ZurichOpenRepositoryand
Archive
UniversityofZurich
UniversityLibrary
Strickhofstrasse39
CH-8057Zurich
www.zora.uzh.ch
Year:2013
Terminalfeedbackoutperformsconcurrentvisual,auditory,andhaptic
feedbackinlearningacomplexrowing-typetask
Sigrist,Roland;Rauter,Georg;Riener,Robert;Wolf,Peter
Abstract:Augmentedfeedback,providedbycoachesordisplays,isawell-establishedstrategytoac-
celeratemotorlearning.Frequentterminalfeedbackandconcurrentfeedbackhavebeenshowntobe
detrimentalforsimplemotortasklearningbutsupportiveforcomplexmotortasklearning.However,
conclusionsonoptimalfeedbackstrategieshavebeenmainlydrawnfromstudiesonarticiallabora-
torytaskswithvisualfeedbackonly.Therefore,theauthorscomparedtheeectivenessoflearninga
complex,3-dimensionalrowing-typetaskwitheitherconcurrentvisual,auditory,orhapticfeedbackto
self-controlledterminalvisualfeedback.Resultsrevealedthatterminalvisualfeedbackwasmosteec-
tivebecauseitemphasizedtheinternalizationoftask-relevantaspects. Incontrast,concurrentfeedback
fosteredthe correctionof task-irrelevanterrors, whichhindered learning.Theconcurrentvisualand
hapticfeedbackgroupperformedmuchbetterduringtrainingwiththefeedbackthaninnonfeedback
trials.Auditoryfeedbackbasedonsonicationofthemovementerrorwasnotpracticalfortrainingthe
3-dimensionalmovementformostparticipants.Concurrentmultimodalfeedbackincombinationwith
terminalfeedbackmaybe mosteective, especiallyifthefeedbackstrategy isadapted toindividual
preferencesandskilllevel.
DOI:https://doi.org/10.1080/00222895.2013.826169
PostedattheZurichOpenRepositoryandArchive,UniversityofZurich
ZORAURL:https://doi.org/10.5167/uzh-89081
JournalArticle
PublishedVersion
Originallypublishedat:
Sigrist, Roland; Rauter, Georg;Riener, Robert;Wolf, Peter(2013).Terminalfeedbackoutperforms
concurrentvisual, auditory, andhapticfeedbackinlearning acomplexrowing-typetask.Journalof
MotorBehavior,45(6):455-472.
DOI:https://doi.org/10.1080/00222895.2013.826169
This art icle was downloaded by: [ UZH Haupt bibliothek / Zent ralbibliot hek Zürich]
On: 15 January 2014, At : 00: 19
Publisher: Routledge
I nform a Lt d Regist ered in England and Wales Registered Num ber: 1072954 Regist ered office: Mortim er House,
37-41 Mortim er St reet , London W1T 3JH, UK
Journal of Motor Behavior
Publicat ion det ails, including inst ruct ions f or aut hors and subscript ion inf ormat ion:
ht t p:/ / www.t andf online. com/ loi/ vj mb20
Terminal Feedback Outperforms Concurrent Visual,
Auditory, and Haptic Feedback in Learning a Complex
Rowing-Type Task
Roland Sigrist
a
, Georg Raut er
a
, Robert Riener
a
& Pet er Wolf
a
a
Sensory-Mot or Syst ems Lab, ETH Zurich & Spinal Cord Inj ury Cent er, Universit y Hospit al
Balgrist , Zurich , Swit zerland
Published online: 05 Sep 2013.
To cite this article: Roland Sigrist , Georg Raut er , Robert Riener & Pet er Wolf (2013) Terminal Feedback Out perf orms
Concurrent Visual, Audit ory, and Hapt ic Feedback in Learning a Complex Rowing-Type Task, Journal of Mot or Behavior, 45: 6,
455-472, DOI: 10.1080/ 00222895. 2013.826169
To link to this article: ht t p: / / dx.doi.org/ 10. 1080/ 00222895.2013. 826169
PLEASE SCROLL DOWN FOR ARTI CLE
Taylor & Francis m akes every effort to ensure t he accuracy of all t he inform at ion (t he “ Content” ) contained
in t he publicat ions on our platform . However, Taylor & Francis, our agents, and our licensors m ake no
represent ations or warranties what soever as t o the accuracy, com pleteness, or suit ability for any purpose of t he
Content. Any opinions and views expressed in t his publicat ion are t he opinions and view s of the aut hors, and
are not t he views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with prim ary sources of inform at ion. Taylor and Francis shall not be liable for
any losses, actions, claim s, proceedings, dem ands, cost s, expenses, dam ages, and other liabilit ies what soever
or howsoever caused arising direct ly or indirect ly in connect ion with, in relat ion to or arising out of t he use of
the Content .
