Learning to throw on a rotating carousel: recalibration based on limb dynamics and projectile kinematics.
Experimental Brain Research (2005)
- PubMed: 15696309
Available from
Hugo Bruggeman's profile on Mendeley.
or
Abstract
Skilled actions exhibit adjustment in calibration to bring about their goals. The sought-after calibrations change as a function of the environmental situation that stages the actions. In these experiments participants sat on one side of a rotating carousel and threw beanbags underhanded at a target fixed on the opposite side. Logically, aimed throwing in this situation involves adjustment to fit changes in limb dynamics (originating from Coriolis forces) and changes in perceived projectile kinematics (originating from the tangential velocity of thrower and target). We studied whether such adjustment involved one or multiple components of recalibration. An initial experiment showed that exposure to rotation while throwing beanbags produced a robust recalibration in the direction of underhanded throws as manifest in throwing at stationary targets from a stationary position. Following some initial decay this recalibration persisted and approached an asymptote. Subsequent experiments suggested two independent components of recalibration. One is based on limb dynamics and accounts for the initial decay. The other is based on the perceived projectile kinematics and accounts for the stable change in throwing direction. These results raised the question of how multiple components of recalibration of an action are related. We propose that movement components are independent and calibrated separately at different levels in the organization of an action.
Author-supplied keywords
Available from
Hugo Bruggeman's profile on Mendeley.
Page 1
Learning to throw on a rotating carousel: recalibration based on limb dynamics and projectile kinematics.
Experimental Brain Research, In Press
Learning to throw on a rotating carousel:
Recalibration based on limb dynamics and
projectile kinematics
1HUGO BRUGGEMAN
2HERBERT L. PICK, JR.
3JOHN J. RIESER
1Department of Cognitive and Linguistic Sciences,
Brown University, Box 1978, Providence, RI 02912-1978, USA
2Institute of Child Development,
University of Minnesota, Minneapolis, MN, USA
3Department of Psychology and Human Development,
Vanderbilt University, Nashville, TN, USA
hugo_bruggeman@brown.edu
(++1) 401.863.1398
(++1) 401.863.2255
Learning to throw on a rotating carousel:
Recalibration based on limb dynamics and
projectile kinematics
1HUGO BRUGGEMAN
2HERBERT L. PICK, JR.
3JOHN J. RIESER
1Department of Cognitive and Linguistic Sciences,
Brown University, Box 1978, Providence, RI 02912-1978, USA
2Institute of Child Development,
University of Minnesota, Minneapolis, MN, USA
3Department of Psychology and Human Development,
Vanderbilt University, Nashville, TN, USA
hugo_bruggeman@brown.edu
(++1) 401.863.1398
(++1) 401.863.2255
Page 2
RECALIBRATION OF THROWING Page 2 of 26
ABSTRACT
Skilled actions exhibit adjustment in calibration to bring about their goals. The sought-after calibrations change
as a function of the environmental situation that stages the actions. In these experiments participants sat on one
side of a rotating carousel and threw beanbags underhanded at a target fixed on the opposite side. Logically,
aimed throwing in this situation involves adjustment to fit changes in limb dynamics (originating from Coriolis
forces) and changes in perceived projectile kinematics (originating from the tangential velocity of thrower and
target). We studied whether such adjustment involved one or multiple components of recalibration. An initial
experiment showed that exposure to rotation while throwing beanbags produced a robust recalibration in the
direction of underhanded throws as manifest in throwing at stationary targets from a stationary position.
Following some initial decay this recalibration persisted and approached an asymptote. Subsequent experiments
suggested two independent components of recalibration. One is based on limb dynamics and accounts for the
initial decay. The other is based on the perceived projectile kinematics and accounts for the stable change in
throwing direction. These results raised the question of how multiple components of recalibration of an action
are related. We propose that movement components are independent and calibrated separately at different levels
in the organization of an action.
Aimed throwing Adaptation Limb dynamics Projectile kinematics Levels of recalibration
ABSTRACT
Skilled actions exhibit adjustment in calibration to bring about their goals. The sought-after calibrations change
as a function of the environmental situation that stages the actions. In these experiments participants sat on one
side of a rotating carousel and threw beanbags underhanded at a target fixed on the opposite side. Logically,
aimed throwing in this situation involves adjustment to fit changes in limb dynamics (originating from Coriolis
forces) and changes in perceived projectile kinematics (originating from the tangential velocity of thrower and
target). We studied whether such adjustment involved one or multiple components of recalibration. An initial
experiment showed that exposure to rotation while throwing beanbags produced a robust recalibration in the
direction of underhanded throws as manifest in throwing at stationary targets from a stationary position.
Following some initial decay this recalibration persisted and approached an asymptote. Subsequent experiments
suggested two independent components of recalibration. One is based on limb dynamics and accounts for the
initial decay. The other is based on the perceived projectile kinematics and accounts for the stable change in
throwing direction. These results raised the question of how multiple components of recalibration of an action
are related. We propose that movement components are independent and calibrated separately at different levels
in the organization of an action.
Aimed throwing Adaptation Limb dynamics Projectile kinematics Levels of recalibration
Page 3
RECALIBRATION OF THROWING Page 3 of 26
INTRODUCTION
Skilled actions exhibit adjustment in calibration to bring about their action goals. The sought-
after calibrations change as a function of environmental situations that stage the actions. The
question is how recalibrations come about, that is, how do people fine tune their actions to fit
variants in environmental situations?
In this paper, we distinguish dynamic recalibration from kinematic recalibration and discuss
this distinction in the context of reaching and throwing tasks. By dynamic recalibration we
mean compensating for perturbing forces to maintain typical movement characteristics such
as straight trajectories and bell-shaped velocity profiles. By kinematic recalibration we mean
compensating for spatial transformations to maintain movements coordinated with respect to
their path and end points. Dynamic recalibration can be achieved purely on the basis of
proprioceptive information, while kinematic recalibration requires exproprioceptive and
exteroceptive information (typically visual and/or auditory information).
Previous research mainly studied recalibration separately for dynamic transformations
(Conditt et al. 1997; Gandolfo et al. 1996b; Goodbody and Wolpert 1998; Lackner and Dizio
1998; Lackner and DiZio 1994; Shadmehr and Mussa-Ivaldi 1994) and kinematic
transformations (Bingham and Romack 1999; Flanagan and Rao 1995; Jakobson and Goodale
1989; Martin et al. 1996; Welch et al. 1993). Only a few studies have exposed participants to
a situation with both altered dynamics and kinematics (Flanagan, 1999; Krakauer et al, 1999;
Tong et al, 2002). The latter allows assessment of whether one or multiple components of
recalibration occur to fit such a dual change in situation.
No more than one study examined a reaching task for a situation with concurrently altered
dynamics and kinematics (Flanagan, 1999). Participants were able to adjust to such a
combination of transformations. Yet, a more rapid adjustment was observed when
participants first learned the kinematic and dynamic transformations separately. Moreover,
upon adjustment to the combined transformations, participants could learn the kinematic
transformation more rapidly. These findings suggest that adjustment to altered dynamics and
kinematics is organized as independent calibrations.
In two other studies that examined a reaching task participants were subjected to serial order
situations with altered dynamics and kinematics, and tested for interference among them. An
INTRODUCTION
Skilled actions exhibit adjustment in calibration to bring about their action goals. The sought-
after calibrations change as a function of environmental situations that stage the actions. The
question is how recalibrations come about, that is, how do people fine tune their actions to fit
variants in environmental situations?
In this paper, we distinguish dynamic recalibration from kinematic recalibration and discuss
this distinction in the context of reaching and throwing tasks. By dynamic recalibration we
mean compensating for perturbing forces to maintain typical movement characteristics such
as straight trajectories and bell-shaped velocity profiles. By kinematic recalibration we mean
compensating for spatial transformations to maintain movements coordinated with respect to
their path and end points. Dynamic recalibration can be achieved purely on the basis of
proprioceptive information, while kinematic recalibration requires exproprioceptive and
exteroceptive information (typically visual and/or auditory information).
Previous research mainly studied recalibration separately for dynamic transformations
(Conditt et al. 1997; Gandolfo et al. 1996b; Goodbody and Wolpert 1998; Lackner and Dizio
1998; Lackner and DiZio 1994; Shadmehr and Mussa-Ivaldi 1994) and kinematic
transformations (Bingham and Romack 1999; Flanagan and Rao 1995; Jakobson and Goodale
1989; Martin et al. 1996; Welch et al. 1993). Only a few studies have exposed participants to
a situation with both altered dynamics and kinematics (Flanagan, 1999; Krakauer et al, 1999;
Tong et al, 2002). The latter allows assessment of whether one or multiple components of
recalibration occur to fit such a dual change in situation.
No more than one study examined a reaching task for a situation with concurrently altered
dynamics and kinematics (Flanagan, 1999). Participants were able to adjust to such a
combination of transformations. Yet, a more rapid adjustment was observed when
participants first learned the kinematic and dynamic transformations separately. Moreover,
upon adjustment to the combined transformations, participants could learn the kinematic
transformation more rapidly. These findings suggest that adjustment to altered dynamics and
kinematics is organized as independent calibrations.
In two other studies that examined a reaching task participants were subjected to serial order
situations with altered dynamics and kinematics, and tested for interference among them. An
Page 4
RECALIBRATION OF THROWING Page 4 of 26
initial study found a lack of interference of a dynamic adjustment with a kinematic adjustment
(Krakauer, et al, 1999). In another study similar adjustments did interfere with each other
when matched on kinematic variables (Tong et al, 2002). The latter suggests that in such a
situation adjustment to altered dynamics and kinematics is no longer organized as independent
calibrations.
Whereas the above studies examined a reaching task, we report on directional recalibration in
a throwing task. Adjustment in direction occurs frequently in various natural hurling actions.
