the study, and was principal
intervention that can be disseminated for astronaut use investigator of the study design. LPS and NK conceived of the study,
contributed to the study design, and draft of the manuscript. BW, JP, and AE
and adapted by other populations. The strengths of the contributed to the study design and draft of the manuscript. WJ contributed
study include the strong conceptual basis of group dy- to the study design. CH and SS contributed to the draft of the manuscript.
namics, using social psychological mechanisms such as All authors read and approved the final manuscript.
social comparison and indispensability to group achieve- Competing interests
ment to boost motivation (i.e., the level of effort) for vig- The authors declare that they have no competing interests.
orous physical exercise [4]. In addition, the use of an SG
Consent for publication
partner embedded in an exergame has several practical
Not applicable.
advantages over a live human partner for astronauts as
well as other populations. For instance, an SG partner’s Ethics approval and consent to participate
abilities can be adjusted automatically over time to per- The Social Science Behavioral/Education Institutional Review Board of the
Human Research Protection Programs at Michigan State University approved
form at a level that is always challenging to the player, this study (reference number: 13-554). Members of the research team obtained
thereby keeping the player engaged and active. written informed consent from participants. Potential participants could decide
Feltz et al. BMC Psychology (2016) 4:54 Page 11 of 11

whether to participate in the study after reading the consent form and asking 19. Raedeke TD, Amorose AJ. A psychometric evaluation of a short exercise
questions. Prior to enrollment of participants, guidelines were established for enjoyment measure. J Sport Exerc Psychol. 2013;35:S110.
identifying adverse events and serious adverse events. Data have been handled 20. Lee RM, Robbins SB. Measuring belongingness: the Social Connectedness
in strict confidence and stored anonymously by identification code only. There and Social Assurance Scale. J Couns Psychol. 1995;42:232–41.
will be no personal identification of participants in scientific communications. 21. Ho CC, MacDorman KF. Revisiting the uncanny valley theory: Developing
Records will be kept in locked file cabinets and secure computer files at the and validating an alternative to the Godspeed indices. Comput Hum Behav.
University, and only the principal investigator, project manager, and formal, 2010;26:1508–18. doi:10.1016/j.chb.2010.05.015.
Institutional Review Board (IRB), research team members will have access to the 22. Irwin BC, Feltz DL, Kerr NL. Silence is golden: effect of encouragement in
data set. Any adverse events or serious adverse events will be communicated motivating the weak link in an online exercise video game. J Med Internet
immediately to the IRB. Res. 2013;15(6):e104. doi:10.2196/jmir.2551.
23. Reeves B, Nass C. The media equation. Stanford: CSLI Publicaqtions; 1996.
Author details
1
Department of Kinesiology, Michigan State University, East Lansing, MI
48824, USA. 2Department of Psychology, Michigan State University, East
Lansing, MI, USA. 3Department of Media and Information, Michigan State
University, East Lansing, MI, USA. 4Universities Space Research Association,
Houston, TX, USA. 5Department of Kinesiology, California State University,
Long Beach, CA, USA.

Received: 21 October 2016 Accepted: 5 November 2016

References
1. Ploutz-Snyder LL, Downs M, Ryder J, Hackney K, Scott J, Buxton R, Goetchius
E, Crowell B. Integrated resistance and aerobic exercise protects fitness
during bed rest. Med Sci Sports Exerc. 2014;358-68. doi:10.1249/MSS.
0b013e3182a62f85.
2. Byrne A, Byrne DG. The effect of exercise on depression, anxiety and other
mood states: a review. J Psychosom Res. 1993;37:565–74.
3. McAuley EB, Blissmer B, Marquez DX, Jerome GJ, Kramer AF, Katula J. Social
relations, physical activity, and well-being in older adults. Prev Med. 2000;31:
608–17.
4. Feltz DL, Kerr NL, Irwin BC. Buddy up: the Kohler effect applied to health
games. J Sport Exerc Psychol. 2011;33:506–26.
5. Bickmore TW, Picard RW. Establishing and maintaining long-term human-
computer relationships. ACM Trans Comput Hum Interact. 2005;12:293–327.
6. Feltz DL, Forlenza ST, Winn B, Kerr NL. Cyber buddy is better than no
buddy: A test of the Köhler motivation effect in exergame. Games Health J.
2014;3:98–105. doi:10.1089/g4h.2013.0088.
7. Max EJ, Samendinger S, Winn B, Kerr NL, Pfeiffer KA, Feltz DL. Enhancing
aerobic exercise with a novel virtual exercise buddy based on the Köhler
effect. Games Health J. 2016;5:1–6. doi:10.1089/g4h.2016.0018.
8. Nass C, Moon Y, Carney P. Are people polite to computers? Responses to
computer-based interviewing systems. J Appl Soc Psychol.
1999;29(5):1093–110.
9. Burke SM, Carron AV, Eys MA, Ntoumanis N, Estabrooks PA. Group versus
individual approach? A meta-analysis of the effectiveness of interventions to
promote physical activity. Sport Exerc Psychol Rev. 2006;2(1):13–29.
10. Dishman RK, Buckworth J. Increasing physical activity: a quantitative
synthesis. Med Sci Sports Exerc. 1996;28:706–19.
11. Yim J, Graham TCN. Using games to increase exercise motivation. Paper
presented at: Conference on Future Play 2007; New York.
12. Lieberman DA. Dance games and other exergames: What the research
says. 2006. http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=0E
6FFDAB4AB375D49E2696948FEFE8E3?doi=10.1.1.403.8924&rep=rep1
&type=pdf.
13. Madsen KA, Yen S, Wlasiuk L, Newman TB, Lustig R. Feasability of a dance Submit your next manuscript to BioMed Central
videogame to promote weight loss among overweight children and and we will help you at every step:
adolescents. Arch Pediatr Adolesc Med. 2007;161:105–7.
14. Weber B, Hertel G. Motivation gains of inferior group members: a meta- • We accept pre-submission inquiries
analytical review. J Pers Soc Psychol. 2007;93(6):973–93. • Our selector tool helps you to find the most relevant journal
15. Parise S, Kiesler S, Sproull L, Waters K. Cooperating with life-like interface
• We provide round the clock customer support
agents. Comput Hum Behav. 1999;15:123–42.
16. Irwin BC, Scorniaenchi J, Kerr NL, Eisenmann JC, Feltz DL. Aerobic exercise is • Convenient online submission
promoted when individual performance affects the group: a test of the • Thorough peer review
Kohler motivation gain effect. Ann Behav Med. 2012;44:151–9.
• Inclusion in PubMed and all major indexing services
17. Borg GAV. Psychophysical bases of perceived exertion. Med Sci Sports Exerc.
1982;14(5):377–81. • Maximum visibility for your research
18. Amann M, Subudhi A, Foster C. Influence of testing protocol on ventilatory
thresholds and cycling performance. Med Sci Sports Exer. 2004;36:613–22. Submit your manuscript at
www.biomedcentral.com/submit