This art icle m ay be used for research, t eaching, and privat e st udy purposes. Any subst antial or system at ic
reproduct ion, redist ribut ion, reselling, loan, sub-licensing, system at ic supply, or distribut ion in any
form t o anyone is expressly forbidden. Term s & Condit ions of access and use can be found at htt p: / /
www.tandfonline.com/ page/ t erm s-and-condit ions
Journal of Motor Behavior, Vol. 45, No. 6, 2013
Copyright
C
Taylor & Francis Group, LLC
RESEARCH ARTICLE
Terminal Feedback Outperforms Concurrent Visual, Auditory,
and Haptic Feedback in Learning a Complex Rowing-Type Task
Roland Sigrist, Georg Rauter, Robert Riener, Peter Wolf
Sensory-Motor Systems Lab, ETH Zurich & Spinal Cord Injury Center, University Hospital Balgrist, Zurich, Switzerland.
ABSTRACT. Augmented feedback, provided by coaches or dis-
plays, is a well-established strategy to accelerate motor learning.
Frequent terminal feedback and concurrent feedback have been
shown to be detrimental for simple motor task learning but sup-
portive for complex motor task learning. However, conclusions on
optimal feedback strategies have been mainly drawn from studies
on artificial laboratory tasks with visual feedback only. Therefore,
the authors compared the effectiveness of learning a complex, 3-
dimensional rowing-type task with either concurrent visual, audi-
tory, or haptic feedback to self-controlled terminal visual feedback.
Results revealed that terminal visual feedback was most effective
because it emphasized the internalization of task-relevant aspects.
In contrast, concurrent feedback fostered the correction of task-
irrelevant errors, which hindered learning. The concurrent visual
and haptic feedback group performed much better during training
with the feedback than in nonfeedback trials. Auditory feedback
based on sonification of the movement error was not practical for
training the 3-dimensional movement for most participants. Concur-
rent multimodal feedback in combination with terminal feedback
may be most effective, especially if the feedback strategy is adapted
to individual preferences and skill level.
Keywords: augmented feedback, haptic guidance, movement soni-
fication, self-controlled feedback, skill learning
A
ugmented feedback is a well-accepted strategy to ac-
celerate motor skill learning. Augmented feedback
provides information about motor performance or result, pre-
sented by an external source, such as a trainer, therapist,
or display (Schmidt & Wrisberg, 2008). To be successfully
applied, research has aimed to determine optimal feedback
principles with respect to frequency, delay, focus of atten-
tion, and content, among others (Schmidt & Wrisberg, 2008;
Wulf & Shea, 2002). In the past, simple or artificial tasks
rather than real-life complex tasks have been investigated,
mainly because the related paradigms are more straight-
forward to test, to set up, and to analyze. However, Wulf
and Shea stated that feedback principles derived from sim-
ple tasks cannot be transferred to complex motor tasks. In
particular, it seems that the guidance hypothesis proposing
that concurrent or frequent feedback is detrimental for mo-
tor learning due to emerging dependency on the feedback
(Salmoni, 1984; Schmidt, 1991; Schmidt, Young, Swinnen,
& Shapiro, 1989), holds true for simple tasks (Schmidt &
Wulf, 1997; Van der Linden, Cauraugh, & Greene, 1993;
Winstein et al., 1996), but not for complex tasks (Marschall,
Bund, & Wiemeyer, 2007; Swinnen, Lee, Verschueren, Ser-
rien, & Bogaerds, 1997; Wulf, Shea, & Matschiner, 1998).