Consider, for example: A person throwing with a side wind; a hockey player making a shot at
a goal while skating a curved path; or even a tennis player coordinating her arm’s swing while
swiveling at the hips. These examples suggest that throwing actions are calibrated for
situations affecting the dynamics of the body and its limbs, and for situations affecting the
kinematics of the projectile (the action’s resultant visual event).
The aim of this report on recalibration in underhand throwing was to examine whether
adjustment based on limb dynamics is organized independently from adjustment based on
projectile kinematics. For this purpose, we developed a new situation to stage aimed throwing
modeled after Lackner and DiZio (1994, 1998). Participants sat on one side of a rotating
beam and threw beanbags in an underhanded style at a target at the opposite end. (The
rotating beam is essentially a carousel.) In such a situation corrections in aimed throwing
consisted of at least two aspects. One aspect is a correction to compensate for the changed
dynamics generated on the arm/hand during the limb trajectory to launch a beanbag. The
other aspect involves correction of the thrower’s aim to compensate for change in the
perceived flight trajectory of a beanbag in relation to the target (i.e. the perceived projectile
kinematics). Both corrections are in the same direction, for instance to the thrower’s left for
counterclockwise rotation of the carousel. (See Task Analysis for detailed information.)
In summary, the main purpose of the present study was to examine the recalibration of an
action by simultaneously altering limb dynamics and projectile kinematics in order to assess
whether one or multiple components of recalibration occur. Throwing on a rotating carousel
provides such a situation. The course of adjustment in the direction of throwing and the
persistence of the recalibration that results were followed. Parts of this study were published
in abstract form (Bruggeman et al. 2003). The study was approved by the Institutional
Review Board of the University of Minnesota.
initial study found a lack of interference of a dynamic adjustment with a kinematic adjustment
(Krakauer, et al, 1999). In another study similar adjustments did interfere with each other
when matched on kinematic variables (Tong et al, 2002). The latter suggests that in such a
situation adjustment to altered dynamics and kinematics is no longer organized as independent
calibrations.
Whereas the above studies examined a reaching task, we report on directional recalibration in
a throwing task. Adjustment in direction occurs frequently in various natural hurling actions.
Consider, for example: A person throwing with a side wind; a hockey player making a shot at
a goal while skating a curved path; or even a tennis player coordinating her arm’s swing while
swiveling at the hips. These examples suggest that throwing actions are calibrated for
situations affecting the dynamics of the body and its limbs, and for situations affecting the
kinematics of the projectile (the action’s resultant visual event).
The aim of this report on recalibration in underhand throwing was to examine whether
adjustment based on limb dynamics is organized independently from adjustment based on
projectile kinematics. For this purpose, we developed a new situation to stage aimed throwing
modeled after Lackner and DiZio (1994, 1998). Participants sat on one side of a rotating
beam and threw beanbags in an underhanded style at a target at the opposite end. (The
rotating beam is essentially a carousel.) In such a situation corrections in aimed throwing
consisted of at least two aspects. One aspect is a correction to compensate for the changed
dynamics generated on the arm/hand during the limb trajectory to launch a beanbag. The
other aspect involves correction of the thrower’s aim to compensate for change in the
perceived flight trajectory of a beanbag in relation to the target (i.e. the perceived projectile
kinematics). Both corrections are in the same direction, for instance to the thrower’s left for
counterclockwise rotation of the carousel. (See Task Analysis for detailed information.)
In summary, the main purpose of the present study was to examine the recalibration of an
action by simultaneously altering limb dynamics and projectile kinematics in order to assess
whether one or multiple components of recalibration occur. Throwing on a rotating carousel
provides such a situation. The course of adjustment in the direction of throwing and the
persistence of the recalibration that results were followed. Parts of this study were published
in abstract form (Bruggeman et al. 2003). The study was approved by the Institutional
Review Board of the University of Minnesota.
Page 5
RECALIBRATION OF THROWING Page 5 of 26
TASK ANALYSIS
In principle, adjustment on the carousel might involve autonomous corrections to compensate
for the altered projectile kinematics and the altered limb dynamics. Each one is discussed
below.
Recalibration based on the perceived projectile kinematics
On the rotating carousel, the perceived flight trajectory of a beanbag in relation to the target is
directly affected by the movement of the thrower and the motion of the target. These motions
result in a thrown beanbag missing the target to the trailing side of the target. Furthermore,
the straight flight trajectory of the beanbag appears as a curved path when perceived from the
rotating carousel. Adjustment to compensate for these effects is governed by visuomotor
control. That is, the recalibration is based on the perceived projectile kinematics.
The effect of the thrower’s movement on the direction of throw is simply explained by a
Pythagorean sum of the aiming velocity (Vaim ? 5m/s) and the tangential velocity of the
beanbag at the moment of release. On the carousel the latter is equal to the tangential
component of the movement of the thrower, Vtan ? 1m/s. Thus, when aiming straight ahead
(towards the center of rotation), the tangential velocity of the thrower alters the direction of
the flight trajectory laterally and causes the beanbag to miss the target by ? ? 11° (Figure 1a).
In addition to the thrower’s movement, there is the continuous angular motion of the target at
45°/s. This would displace the target by ? ? 23° relative to the center of rotation during the
time of the beanbag’s flight (estimated at 0.5s). Relative to the position of the thrower, such
displacement results in an error of ? ? 11° (Figure 1a). Accordingly, the thrower has to adjust
her throwing direction by about 22° (? + ?).
However changing the direction of throw on a rotating carousel to compensate for both one’s
own movement and the target’s motion is complex. The issue lies in the specification of these
displacements to the thrower. They might be specified by two different frames; one frame is
provided by the carousel and another frame is provided by the room. Seated on the carousel
the thrower’s perspective is stationary relative to the carousel and undergoes a
rotation/translation relative to the room.
TASK ANALYSIS
In principle, adjustment on the carousel might involve autonomous corrections to compensate
for the altered projectile kinematics and the altered limb dynamics. Each one is discussed
below.
Recalibration based on the perceived projectile kinematics
On the rotating carousel, the perceived flight trajectory of a beanbag in relation to the target is
directly affected by the movement of the thrower and the motion of the target. These motions
result in a thrown beanbag missing the target to the trailing side of the target. Furthermore,
the straight flight trajectory of the beanbag appears as a curved path when perceived from the
rotating carousel. Adjustment to compensate for these effects is governed by visuomotor
control. That is, the recalibration is based on the perceived projectile kinematics.
The effect of the thrower’s movement on the direction of throw is simply explained by a
Pythagorean sum of the aiming velocity (Vaim ? 5m/s) and the tangential velocity of the
beanbag at the moment of release. On the carousel the latter is equal to the tangential
component of the movement of the thrower, Vtan ? 1m/s. Thus, when aiming straight ahead
(towards the center of rotation), the tangential velocity of the thrower alters the direction of
the flight trajectory laterally and causes the beanbag to miss the target by ? ? 11° (Figure 1a).
In addition to the thrower’s movement, there is the continuous angular motion of the target at
45°/s. This would displace the target by ? ? 23° relative to the center of rotation during the
time of the beanbag’s flight (estimated at 0.5s). Relative to the position of the thrower, such
displacement results in an error of ? ? 11° (Figure 1a). Accordingly, the thrower has to adjust
her throwing direction by about 22° (? + ?).
However changing the direction of throw on a rotating carousel to compensate for both one’s
own movement and the target’s motion is complex. The issue lies in the specification of these
displacements to the thrower. They might be specified by two different frames; one frame is
provided by the carousel and another frame is provided by the room. Seated on the carousel
the thrower’s perspective is stationary relative to the carousel and undergoes a
rotation/translation relative to the room.
Page 6
RECALIBRATION OF THROWING Page 6 of 26
In principle, the visible frame provided by the room would provide information on self
movement and target motion relative to the room. (Participants can perceive that they
themselves and the target are in motion). Yet as the thrower’s task was to toss a beanbag at a
target on the opposite side of the carousel, it is the visible frame provided by the carousel that
controls the task. In this frame, there is no target motion to perceive egocentrically and there
is no target motion to perceive relative to the carousel, the target’s immediate frame of
reference. Furthermore, this frame does not specify self movement relative to the target. In
this frame, however, from trial to trial participants have the chance to watch the beanbag’s
flight path in relation to the target after it is thrown. This visual event reflects self movement
and target motion and is likely used to adjust the direction of throw.
Perceived from a top view perspective (attached to the room) the beanbag, although perturbed
in direction from the goal, travels in a straight trajectory determined by its linear momentum
(Figure 1a). However, perceived from the rotating carousel, the thrower observes the (erring)
beanbag to follow a curved trajectory away from the target (Figure 1b). When asked to judge
whether beanbag paths looked straight or curved all eight participants in a pilot study judged
the path of all their thrown beanbags as curved. The thrower can not change the observed
curvature, however she can compensate for the lateral shift in landing position by changing
the aiming direction towards the left by about 22° for counter clockwise rotation (Figure 1c).
This is the recalibration based on the perceived projectile kinematics.
In principle, the visible frame provided by the room would provide information on self
movement and target motion relative to the room. (Participants can perceive that they
themselves and the target are in motion). Yet as the thrower’s task was to toss a beanbag at a
target on the opposite side of the carousel, it is the visible frame provided by the carousel that
controls the task. In this frame, there is no target motion to perceive egocentrically and there
is no target motion to perceive relative to the carousel, the target’s immediate frame of
reference. Furthermore, this frame does not specify self movement relative to the target. In
this frame, however, from trial to trial participants have the chance to watch the beanbag’s
flight path in relation to the target after it is thrown. This visual event reflects self movement
and target motion and is likely used to adjust the direction of throw.
Perceived from a top view perspective (attached to the room) the beanbag, although perturbed
in direction from the goal, travels in a straight trajectory determined by its linear momentum
(Figure 1a). However, perceived from the rotating carousel, the thrower observes the (erring)
beanbag to follow a curved trajectory away from the target (Figure 1b). When asked to judge
whether beanbag paths looked straight or curved all eight participants in a pilot study judged
the path of all their thrown beanbags as curved. The thrower can not change the observed
curvature, however she can compensate for the lateral shift in landing position by changing
the aiming direction towards the left by about 22° for counter clockwise rotation (Figure 1c).
This is the recalibration based on the perceived projectile kinematics.
Page 7
RECALIBRATION OF THROWING Page 7 of 26
Figure 1. Calibration based on the perceived projectile kinematics for a carousel rotating
counterclockwise. Throws miss the target when aimed directly towards the target. In ‘a’, a
top view perspective - attached to earth - demonstrates the effects of the tangential velocity of
the thrower (? ? 11°) and target (? ? 11°). In ‘b’, a top view perspective - rotating with the
carousel - displays the curved path of a beanbag due to rotation. Finally, in ‘c’ an illustration
of a participant perspective which demonstrates the path of a beanbag for a throw that is
adjusted for these factors.
Recalibration based on the limb dynamics
During throwing on the rotating carousel, the forearm/hand is subject to a transient Coriolis
force causing it to follow a curved path. This results in a lateral shift in launching in the same
direction as the deflected path of a beanbag. However, unlike the curved path of a beanbag
the thrower can straighten the path of the forearm/hand by making the limb extremely rigid
through co-contraction of musculature, or by exerting a directly counteracting force. The
latter is preferred as a rigid limb complicates smooth coordination. Adjustment to compensate
for these effects is governed by sensorimotor control. That is, the recalibration is based on
limb dynamics1.
1 In our situation participants are seated at a radius of 1.5 m. In addition to a Coriolis force,
the limb dynamics are also transformed by a centrifugal force. As the centrifugal force is a
constant background force acting upon the body it alters both magnitude and direction of the
gravitoinertial force. Lackner and DiZio (1998) showed that a slight increase in the
background force level hindered adaptation to the Coriolis force in the absence of visual
feedback. That is, participants did not achieve complete adaptation for the lateral shift in
pointing. Provided with visual feedback, participants did achieve full adaptation (Bourdin et
al. 2001). This implies that in our situation of throwing on a carousel visual information
might be necessary to achieve complete adaptation for the transformed limb dynamics.
Figure 1. Calibration based on the perceived projectile kinematics for a carousel rotating
counterclockwise. Throws miss the target when aimed directly towards the target. In ‘a’, a
top view perspective - attached to earth - demonstrates the effects of the tangential velocity of
the thrower (? ? 11°) and target (? ? 11°). In ‘b’, a top view perspective - rotating with the
carousel - displays the curved path of a beanbag due to rotation. Finally, in ‘c’ an illustration
of a participant perspective which demonstrates the path of a beanbag for a throw that is
adjusted for these factors.
Recalibration based on the limb dynamics
During throwing on the rotating carousel, the forearm/hand is subject to a transient Coriolis
force causing it to follow a curved path. This results in a lateral shift in launching in the same
direction as the deflected path of a beanbag. However, unlike the curved path of a beanbag
the thrower can straighten the path of the forearm/hand by making the limb extremely rigid
through co-contraction of musculature, or by exerting a directly counteracting force. The
latter is preferred as a rigid limb complicates smooth coordination. Adjustment to compensate
for these effects is governed by sensorimotor control. That is, the recalibration is based on
limb dynamics1.
1 In our situation participants are seated at a radius of 1.5 m. In addition to a Coriolis force,
the limb dynamics are also transformed by a centrifugal force. As the centrifugal force is a
constant background force acting upon the body it alters both magnitude and direction of the
gravitoinertial force. Lackner and DiZio (1998) showed that a slight increase in the
background force level hindered adaptation to the Coriolis force in the absence of visual
feedback. That is, participants did not achieve complete adaptation for the lateral shift in
pointing. Provided with visual feedback, participants did achieve full adaptation (Bourdin et
al. 2001). This implies that in our situation of throwing on a carousel visual information
might be necessary to achieve complete adaptation for the transformed limb dynamics.
Page 8
RECALIBRATION OF THROWING Page 8 of 26
Further insight of the effect of the Coriolis force requires some knowledge of objects moving
in a rotating frame. In a rotating frame, a Coriolis force acts perpendicular to the trajectory of
a moving object, and its direction depends on the direction of rotation (to the right for counter
clockwise rotation). The acceleration component of the Coriolis force is equal to two times
the cross product of the velocity of a moving object (originating from its linear momentum)
and the rotation (angular velocity) of the frame; 2(?v).
In throwing from a seated position, the fingers release a beanbag when the forearm/hand is
moving at the highest speed (Debicki et al, 2004). In the situation of throwing on the rotating
carousel, we estimated the velocity of the forearm/hand to peak at about 5 m/s (Vaim), which
results in an acceleration component of the coriolis force of 7.9 m/s2 (? 0.8 g). Consequently,
a peak curvature in the path of the forearm/hand is experienced upon beanbag release.
METHODS
The current study reports a series of four experiments to examine the nature of adjustment on
a carousel resulting in a recalibration in throwing direction. All experiments analyzed
recalibration as pretest posttest comparisons. Pretest and posttest consisted of throwing at
stationary targets from a stationary position without access to auditory and visual feedback. It
was hypothesized that an adjustment on the carousel should manifest itself in the posttest as a
shift in throwing direction from the pretest.
Experiment 1: Testing the basic recalibration
Participants
Twelve right-handed undergraduate students, 8 females and 4 males, participated in the study
with an average age of 20 ± 3 years (mean ±SD). All participants gave informed consent and
were compensated for their time with extra credit points for a psychology course.
Further insight of the effect of the Coriolis force requires some knowledge of objects moving
in a rotating frame. In a rotating frame, a Coriolis force acts perpendicular to the trajectory of
a moving object, and its direction depends on the direction of rotation (to the right for counter
clockwise rotation). The acceleration component of the Coriolis force is equal to two times
the cross product of the velocity of a moving object (originating from its linear momentum)
and the rotation (angular velocity) of the frame; 2(?v).
In throwing from a seated position, the fingers release a beanbag when the forearm/hand is
moving at the highest speed (Debicki et al, 2004). In the situation of throwing on the rotating
carousel, we estimated the velocity of the forearm/hand to peak at about 5 m/s (Vaim), which
results in an acceleration component of the coriolis force of 7.9 m/s2 (? 0.8 g). Consequently,
a peak curvature in the path of the forearm/hand is experienced upon beanbag release.
METHODS
The current study reports a series of four experiments to examine the nature of adjustment on
a carousel resulting in a recalibration in throwing direction. All experiments analyzed
recalibration as pretest posttest comparisons. Pretest and posttest consisted of throwing at
stationary targets from a stationary position without access to auditory and visual feedback. It
was hypothesized that an adjustment on the carousel should manifest itself in the posttest as a
shift in throwing direction from the pretest.
Experiment 1: Testing the basic recalibration
Participants
Twelve right-handed undergraduate students, 8 females and 4 males, participated in the study
with an average age of 20 ± 3 years (mean ±SD). All participants gave informed consent and
were compensated for their time with extra credit points for a psychology course.
Page 9
RECALIBRATION OF THROWING Page 9 of 26
Materials
The test space contained a stationary chair facing four target points on the floor, arranged in a
diamond like shape. Two targets were located straight ahead at a distance of 3.0m and 6.0m
(the Close and Far Target). The other two targets were placed 16º to the right and left at a
distance of 4.5m (the Right and Left Target). A sighting device stood immediately behind the
chair, which measured beanbag landing positions accurate to .1?.
The carousel for manipulating the direction of throwing was constructed as a metal beam with
a radius of 1.85m (Figure 2). On each end of the beam was a small platform. One platform
held a chair to seat a participant and a bucket containing 45 beanbags. The other platform
held a small object (the size of a tennis ball) that was the throwing target. An electric motor
turned the carousel at 7.5 rpm.
Participants threw beanbags weighing 0.272 kg (9.6 oz). When throwing at stationary targets
(pretest and posttest) the participant was equipped with a blindfold and a headset connected to
a sound system. The sound system prevented the participant from localizing sound and
obtaining auditory feedback about accuracy of throwing during tests. The blindfold
eliminated visual feedback of throwing performance; during tests it was lowered to cover the
eyes immediately before any attempt to throw at a target.
Figure 2. A line drawing of the carousel, constructed as a metal beam with a platform on
each side. On the left is a platform that held a chair to seat a participant. On the opposite site
is a platform that held the target (or a second chair in Experiment 2).
Procedure
Participants were divided in two groups (TURNING DIRECTION); one group (n=6) that
turned clockwise (CW) on the carousel, and another group (n=6) that turned counterclockwise
(CCW).
Materials
The test space contained a stationary chair facing four target points on the floor, arranged in a
diamond like shape. Two targets were located straight ahead at a distance of 3.0m and 6.0m
(the Close and Far Target). The other two targets were placed 16º to the right and left at a
distance of 4.5m (the Right and Left Target). A sighting device stood immediately behind the
chair, which measured beanbag landing positions accurate to .1?.
The carousel for manipulating the direction of throwing was constructed as a metal beam with
a radius of 1.85m (Figure 2). On each end of the beam was a small platform. One platform
held a chair to seat a participant and a bucket containing 45 beanbags. The other platform
held a small object (the size of a tennis ball) that was the throwing target. An electric motor
turned the carousel at 7.5 rpm.