In an early stage of complex task learning, concurrent feed-
back may accelerate learning by mediating a general idea
of the movement (Huegel & O’Malley, 2010; Liebermann
et al., 2002) and by preventing cognitive overload (Wulf &
Shea, 2002). Indeed, concurrent visual feedback has facili-
tated learning of different complex tasks (Kovacs & Shea,
2011; Snodgrass, Rivett, Robertson, & Stojanovski, 2010;
Swinnen et al., 1997; Todorov, Shadmehr, & Bizzi, 1997;
Wishart, Lee, Cunningham, & Murdoch, 2002; Wulf, H
¨
orger,
& Shea, 1999). Research on feedback principles such as the
guidance hypothesis has predominantly addressed the visual
modality, thereby neglecting a comparison with feedback
in other modalities such as auditory and haptic feedback
(Sigrist, Rauter, Riener, & Wolf, 2013).
Concurrent augmented visual feedback may interfere with
motor tasks that depend on visually perceived informa-
tion. In such tasks, the attention on the environment and
the augmented feedback compete and may overload the
learner. Auditory feedback might be less interfering and
distracting as, in contrast to visual information, neither a
specific focus on the display nor a specific orientation of
the head in space is required (Eldridge, 2006; Grond, Her-
mann, Verfaille, & Wanderley, 2010; Secoli, Milot, Rosati,
& Reinkensmeyer, 2011). Auditory feedback such as move-
ment sonification (i.e., the mapping of a movement variable
to a parameter of sound such as pitch or volume) enhanced
learning of time-dependent dynamic coordination in rowing
(Effenberg & Mechling, 1998) and swimming (Chollet,
Madani, & Micallef, 1992; Chollet, Micallef, & Rabischong,
1988). In a task on the German Wheel, which is a large wheel
for gymnastics, movement sonification was only effective
for experts, but not for novices (Hummel, Hermann, Frauen-
berger, & Stockman, 2010). This supports the assumption
that the effectiveness of movement sonification is limited to
athletes who already have a high skill level and thus can in-
terpret the sonification and relate it to an optimal movement
(Sigrist et al., 2013). Novices may not benefit, as they cannot
estimate how the optimal movement should sound. To sup-
port also novices, error sonification may be effective (i.e.,
the auditory representation of the current deviation from a
target movement to a parameter of sound, instead of directly
mapping the movement variable to a parameter of sound as
done in movement sonification). In a study on pistol shooting,
mapping of the deviation from the aiming point to changes in
pitch was more effective than feedback about the score only,
i.e., knowledge of results (Konttinen, Mononen, Viitasalo,
& Mets, 2004; Mononen, 2007). Although multidimensional
error sonification is very challenging to design, it has already
Correspondence address: Roland Sigrist, ETH Zurich, Institute
of Robotics and Intelligent Systems (IRIS), Sonneggstrasse 3 (ML G
57), 8092 Zurich, Switzerland. e-mail: roland.sigrist@hest.ethz.ch
455
Downloaded by [UZH Hauptbibliothek / Zentralbibliothek Zürich] at 00:19 15 January 2014
R.Sigristetal.
been shown to be interpretable for a rowing-type movement
(Sigrist, Schellenberg, et al., 2011). However, its effective-
ness in enhancing complex motor learning has not yet been
evaluated.
Haptic guidance is another possibility for providing con-
current augmented feedback. Haptic guidance leads a person
through or toward a target movement by addressing kines-
thetic and sometimes also tactile perception. In contrast to
visual or auditory feedback, haptic guidance can completely
restrict the movement to the desired one (e.g., by a robot
in position control; Bluteau, Coquillart, Payan, & Gentaz,
2008; Liu, Cramer, & Reinkensmeyer, 2006). However, it is
believed that some freedom in performing the movement is
needed. First, because errors drive motor learning (Emken,
Benitez, & Reinkensmeyer, 2007; Patton, Stoykov, Kovic, &
Mussa-Ivaldi, 2006; van Beers, 2009) and, second, to pre-
vent passivity (Israel, Campbell, Kahn, & Hornby, 2006)
and slackness (Reinkensmeyer, Akoner, Ferris, & Gordon,
2009) of the user that may hinder motor learning (Shad-
mehr & Mussa-Ivaldi, 1994). According concepts have of-
ten been based on a path controller (Khatib, 1986; Rauter
et al., 2010; Rauter, Sigrist, Marchal-Crespo, et al., 2011;
Vallery, Guidali, Duschau-Wicke, & Riener, 2009). A path
controller is a controller that provides spatial guidance us-
ing, for example, conservative force fields to generate a
virtual tunnel/path with elastic walls (Duschau-Wicke, von
Zitzewitz, Caprez, Lunenburger, & Riener, 2010; Marchal-
Crespo, Rauter, Wyss, von Zitzewitz, & Riener, 2012). Such
a haptic guidance concept outperformed a control group
without haptic guidance in terms of driving accuracy in a
steering task (Marchal-Crespo & Reinkensmeyer, 2008) and
in a wheelchair driving task (Marchal-Crespo, Furumasu,
& Reinkensmeyer, 2010). However, complex tasks or even
sportive tasks have not been investigated to date.