Participants threw beanbags weighing 0.272 kg (9.6 oz). When throwing at stationary targets
(pretest and posttest) the participant was equipped with a blindfold and a headset connected to
a sound system. The sound system prevented the participant from localizing sound and
obtaining auditory feedback about accuracy of throwing during tests. The blindfold
eliminated visual feedback of throwing performance; during tests it was lowered to cover the
eyes immediately before any attempt to throw at a target.
Figure 2. A line drawing of the carousel, constructed as a metal beam with a platform on
each side. On the left is a platform that held a chair to seat a participant. On the opposite site
is a platform that held the target (or a second chair in Experiment 2).
Procedure
Participants were divided in two groups (TURNING DIRECTION); one group (n=6) that
turned clockwise (CW) on the carousel, and another group (n=6) that turned counterclockwise
(CCW).
Page 10
RECALIBRATION OF THROWING Page 10 of 26
A pretest-experimental manipulation-posttest design was used, preceded by a short training
phase. During training (about 5 throws from a stationary chair), the participant was instructed
to use an underhand throwing style that started next to the hips and produced an arch-like
trajectory with the top of the arc approximately between eye and shoulder height.
During pretest and posttest the task of the participant was to throw (blindfolded) at the
stationary targets. For each trial, the participant – who was seated on the stationary chair -
was instructed to look at one of the targets, lower the blindfold and throw a beanbag. Next,
the landing position of the beanbag was recorded and it was removed before the participant
was instructed to raise the blindfold and to continue with a next trial. During both the pretest
and posttest, the participant threw a total of 24 beanbags. These throws were arranged in six
blocks (BLOCK) of four throws, each of these throws aimed at one of the four targets
(TARGET).
During the experimental manipulation, the participant sat on the chair on one end of the
rotating beam and engaged in throwing beanbags at the target on the opposite end. Again, the
task of the participant was to throw by means of the practiced underhand throwing style. In
addition, the participant was instructed to keep their gaze fixed at the target (notice that here
the participant was not equipped with the blindfold or headset/sound system and hence had
unobstructed vision and hearing). Each participant threw a total of 60 beanbags. After 45
throws a short break was required to refill the bucket with beanbags (this included stopping
and restarting rotation of the carousel). Upon completion of the 60 throws the carousel was
brought to a standstill. Participants were given a short break (about 1-2 minutes) before they
dismounted the carousel and prepared to continue with the posttest (this included traversing
about 30 feet).
Measurements & Analysis
During the pre- and posttest, landing positions of the beanbags were measured in degrees as
seen from a position just behind the seated participant. The adjustment due to the carousel
experience was expected to manifest itself in the posttest as a shift in throwing direction from
the pretest.
The predicted shift differed in direction for the CW and CCW groups. Participants rotating
CCW had to adjust their direction of throwing to the left (Figure 1c). Consequently it was
A pretest-experimental manipulation-posttest design was used, preceded by a short training
phase. During training (about 5 throws from a stationary chair), the participant was instructed
to use an underhand throwing style that started next to the hips and produced an arch-like
trajectory with the top of the arc approximately between eye and shoulder height.
During pretest and posttest the task of the participant was to throw (blindfolded) at the
stationary targets. For each trial, the participant – who was seated on the stationary chair -
was instructed to look at one of the targets, lower the blindfold and throw a beanbag. Next,
the landing position of the beanbag was recorded and it was removed before the participant
was instructed to raise the blindfold and to continue with a next trial. During both the pretest
and posttest, the participant threw a total of 24 beanbags. These throws were arranged in six
blocks (BLOCK) of four throws, each of these throws aimed at one of the four targets
(TARGET).
During the experimental manipulation, the participant sat on the chair on one end of the
rotating beam and engaged in throwing beanbags at the target on the opposite end. Again, the
task of the participant was to throw by means of the practiced underhand throwing style. In
addition, the participant was instructed to keep their gaze fixed at the target (notice that here
the participant was not equipped with the blindfold or headset/sound system and hence had
unobstructed vision and hearing). Each participant threw a total of 60 beanbags. After 45
throws a short break was required to refill the bucket with beanbags (this included stopping
and restarting rotation of the carousel). Upon completion of the 60 throws the carousel was
brought to a standstill. Participants were given a short break (about 1-2 minutes) before they
dismounted the carousel and prepared to continue with the posttest (this included traversing
about 30 feet).
Measurements & Analysis
During the pre- and posttest, landing positions of the beanbags were measured in degrees as
seen from a position just behind the seated participant. The adjustment due to the carousel
experience was expected to manifest itself in the posttest as a shift in throwing direction from
the pretest.
The predicted shift differed in direction for the CW and CCW groups. Participants rotating
CCW had to adjust their direction of throwing to the left (Figure 1c). Consequently it was
Page 11
RECALIBRATION OF THROWING Page 11 of 26
predicted that they would err in the posttest by throwing too far to the left. For participants
rotating CW the prediction was just the opposite. To analyze for differences in amount of
shift, error measurements are considered positive (+) if they are in the predicted direction and
negative (-) if in the opposite direction.
For each group as well as for each participant within a group a t-test was applied to test
whether the amount of shift was different from zero. In addition, a 2x4x6 ANOVA was
carried out to analyze the measure of shift for the between-subject effect of TURNING
DIRECTION, and the within-subject effects of TARGET and BLOCK. The Bonferroni
method (Dunn) of pair-wise comparisons was used to test for multiple comparisons for
TARGET and BLOCK.
Experiment 2: Control for effects due to rotating on carousel
To explore whether simply rotating on the carousel by itself would produce an aftereffect on
throwing direction, participants simply sat on a rotating carousel without throwing. Instead
they talked to another participant who was seated on the opposite side of the carousel.
Participants
Twelve right-handed undergraduate students, 8 females and 4 males, participated in the study
with an average age of 20 ± 2 years. All participants gave informed consent and were
compensated for their time with extra credit points for a psychology course.
Materials, Procedure, Measurements & Analysis
These parts of the experiment were identical to those in Experiment 1 with one exception. A
second chair was mounted on the carousel that replaced the target, so two participants could
be seated on the carousel during the experimental manipulation. Instead of throwing
beanbags, participants were instructed to talk with each other during a 10 minute period of
rotating on the carousel.
predicted that they would err in the posttest by throwing too far to the left. For participants
rotating CW the prediction was just the opposite. To analyze for differences in amount of
shift, error measurements are considered positive (+) if they are in the predicted direction and
negative (-) if in the opposite direction.
For each group as well as for each participant within a group a t-test was applied to test
whether the amount of shift was different from zero. In addition, a 2x4x6 ANOVA was
carried out to analyze the measure of shift for the between-subject effect of TURNING
DIRECTION, and the within-subject effects of TARGET and BLOCK. The Bonferroni
method (Dunn) of pair-wise comparisons was used to test for multiple comparisons for
TARGET and BLOCK.
Experiment 2: Control for effects due to rotating on carousel
To explore whether simply rotating on the carousel by itself would produce an aftereffect on
throwing direction, participants simply sat on a rotating carousel without throwing. Instead
they talked to another participant who was seated on the opposite side of the carousel.
Participants
Twelve right-handed undergraduate students, 8 females and 4 males, participated in the study
with an average age of 20 ± 2 years. All participants gave informed consent and were
compensated for their time with extra credit points for a psychology course.
Materials, Procedure, Measurements & Analysis
These parts of the experiment were identical to those in Experiment 1 with one exception. A
second chair was mounted on the carousel that replaced the target, so two participants could
be seated on the carousel during the experimental manipulation. Instead of throwing
beanbags, participants were instructed to talk with each other during a 10 minute period of
rotating on the carousel.
Page 12
RECALIBRATION OF THROWING Page 12 of 26
Experiments 3: Testing the component linked to the Coriolis force
To distinguish recalibration based on limb dynamics from recalibration based on the
perceived projectile kinematics, participants were blindfolded and instructed to throw straight
ahead of their nose during throwing on the carousel. During initial throws the forearm/hand
might deviate laterally due to the Coriolis force by itself. Although our apparatus did not
measure limb movements online, we hypothesized that, if participants adapt to the Coriolis
force, such adaptation should show up during the posttest as a shift from throwing direction
during the pretest.
Participants
Sixteen right-handed undergraduate students, 9 females and 7 males, participated in the study
with an average age of 20 ± 1 years. All participants gave informed consent and were
compensated for their time with extra credit points for a psychology course.
Materials, Procedure, Measurements & Analysis
These parts of the experiment were identical to those in Experiment 1 with one exception.
While on the carousel, instead of throwing at a visually specified target at the opposite site,
participants were blindfolded and instructed to throw straight ahead of their nose. Participants
were also equipped with the headset which was tuned to a radio signal (rock music station)
and served to mask auditory information on the direction of throwing, as for example might
be specified during the landings of beanbags. To simply prevent occurrence of audible
landings, one of the experimenters stepped along with the rotating carousel and caught the
thrown beanbags.
Experiment 4: Testing the stability of the recalibration
To test the stability of the recalibration in throwing direction due to adjustment on the
carousel, this experiment examined the effect of a 30 minutes delay period preceding posttest
measurements. During the delay period participants filled in a questionnaire requiring the
eye-hand coordination of writing.
Experiments 3: Testing the component linked to the Coriolis force
To distinguish recalibration based on limb dynamics from recalibration based on the
perceived projectile kinematics, participants were blindfolded and instructed to throw straight
ahead of their nose during throwing on the carousel. During initial throws the forearm/hand
might deviate laterally due to the Coriolis force by itself. Although our apparatus did not
measure limb movements online, we hypothesized that, if participants adapt to the Coriolis
force, such adaptation should show up during the posttest as a shift from throwing direction
during the pretest.
Participants
Sixteen right-handed undergraduate students, 9 females and 7 males, participated in the study
with an average age of 20 ± 1 years. All participants gave informed consent and were
compensated for their time with extra credit points for a psychology course.
Materials, Procedure, Measurements & Analysis
These parts of the experiment were identical to those in Experiment 1 with one exception.
While on the carousel, instead of throwing at a visually specified target at the opposite site,
participants were blindfolded and instructed to throw straight ahead of their nose. Participants
were also equipped with the headset which was tuned to a radio signal (rock music station)
and served to mask auditory information on the direction of throwing, as for example might
be specified during the landings of beanbags. To simply prevent occurrence of audible
landings, one of the experimenters stepped along with the rotating carousel and caught the
thrown beanbags.
Experiment 4: Testing the stability of the recalibration
To test the stability of the recalibration in throwing direction due to adjustment on the
carousel, this experiment examined the effect of a 30 minutes delay period preceding posttest
measurements. During the delay period participants filled in a questionnaire requiring the
eye-hand coordination of writing.
Page 13
RECALIBRATION OF THROWING Page 13 of 26
In addition, the rate of adaptation in throwing while on the rotating carousel was documented
based on experimenter logged judgments of landing positions of the beanbags. We expected
to confirm a gradual rate of adaptation as casually observed during Experiment 1.
Participants
Twelve right-handed undergraduate students, 6 females and 6 males, participated in the study
with an average age of 20 ± 4 years. All participants gave informed consent and were
compensated for their time with extra credit points for a psychology course.
Materials, Procedure, Measurements & Analysis
These parts of the experiment were identical to those in Experiment 1 with two exceptions.
First, an interval of thirty minutes was placed between completion of the experimental
manipulation (throwing on the carousel) and the start of the posttest. During this interval
participants completed a four page questionnaire on driving habits and orientation skills in
approximately twenty minutes, and spent the remainder of the time talking with the
experimenters.
Second, during the experimental manipulation two experimenters judged the throwing
accuracy categorically by logging the landing position of a participant’s throws. For this
purpose the platform holding the target was divided into five equal panels, 11 cm wide.
Going from left to right these could be thought of as numbered 1, 2, 3, 4, and 5. The center
panel holding the target is no. 3. Numbers 2 and 4 are adjacent to the center and numbers 1
and 5 are two panels away from the target. It was easy for the experimenters to identify
which panel a beanbag landed on. A quantitative error score for each throw was given by
arbitrarily assigning the center panel as zero, panels 2 and 4 as scores of one, and panels 2 and
5 as two. Outside the platform in either direction was assigned a score of three.
To take into account direction of error, throws missing in the expected direction were scored
positive, whereas throws in opposite direction were scored negative. Furthermore, the sixty
throws during the experimental manipulation were divided into twelve series of five
consecutive throws. For each series, a total numerical Score was calculated by summing the
scores of these five throws.
In addition, the rate of adaptation in throwing while on the rotating carousel was documented
based on experimenter logged judgments of landing positions of the beanbags. We expected
to confirm a gradual rate of adaptation as casually observed during Experiment 1.
Participants
Twelve right-handed undergraduate students, 6 females and 6 males, participated in the study
with an average age of 20 ± 4 years. All participants gave informed consent and were
compensated for their time with extra credit points for a psychology course.
Materials, Procedure, Measurements & Analysis
These parts of the experiment were identical to those in Experiment 1 with two exceptions.
First, an interval of thirty minutes was placed between completion of the experimental
manipulation (throwing on the carousel) and the start of the posttest. During this interval
participants completed a four page questionnaire on driving habits and orientation skills in
approximately twenty minutes, and spent the remainder of the time talking with the
experimenters.
Second, during the experimental manipulation two experimenters judged the throwing
accuracy categorically by logging the landing position of a participant’s throws. For this
purpose the platform holding the target was divided into five equal panels, 11 cm wide.
Going from left to right these could be thought of as numbered 1, 2, 3, 4, and 5. The center
panel holding the target is no. 3. Numbers 2 and 4 are adjacent to the center and numbers 1
and 5 are two panels away from the target. It was easy for the experimenters to identify
which panel a beanbag landed on. A quantitative error score for each throw was given by
arbitrarily assigning the center panel as zero, panels 2 and 4 as scores of one, and panels 2 and
5 as two. Outside the platform in either direction was assigned a score of three.
To take into account direction of error, throws missing in the expected direction were scored
positive, whereas throws in opposite direction were scored negative. Furthermore, the sixty
throws during the experimental manipulation were divided into twelve series of five
consecutive throws. For each series, a total numerical Score was calculated by summing the
scores of these five throws.
Page 14
RECALIBRATION OF THROWING Page 14 of 26
RESULTS
The results of Experiment 1 through 4 mostly concern pretest posttest analysis. A report on
the course of adjustment in underhand throws on the rotating carousel is included in the
results of Experiment 4 (Figure 5).
Experiment 1: Demonstration of the basic recalibration of throwing
direction
Pretest/posttest comparisons revealed that eleven of the twelve participants (one in the CCW
group) shifted significantly in the predicted compensatory direction: the adjustment on the
carousel manifested itself as shifts when throwing at stationary targets in the posttest. Figure
3 depicts the shifts in throwing direction both for the individual participants (circles) and for
the groups (bars). The effect of TURNING DIRECTION on the SIZE of the shift was not
statistically significant, F(1,10)=.43, p=.53. Collapsed over the two groups the magnitude of
the shift in throwing direction averaged 3.7 ± 1.6? (mean, ±SD), which was significantly
different from a 0? shift, t(11)=8.03, p<.0001.
RESULTS
The results of Experiment 1 through 4 mostly concern pretest posttest analysis. A report on
the course of adjustment in underhand throws on the rotating carousel is included in the
results of Experiment 4 (Figure 5).
Experiment 1: Demonstration of the basic recalibration of throwing
direction
Pretest/posttest comparisons revealed that eleven of the twelve participants (one in the CCW
group) shifted significantly in the predicted compensatory direction: the adjustment on the
carousel manifested itself as shifts when throwing at stationary targets in the posttest. Figure
3 depicts the shifts in throwing direction both for the individual participants (circles) and for
the groups (bars). The effect of TURNING DIRECTION on the SIZE of the shift was not
statistically significant, F(1,10)=.43, p=.53. Collapsed over the two groups the magnitude of
the shift in throwing direction averaged 3.7 ± 1.6? (mean, ±SD), which was significantly
different from a 0? shift, t(11)=8.03, p<.0001.
Page 15
RECALIBRATION OF THROWING Page 15 of 26
Figure 3. Results of Experiment 1 & 4. The bars plot the shift in throwing direction as a
group average (TURNING DIRECTION). The circles represent shifts per participant as an
average across 24 trials. See Figure 4 for a time sensitive measure.
The effect of TARGET was statistically significant, F(3,30)=4.34, p=.012. Tests for multiple
comparisons revealed that the shift towards the Left Target (4.6 ± 3.1?) was larger than shifts
towards the Close Target (3.4 ± 2.7?) and the Right Target (3.3 ± 2.9?). During the debriefing
session many of the participants reported difficulty throwing at the Left Target, and attributed
this to the need to avoid the knee when throwing across the body. With exception of the Left
Target, the data suggested an almost even distribution of the recalibration in throwing
direction across space.
The effect of BLOCK was statistically significant, F(5,50)=7.43, p<.0001. Tests for multiple
comparisons revealed that the shift in Block 1 (5.1 ± 2.9?) was significantly larger than shifts
in Block 3, 4 & 5. In addition, the shift in Block 2 (4.2 ± 2.3?) was significantly larger than
the shift in Block 5. An exponential fit of these data suggested that the decay in shift over
blocks approached an asymptote (Figure 4).
Figure 3. Results of Experiment 1 & 4. The bars plot the shift in throwing direction as a
group average (TURNING DIRECTION). The circles represent shifts per participant as an
average across 24 trials. See Figure 4 for a time sensitive measure.
The effect of TARGET was statistically significant, F(3,30)=4.34, p=.012. Tests for multiple
comparisons revealed that the shift towards the Left Target (4.6 ± 3.1?) was larger than shifts
towards the Close Target (3.4 ± 2.7?) and the Right Target (3.3 ± 2.9?). During the debriefing
session many of the participants reported difficulty throwing at the Left Target, and attributed
this to the need to avoid the knee when throwing across the body. With exception of the Left
Target, the data suggested an almost even distribution of the recalibration in throwing
direction across space.
The effect of BLOCK was statistically significant, F(5,50)=7.43, p<.0001. Tests for multiple
comparisons revealed that the shift in Block 1 (5.1 ± 2.9?) was significantly larger than shifts
in Block 3, 4 & 5. In addition, the shift in Block 2 (4.2 ± 2.3?) was significantly larger than
the shift in Block 5. An exponential fit of these data suggested that the decay in shift over
blocks approached an asymptote (Figure 4).
Page 16
RECALIBRATION OF THROWING Page 16 of 26
Figure 4. Shift in throwing direction for Experiment 1 (?), Experiment 3 (?) and
Experiment 4 (?). The dashed and dotted lines are exponential fits of the decay in shift over
block for respectively Experiment 1 & 3. For Experiment 4 one data point plots the change in
throwing direction as an average over 6 blocks. Bars indicate the size of the 95% confidence
intervals.
This experiment revealed that throwing on a rotating carousel produced a robust recalibration
in the direction of throw. We hypothesized that such recalibration was based on two
components, the perceived projectile kinematics and the limb dynamics. However before
examining these possibilities in more detail we assessed whether simply rotating on the
carousel by itself might induce some sort of sensory aftereffect that produced a shift in
throwing direction.