In summary, studies on concurrent augmented feedback in
complex motor learning have rarely considered the auditory
or haptic modality. Furthermore, training with one feedback
strategy has usually been compared to training without feed-
back or to another feedback strategy of the same modality.
However, in order to optimize motor learning in general, the
effects of feedback strategies provided in different modali-
ties should be systematically exploited (Sigrist et al., 2013).
Thereby, it is important to consider the interaction of the
feedback presentation with the stage of learning (Magill &
Anderson, 2012). Consequently, in this study, we compared
and contrasted learning of groups that trained a complex,
multidimensional rowing-type task in several sessions with
either concurrent visual, auditory, or haptic feedback.
A control group trained with terminal self-controlled feed-
back (i.e., feedback provided after task execution) where
the time point and frequency of feedback were selected
by the learner. A self-controlled feedback strategy was ap-
plied because it has been shown to be more effective than
an externally imposed feedback strategy (Chiviacowsky &
Wulf, 2002, 2005; Huet, Camachon, Fernandez, Jacobs, &
Montagne, 2009; Janelle, Barba, Frehlich, Tennant, & Cau-
raugh, 1997; Janelle, Kim, & Singer, 1995). Benefits of a self-
controlled strategy are seen in the adaptation to the learner’s
needs, focus on the current aspect the learner wants to train,
and involvement of the learner in the learning process result-
ing in an increased motivation (Wulf, 2007). In this study,
terminal feedback was delayed for 10 s in order to enable
self-estimation and information processing (Swinnen et al.,
1990; Winstein, 1991).
We hypothesized, first, that the benefits of terminal feed-
back should be observable only after the initial learning stage
(i.e., after a few training sessions) because the learners must
first develop a general internal movement representation to
which the terminal feedback could be compared to. Second,
in contrast, concurrent visual and haptic feedback should ac-
celerate learning in a very early stage because they should
effectively mediate the general idea of the complex move-
ment and, thus, the understanding of the movement pattern.
Thereby, visual feedback should be most beneficial for learn-
ing spatial aspects of the trajectory (Feygin, Keehner, &
Tendick 2002), due to the distinctive ability of humans to
visually perceive spatial information (Freides, 1974; Nesbitt,
2003; Welch & Warren, 1980). Haptic feedback should con-
tribute most to learning of temporal (Feygin et al., 2002)
but also of spatiotemporal features (Marchal-Crespo et al.,
2010; Marchal-Crespo & Reinkensmeyer, 2008) as haptic
perception is well developed for both aspects (Nesbitt, 2003).
One—dimensional error sonification was shown to reduce a
spatial error (Konttinen et al., 2004; Mononen, 2007). Thus,
third, we hypothesized that the three-dimensional error soni-
fication would facilitate motor learning; however, a longer
familiarization time to the auditory feedback is expected than
to the visual, haptic, or terminal feedback.
Method
Participants
All 36 participants (8 women, 28 men; age range =
22–40 years; M age = 28 years, SD = 3.7 years) were healthy,
nonrowers, without prior experience of the task, and had nor-
mal hearing and normal or corrected-to-normal vision. The
participants signed an agreement following the guidelines of
the local ethics commission, which had approved the study.
Apparatus and Task
During each experiment, the participant was seated in an
actual (trimmed) rowing boat, set up in the middle of a CAVE
(Cave Automated Virtual Environment; von Zitzewitz et al.,
2008). Reflective markers were attached to a trimmed sweep
rowing oar to track its movement with an optoelectrical mo-
tion tracking system (Qualisys, Gothenburg, Sweden). The
kinematic data were used to virtually elongate the physical
oar in real-time on a 4.44 m × 3.33 m screen as well as to
provide input for the concurrent visual, auditory, and termi-
nal feedback, and to analyze movement errors at 62.5 Hz.