Experiment 2: Recalibration in throwing direction: not simply an
aftereffect of rotating on a carousel
Pretest/posttest comparisons revealed that participants did not exhibit a shift in throwing
direction that correlated with their direction of rotating on the carousel. The effect of
TURNING DIRECTION was not statistically significant, F(1,10)=.36, p=.56. Collapsed over
Figure 4. Shift in throwing direction for Experiment 1 (?), Experiment 3 (?) and
Experiment 4 (?). The dashed and dotted lines are exponential fits of the decay in shift over
block for respectively Experiment 1 & 3. For Experiment 4 one data point plots the change in
throwing direction as an average over 6 blocks. Bars indicate the size of the 95% confidence
intervals.
This experiment revealed that throwing on a rotating carousel produced a robust recalibration
in the direction of throw. We hypothesized that such recalibration was based on two
components, the perceived projectile kinematics and the limb dynamics. However before
examining these possibilities in more detail we assessed whether simply rotating on the
carousel by itself might induce some sort of sensory aftereffect that produced a shift in
throwing direction.
Experiment 2: Recalibration in throwing direction: not simply an
aftereffect of rotating on a carousel
Pretest/posttest comparisons revealed that participants did not exhibit a shift in throwing
direction that correlated with their direction of rotating on the carousel. The effect of
TURNING DIRECTION was not statistically significant, F(1,10)=.36, p=.56. Collapsed over
Page 17
RECALIBRATION OF THROWING Page 17 of 26
the two groups the shift in throwing direction averaged .14 ± 1.5?, which was not significantly
different from a 0? shift, t(11)=.51, p=.76.
In summary, the findings of this experiment indicated that the observed shifts in throwing
direction of Experiment 1 were not an aftereffect of simply turning on the carousel by itself.
Instead, we hypothesized those findings to result from a recalibration for multiple
components. The next two experiments examined respectively recalibration based on limb
dynamics and recalibration based on projectile kinematics.
Experiments 3: Recalibration based on limb dynamics shows rapid
decay
Pretest/posttest comparisons revealed that the effect of TURNING DIRECTION on the SIZE
of the shift was not statistically significant, F(1,14)=.77, p=.40. Collapsed over the two
groups the magnitude of the shift in throwing direction averaged 0.9 ± 2.3?, which was not
significantly different from a 0? shift, t(15)=1.54, p=.15.
The effect of BLOCK was statistically significant, F(5,70)=4.62, p=.001. Tests for multiple
comparisons revealed that the shift in BLOCK 1 (2.3 ± 3.1?) was significantly larger than
shifts in BLOCK 3, 4, 5 & 6, which ranged from .7 ± 3.7? to .3 ± 3.2?. Furthermore, an
additional series of t-tests found the shift in throwing direction to differ from a 0? shift only
for Block 1 & 2. An exponential fit of these data suggested that the decay in shift over blocks
approached an asymptote of zero, indicating rapid re-adaptation (Figure 4).
In this experiment participants did not have visual or auditory information of the accuracy of
their throw while on the carousel. This allowed us to distinguish the effect of the Coriolis
force from the effect of the perceived projectile kinematics on the recalibration in the
direction of throw. A small shift was found at the beginning of the posttest, which decayed
rapidly to a zero shift. Such a finding of rapid re-adaptation is congruent with findings by
Lackner and DiZio (1995, 1998). This decay is essentially the same as the amount of decay
found in Experiment 1 (see Figure 4). That correspondence suggests that the results of
Experiment 1 might be due to the two components: 1) a rapidly decaying Coriolis component,
and 2) a perceived projectile kinematics dependent component which is stable over blocks of
posttest trials in the absence of visual feedback.
the two groups the shift in throwing direction averaged .14 ± 1.5?, which was not significantly
different from a 0? shift, t(11)=.51, p=.76.
In summary, the findings of this experiment indicated that the observed shifts in throwing
direction of Experiment 1 were not an aftereffect of simply turning on the carousel by itself.
Instead, we hypothesized those findings to result from a recalibration for multiple
components. The next two experiments examined respectively recalibration based on limb
dynamics and recalibration based on projectile kinematics.
Experiments 3: Recalibration based on limb dynamics shows rapid
decay
Pretest/posttest comparisons revealed that the effect of TURNING DIRECTION on the SIZE
of the shift was not statistically significant, F(1,14)=.77, p=.40. Collapsed over the two
groups the magnitude of the shift in throwing direction averaged 0.9 ± 2.3?, which was not
significantly different from a 0? shift, t(15)=1.54, p=.15.
The effect of BLOCK was statistically significant, F(5,70)=4.62, p=.001. Tests for multiple
comparisons revealed that the shift in BLOCK 1 (2.3 ± 3.1?) was significantly larger than
shifts in BLOCK 3, 4, 5 & 6, which ranged from .7 ± 3.7? to .3 ± 3.2?. Furthermore, an
additional series of t-tests found the shift in throwing direction to differ from a 0? shift only
for Block 1 & 2. An exponential fit of these data suggested that the decay in shift over blocks
approached an asymptote of zero, indicating rapid re-adaptation (Figure 4).
In this experiment participants did not have visual or auditory information of the accuracy of
their throw while on the carousel. This allowed us to distinguish the effect of the Coriolis
force from the effect of the perceived projectile kinematics on the recalibration in the
direction of throw. A small shift was found at the beginning of the posttest, which decayed
rapidly to a zero shift. Such a finding of rapid re-adaptation is congruent with findings by
Lackner and DiZio (1995, 1998). This decay is essentially the same as the amount of decay
found in Experiment 1 (see Figure 4). That correspondence suggests that the results of
Experiment 1 might be due to the two components: 1) a rapidly decaying Coriolis component,
and 2) a perceived projectile kinematics dependent component which is stable over blocks of
posttest trials in the absence of visual feedback.
Page 18
RECALIBRATION OF THROWING Page 18 of 26
A final experiment further examined the stability of the recalibration based on the perceived
projectile kinematics. Previous theories on recalibration based on visual event kinematics
(Rieser et al., 1995) implied that a recalibration was stable over time until action parameters
were subjected to another context. The hypothesis is that adjustment (and readjustment) of an
action for the visual event component depended on the visual information from such event.
Consequently such a recalibration should remain the same if no action specific visual
information to the contrary is available.
Experiment 4: Recalibration based on perceived projectile
kinematics is stable over a period of 30 minutes
This experiment documented adjustment in throwing on the carousel. The course of
adjustment was characterized by two stages; an initial rapid adjustment was followed by
gradual adjustment. In their first throw all participants missed the target’s platform to the
side; they made significant errors in the range of the estimated +20°. Within 2 series of 5
throws they adjusted so that most beanbags (75%) landed on the platform (the 75 percentile
Score is smaller than 11, Figure 5). During the subsequent 10 throw series participants
gradually adjusted until they threw near the target again, with a small bias in the direction of
the initial errors as indicated by a median Score of about 4 for the last series of five throws.
A final experiment further examined the stability of the recalibration based on the perceived
projectile kinematics. Previous theories on recalibration based on visual event kinematics
(Rieser et al., 1995) implied that a recalibration was stable over time until action parameters
were subjected to another context. The hypothesis is that adjustment (and readjustment) of an
action for the visual event component depended on the visual information from such event.
Consequently such a recalibration should remain the same if no action specific visual
information to the contrary is available.
Experiment 4: Recalibration based on perceived projectile
kinematics is stable over a period of 30 minutes
This experiment documented adjustment in throwing on the carousel. The course of
adjustment was characterized by two stages; an initial rapid adjustment was followed by
gradual adjustment. In their first throw all participants missed the target’s platform to the
side; they made significant errors in the range of the estimated +20°. Within 2 series of 5
throws they adjusted so that most beanbags (75%) landed on the platform (the 75 percentile
Score is smaller than 11, Figure 5). During the subsequent 10 throw series participants
gradually adjusted until they threw near the target again, with a small bias in the direction of
the initial errors as indicated by a median Score of about 4 for the last series of five throws.
Page 19
RECALIBRATION OF THROWING Page 19 of 26
Figure 5. Adjustments in throwing direction while throwing on the rotating carousel for
Experiment 4. The error in throwing direction, Score with its range of degrees besides the
target depicted underneath, is plotted against the number of throws on the carousel (12 series
of 5 throws). The filled circles (?) indicate the median Score and the solid line is an
exponential fit of these data (R2=.72). The dotted lines indicate the variance in Score by
depicting and exponential fit for the 25% and 75% percentiles.
The adjustment on the carousel manifested itself as shifts when throwing at stationary targets
in the posttest. Pretest/posttest comparisons revealed that eleven of the twelve participants
(one in the CCW group) shifted in a compensatory direction. Figure 3 depicts the shifts in
throwing direction both for the individual participants (circles) and for the groups (bars). The
effect of TURNING DIRECTION on the SIZE of the shift was not statistically significant,
F(1,10)=.42, p=.53. Collapsed over the two groups the magnitude of the shift in throwing
direction averaged 3.1 ± 1.7?, which was significantly different from a 0? shift, t(11)=6.25,
p<.0001.
Across the four targets the shift in throwing direction ranged from 2.6 ± 3.0? to 3.4 ± 2.7?.
The effect of TARGET was not statistically significant, F(3,30)=1.33, p=.29, indicating an
even generalization of the recalibration in throwing direction across space.
Figure 5. Adjustments in throwing direction while throwing on the rotating carousel for
Experiment 4. The error in throwing direction, Score with its range of degrees besides the
target depicted underneath, is plotted against the number of throws on the carousel (12 series
of 5 throws). The filled circles (?) indicate the median Score and the solid line is an
exponential fit of these data (R2=.72). The dotted lines indicate the variance in Score by
depicting and exponential fit for the 25% and 75% percentiles.
The adjustment on the carousel manifested itself as shifts when throwing at stationary targets
in the posttest. Pretest/posttest comparisons revealed that eleven of the twelve participants
(one in the CCW group) shifted in a compensatory direction. Figure 3 depicts the shifts in
throwing direction both for the individual participants (circles) and for the groups (bars). The
effect of TURNING DIRECTION on the SIZE of the shift was not statistically significant,
F(1,10)=.42, p=.53. Collapsed over the two groups the magnitude of the shift in throwing
direction averaged 3.1 ± 1.7?, which was significantly different from a 0? shift, t(11)=6.25,
p<.0001.
Across the four targets the shift in throwing direction ranged from 2.6 ± 3.0? to 3.4 ± 2.7?.
The effect of TARGET was not statistically significant, F(3,30)=1.33, p=.29, indicating an
even generalization of the recalibration in throwing direction across space.
Page 20
RECALIBRATION OF THROWING Page 20 of 26
Over the six blocks the shift in throwing direction ranged from 2.7 ± 2.4? to 3.6 ± 3.2?. The
effect of BLOCK was not statistically significant, F(5,50)=0.90, p=.49, indicating that the
shift in throwing direction was stable over blocks.
As expected, the activities during the 30 minute period had a differential effect on the two
components of adjustment in throwing direction. Results of previous studies (Lackner and
Dizio 1998; Lackner and DiZio 1994) predict rapid re-adaptation of limb dynamics during the
30 minute period that participants freely moved their arms. Accordingly, the delayed posttest
measurement of a shift in throwing direction did no longer reflect the adjustment based on
limb dynamics as indicated by the steady level in shift of throwing direction over blocks.
Moreover the activities - and their resultant visual feedback - of filling out the questionnaire
(writing, marking text or symbols and turning pages) did not result in a complete re-
adaptation of throwing for a stationary situation. After the 30 minute period, the shift in
throwing direction was stable at a magnitude of about 3.1°. A shift of such magnitude
resembles the value of the asymptotic shift observed in Experiment 1 (Figure 4).
These findings support the notion that the recalibration based on the perceived projectile
kinematics is stable over time when no action specific visual information to the contrary is
available. We consider that renewed practice in throwing with the consequent projectile
kinematics is required to readjust to throwing for a stationary situation.
GENERAL DISCUSSION
While rotating on the carousel all participants initially made relatively large lateral errors
(errors>20°), and these decreased with practice. A rapid adjustment, which reduced errors to
less then 6°, was followed by a gradual adjustment towards the target. This adaptation
produced a robust recalibration in the direction of underhand throws as tested in throwing at
stationary targets from a stationary position. Following some initial decay the recalibration in
throwing direction persisted and approached an asymptote when no action specific visual
information to the contrary was available. The recalibration was general for targets at various
locations.
Over the six blocks the shift in throwing direction ranged from 2.7 ± 2.4? to 3.6 ± 3.2?. The
effect of BLOCK was not statistically significant, F(5,50)=0.90, p=.49, indicating that the
shift in throwing direction was stable over blocks.
As expected, the activities during the 30 minute period had a differential effect on the two
components of adjustment in throwing direction. Results of previous studies (Lackner and
Dizio 1998; Lackner and DiZio 1994) predict rapid re-adaptation of limb dynamics during the
30 minute period that participants freely moved their arms. Accordingly, the delayed posttest
measurement of a shift in throwing direction did no longer reflect the adjustment based on
limb dynamics as indicated by the steady level in shift of throwing direction over blocks.
Moreover the activities - and their resultant visual feedback - of filling out the questionnaire
(writing, marking text or symbols and turning pages) did not result in a complete re-
adaptation of throwing for a stationary situation. After the 30 minute period, the shift in
throwing direction was stable at a magnitude of about 3.1°. A shift of such magnitude
resembles the value of the asymptotic shift observed in Experiment 1 (Figure 4).
These findings support the notion that the recalibration based on the perceived projectile
kinematics is stable over time when no action specific visual information to the contrary is
available. We consider that renewed practice in throwing with the consequent projectile
kinematics is required to readjust to throwing for a stationary situation.
GENERAL DISCUSSION
While rotating on the carousel all participants initially made relatively large lateral errors
(errors>20°), and these decreased with practice. A rapid adjustment, which reduced errors to
less then 6°, was followed by a gradual adjustment towards the target. This adaptation
produced a robust recalibration in the direction of underhand throws as tested in throwing at
stationary targets from a stationary position. Following some initial decay the recalibration in
throwing direction persisted and approached an asymptote when no action specific visual
information to the contrary was available. The recalibration was general for targets at various
locations.
Page 21
RECALIBRATION OF THROWING Page 21 of 26
The results of subsequent experiments supported an hypothesis of two components in the
recalibration of throwing direction (all experiments analyzed recalibration as pretest/posttest
comparisons). A control experiment indicated that simply rotating on the carousel by itself
did not produce an adjustment in throwing direction (Experiment 2). Next, an experiment that
isolated an adjustment to the Coriolis force showed a small initial shift which decayed rapidly
to zero over trials (Experiment 3). Such rapid re-adaptation to the Coriolis force was
equivalent to the decay in shift over blocks in the original experiment. A final experiment
indicated that the recalibration in throwing direction remained stable after a 30 minute period,
even if, during such a period, participants filled in a questionnaire requiring the hand-eye
coordinated of writing (Experiment 4). This suggested that recalibration for the kinematics of
the visual event (beanbag trajectory and its landing position) accounted for the stable change
in throwing direction when no current action specific visual information to the contrary is
available. Thus, our observations on adjustment in throwing direction suggest that multiple
independent components of recalibration occur.
Previous studies found evidence for independent components of recalibration for limb
dynamics and limb kinematics in a reaching task (Flanagan et al., 1999; Krakauer et al, 1999).
However, these studies examined reaching instead of throwing, and involved different types
of spatial information: In reaching the spatial information is largely based on perceived limb
kinematics whereas in throwing the spatial information is largely based on the perceived
projectile kinematics.
The study of Flanagan et al. (1999) is similar to the present study since participants needed to
compensate for concurrently altered dynamics and kinematics. Still, Flanagan et al. and the
present study differ in several ways that may be important. For example in the present study
the dynamics and kinematics were perturbed in the same direction whereas in Flanagan et al.
they were perturbed in opposite directions. Furthermore, in the present study the dynamics
and kinematics were altered by staging the action of throwing in a new environmental
situation, whereas in Flanagan et al. the alterations were engineered by asking participants to
make reaches while holding a manipulandum. Consequently, the observed adjustment in
reaching compensates for changes of the tool hence the study could be conceived as tool
learning (see Lackner and Dizio (1994) for a similar argument).
The results of subsequent experiments supported an hypothesis of two components in the
recalibration of throwing direction (all experiments analyzed recalibration as pretest/posttest
comparisons). A control experiment indicated that simply rotating on the carousel by itself
did not produce an adjustment in throwing direction (Experiment 2). Next, an experiment that
isolated an adjustment to the Coriolis force showed a small initial shift which decayed rapidly
to zero over trials (Experiment 3). Such rapid re-adaptation to the Coriolis force was
equivalent to the decay in shift over blocks in the original experiment. A final experiment
indicated that the recalibration in throwing direction remained stable after a 30 minute period,
even if, during such a period, participants filled in a questionnaire requiring the hand-eye
coordinated of writing (Experiment 4). This suggested that recalibration for the kinematics of
the visual event (beanbag trajectory and its landing position) accounted for the stable change
in throwing direction when no current action specific visual information to the contrary is
available. Thus, our observations on adjustment in throwing direction suggest that multiple
independent components of recalibration occur.
Previous studies found evidence for independent components of recalibration for limb
dynamics and limb kinematics in a reaching task (Flanagan et al., 1999; Krakauer et al, 1999).
However, these studies examined reaching instead of throwing, and involved different types
of spatial information: In reaching the spatial information is largely based on perceived limb
kinematics whereas in throwing the spatial information is largely based on the perceived
projectile kinematics.
The study of Flanagan et al. (1999) is similar to the present study since participants needed to
compensate for concurrently altered dynamics and kinematics. Still, Flanagan et al. and the
present study differ in several ways that may be important. For example in the present study
the dynamics and kinematics were perturbed in the same direction whereas in Flanagan et al.
they were perturbed in opposite directions. Furthermore, in the present study the dynamics
and kinematics were altered by staging the action of throwing in a new environmental
situation, whereas in Flanagan et al. the alterations were engineered by asking participants to
make reaches while holding a manipulandum. Consequently, the observed adjustment in
reaching compensates for changes of the tool hence the study could be conceived as tool
learning (see Lackner and Dizio (1994) for a similar argument).
Page 22
RECALIBRATION OF THROWING Page 22 of 26
Changing the environmental situation that stages throwing, in the absence of an intermediate
tool, allows for straightforward interpretations regarding the organization of the adjustment.
That is, the recalibration based on limb dynamics on the carousel is a family member of
adjustment to gravitational-inertial forces. Therefore, we consider such change in calibration
of limb dynamics to serve a background function for various limb movements. The
recalibration based on projectile kinematics concerns the beanbag’s flight trajectory and
landing position (the action’s resultant visual event). We consider such change in calibration
to serve a goal level of any throwing task or, for that matter, any hurling task that would
generate an equivalent visual event. The idea of different levels in the organization of a
movement task is of relevance.
Tests for transfer of recalibration to limbs and/or movement tasks other than the one practiced
may permit examination of how multiple components of recalibration are organized. The
literature suggests a profound difference between transfer of a recalibration based on limb
dynamics and transfer of a recalibration based on visual event kinematics.