Standard headphones (Sennheiser HDR 170, GmbH & Co.
456 Journal of Motor Behavior
Downloaded by [UZH Hauptbibliothek / Zentralbibliothek Zürich] at 00:19 15 January 2014
Visual, Auditory, and Haptic Augmented Feedback
KG, Wedemark Wennebostel, Germany; frequency response:
18Hz to 21kHz) were used to display the auditory feedback.
The haptic augmented feedback was realized via a rope-robot
attached to the end of the trimmed oar (Rauter et al., 2010).
The horizontal and vertical oar angles were thereby measured
with the rope robot, while the blade rotation was computed
from the data of two wire potentiometers at 1000 Hz. During
the entire study, the rope robot was attached to the end effec-
tor (i.e., the outer end of the trimmed oar) for the participants
of the haptic group only.
The task to be learned was very similar to body-arm rowing
taught to beginners in their first rowing lessons and used by
experienced rowers as a warm-up exercise. The participant
moved the sweep rowing oar with both hands and both arms
as well as the trunk, without moving the legs. The requested
rowing-type oar movement required a horizontal range of
the outer hand of about 25
◦
(0.50 m) and a vertical range
of about 13
◦
(0.26 m). As in real rowing, the blade had to
be turned to a vertical orientation (about 90
◦
) before it was
pulled (drive phase) and turned to a flat orientation (0
◦
) when
it was pushed (recovery phase). To increase task complexity,
an angular velocity profile that is typical for rowing, was
applied to the oar movement (i.e., the angular velocity was
two times higher in the pulling phase than in the pushing
phase). As water resistance was not simulated in this study,
the participant only felt the inertia and the remaining friction
of the oar throughout the whole movement. One cycle lasted
6 s, resulting in a stroke rate of 10 strokes/min.
Feedback Designs
Concurrent Visual Feedback
The visual feedback was created according to an evalua-
tion of different designs performed in a prior study (Sigrist,
Schellenberg, et al., 2011). The feedback consisted of a su-
perposition of the virtual blade on the transparent virtual
target blade, which was moving along the displayed target
trajectory (Figure 1). The feedback was programmed in Lua
(Ierusalimschy, 2013; http://www.lua.org).
Concurrent Auditory Feedback
Conclusions from earlier experiments revealed that uni-
modal auditory feedback for a three-dimensional rowing-
type movement is unambiguous and practical if the current
deviation from the target is sonified in each degree of free-
dom separately (Sigrist, Schellenberg, et al., 2011). There-
fore, the auditory feedback applied in this study was based
on the error sonification introduced in earlier experiments.
The deviation from the target oar was mapped to a sound
modulation. Changes in stereo balance represented changes
in horizontal deviation, changes in pitch represented changes
in vertical deviation, changes in volume represented the com-
bined vertical and horizontal deviation, and a modulated tim-
bre represented incorrect blade orientation (Figure 2).
Stereo balance (i.e., information from the left and right
headphones) was correlated to the horizontal deviation. The
FIGURE 1. Superposition-based visual feedback. The vir-
tual elongation of the oar in brown was superimposed on a
transparent virtual target oar in blue, moving along the target
trajectory. The brightness of the target trajectory indicated
the target oar velocity (the brighter, the faster). Vertical bars
on the target trajectory indicated the location of blade ro-
tation. To enhance the perception of the relation of the oar
blade positions in relation to the target trajectory, virtual
spheres were attached to the blades. (Color figure available
online).
deviation was minimal when a center-balanced signal (vol-
ume left channel = volume right channel) was heard. To
facilitate the interpretation of stereo balance mapping, the
participant was instructed to orient the head toward the outer
end of the real oar.
Pitch is commonly described by terms such as high or
low; therefore, vertical deviation was mapped to pitch. The
frequency for a correct vertical angle was 370 Hz (f#0).
Around this frequency, the pitch ranged within ±0.9 octaves
FIGURE 2. Auditory feedback mapping. The deviation
from the participant’s oar to the target oar was represented
by modulations of stereo balance, pitch, volume, and timbre.
(Color figure available online).
2013, Vol. 45, No. 6 457
Downloaded by [UZH Hauptbibliothek / Zentralbibliothek Zürich] at 00:19 15 January 2014