Studies of transfer of recalibration based on altered limb dynamics suggest generalization to
other movement tasks dependent on the spatial and anatomical properties of trajectory
formation. In tests of direction, recalibration of pointing in a viscous force field showed
generalization near the areas of original adaptation (Gandolfo et al. 1996a). For a similar
situation, almost complete generalization was found for pointing movements at different
speeds as well as for movements to targets at a twofold distance (Goodbody and Wolpert
1998). Furthermore, recalibration in reaching movements transferred to circle drawing
movements when performed in the same region (Condit et al. 1997). Finally, for pointing,
intermanual transfer was found for the shifts in end-point accuracy but not for the curvature in
movement trajectory (DiZio and Lackner 1995).
Studies of transfer of recalibration based on altered kinematics of the visual event suggest
generalization to other movement tasks by means of a goal relationship. This occurs in
recalibration of walking, for example. The kinematics of the visual event was altered by a
transformation equivalent to walking on a people conveyer. Such a situation specifically
altered the relationship between the biomechanically specified stepping speed (proprioception,
efference, and effort) and the visually specified walking speed (rate of optic flow).
Recalibration in forward walking transferred to sidestepping but did not transfer to throwing
Changing the environmental situation that stages throwing, in the absence of an intermediate
tool, allows for straightforward interpretations regarding the organization of the adjustment.
That is, the recalibration based on limb dynamics on the carousel is a family member of
adjustment to gravitational-inertial forces. Therefore, we consider such change in calibration
of limb dynamics to serve a background function for various limb movements. The
recalibration based on projectile kinematics concerns the beanbag’s flight trajectory and
landing position (the action’s resultant visual event). We consider such change in calibration
to serve a goal level of any throwing task or, for that matter, any hurling task that would
generate an equivalent visual event. The idea of different levels in the organization of a
movement task is of relevance.
Tests for transfer of recalibration to limbs and/or movement tasks other than the one practiced
may permit examination of how multiple components of recalibration are organized. The
literature suggests a profound difference between transfer of a recalibration based on limb
dynamics and transfer of a recalibration based on visual event kinematics.
Studies of transfer of recalibration based on altered limb dynamics suggest generalization to
other movement tasks dependent on the spatial and anatomical properties of trajectory
formation. In tests of direction, recalibration of pointing in a viscous force field showed
generalization near the areas of original adaptation (Gandolfo et al. 1996a). For a similar
situation, almost complete generalization was found for pointing movements at different
speeds as well as for movements to targets at a twofold distance (Goodbody and Wolpert
1998). Furthermore, recalibration in reaching movements transferred to circle drawing
movements when performed in the same region (Condit et al. 1997). Finally, for pointing,
intermanual transfer was found for the shifts in end-point accuracy but not for the curvature in
movement trajectory (DiZio and Lackner 1995).
Studies of transfer of recalibration based on altered kinematics of the visual event suggest
generalization to other movement tasks by means of a goal relationship. This occurs in
recalibration of walking, for example. The kinematics of the visual event was altered by a
transformation equivalent to walking on a people conveyer. Such a situation specifically
altered the relationship between the biomechanically specified stepping speed (proprioception,
efference, and effort) and the visually specified walking speed (rate of optic flow).
Recalibration in forward walking transferred to sidestepping but did not transfer to throwing
Page 23
RECALIBRATION OF THROWING Page 23 of 26
or turning by stepping in place (Rieser et al. 1995). A more recent study found transfer from
walking to crawling (Withagen and Michaels 2002). Furthermore, recalibration in rotational
locomotion (sidestepping in a turning fashion) did not transfer to forward walking, however
the new calibration of rotation transferred to arm movements to turn in place; participants
stood on a pivot and pulled themselves around by hands along a rail (Berry and Rieser 1999).
These findings reveal a spatial coordination in action specific terms and imply an action goal
relationship in movement organization.
The profound difference in generalization between adjustment based on limb dynamics and
adjustment based on visual event kinematics implies theoretical concepts of multiple
(hierarchical) levels in movement organization. Previously several authors presented such
ideas (Bernstein 1967, 1988/1947, 1996/1947; Greene 1972; Saltzman 1979; Schöner 1995).
Let us consider Bernstein’s ideas.
Bernstein suggested that the movements of a motor skill are organized over different levels.
Some of these levels (the Level of Tone and the Level of Muscular Articular Links) mainly
serve a background function for the organization of movements, allowing other levels (the
Level of Space and the Level of Actions) to coordinate movements chiefly concerned with the
goal of the motor skill. We consider the differences in generalization between recalibration
based on limb dynamics and recalibration based on kinematics of the visual event as
congruent with Bernstein’s concept of levels in motor skill organization. That is, the limited
generalization of recalibrations for altered limb dynamics suggests an organization in
background function (likely the formation of synergies). The pattern of generalization of
recalibrations for altered kinematics of the visual event suggests an organization in goal
relationship (likely as a configuration of space in action specific terms). Further studies
should address this more specifically and might also exploit on Reed’s (1982, 1996) concept
of Action Systems.
The situation of throwing on the rotating carousel is unusual in that it allows for examination
of recalibration based on limb dynamics and perceived projectile kinematics. Studies are
currently underway to examine differences in recalibration by means of tests for transfer. We
expect that recalibration for the Coriolis force transfers to other arm movements (whatever
their function) probably depending on the spatial and anatomical properties of trajectory
or turning by stepping in place (Rieser et al. 1995). A more recent study found transfer from
walking to crawling (Withagen and Michaels 2002). Furthermore, recalibration in rotational
locomotion (sidestepping in a turning fashion) did not transfer to forward walking, however
the new calibration of rotation transferred to arm movements to turn in place; participants
stood on a pivot and pulled themselves around by hands along a rail (Berry and Rieser 1999).
These findings reveal a spatial coordination in action specific terms and imply an action goal
relationship in movement organization.
The profound difference in generalization between adjustment based on limb dynamics and
adjustment based on visual event kinematics implies theoretical concepts of multiple
(hierarchical) levels in movement organization. Previously several authors presented such
ideas (Bernstein 1967, 1988/1947, 1996/1947; Greene 1972; Saltzman 1979; Schöner 1995).
Let us consider Bernstein’s ideas.
Bernstein suggested that the movements of a motor skill are organized over different levels.
Some of these levels (the Level of Tone and the Level of Muscular Articular Links) mainly
serve a background function for the organization of movements, allowing other levels (the
Level of Space and the Level of Actions) to coordinate movements chiefly concerned with the
goal of the motor skill. We consider the differences in generalization between recalibration
based on limb dynamics and recalibration based on kinematics of the visual event as
congruent with Bernstein’s concept of levels in motor skill organization. That is, the limited
generalization of recalibrations for altered limb dynamics suggests an organization in
background function (likely the formation of synergies). The pattern of generalization of
recalibrations for altered kinematics of the visual event suggests an organization in goal
relationship (likely as a configuration of space in action specific terms). Further studies
should address this more specifically and might also exploit on Reed’s (1982, 1996) concept
of Action Systems.
The situation of throwing on the rotating carousel is unusual in that it allows for examination
of recalibration based on limb dynamics and perceived projectile kinematics. Studies are
currently underway to examine differences in recalibration by means of tests for transfer. We
expect that recalibration for the Coriolis force transfers to other arm movements (whatever
their function) probably depending on the spatial and anatomical properties of trajectory
Page 24
RECALIBRATION OF THROWING Page 24 of 26
formation, whereas transfer of a recalibration for the visual event in throwing would follow a
goal relationship and therefore be restricted to hurling actions.
In conclusion, throwing on a rotating carousel creates a situation that requires adjustment in
direction to compensate for concurrently transformed limb dynamics and perceived projectile
kinematics. Our findings suggest that two independent components of recalibration occur.
These observations promote consideration of theoretical concepts on how changes in multiple
components of calibration are organized. Further empirical studies that address such a
question might eventually help us in better understanding how an agent achieves coordinated
motor skills.
ACKNOWLEDGEMENTS
We thank Jürgen Konczak and Michael Wade of the School of Kinesiology for providing
space for the conduct of this research. We gratefully acknowledge the stimulating discussions
of this research with Geoffrey Bingham and Jürgen Konczak. We also thank Esther Thelen
for suggestions which shaped Experiment 3, and Denise Hendriques for suggestions about the
format of Figure 3, and Paul Schrater for a discussion on the mechanism of the Coriolis force.
GRANTS
This research was partially supported by a grant from the National Science Foundation to the
University of Minnesota, NSF IIS-0121044, ITR/SY: Collaborative/RUI Research on the
Perceptual Aspects of Locomotion and by the Center for Cognitive Sciences of the University
of Minnesota.
formation, whereas transfer of a recalibration for the visual event in throwing would follow a
goal relationship and therefore be restricted to hurling actions.
In conclusion, throwing on a rotating carousel creates a situation that requires adjustment in
direction to compensate for concurrently transformed limb dynamics and perceived projectile
kinematics. Our findings suggest that two independent components of recalibration occur.
These observations promote consideration of theoretical concepts on how changes in multiple
components of calibration are organized. Further empirical studies that address such a
question might eventually help us in better understanding how an agent achieves coordinated
motor skills.
ACKNOWLEDGEMENTS
We thank Jürgen Konczak and Michael Wade of the School of Kinesiology for providing
space for the conduct of this research. We gratefully acknowledge the stimulating discussions
of this research with Geoffrey Bingham and Jürgen Konczak. We also thank Esther Thelen
for suggestions which shaped Experiment 3, and Denise Hendriques for suggestions about the
format of Figure 3, and Paul Schrater for a discussion on the mechanism of the Coriolis force.
GRANTS
This research was partially supported by a grant from the National Science Foundation to the
University of Minnesota, NSF IIS-0121044, ITR/SY: Collaborative/RUI Research on the
Perceptual Aspects of Locomotion and by the Center for Cognitive Sciences of the University
of Minnesota.
Page 25
RECALIBRATION OF THROWING Page 25 of 26
REFERENCES
Bernstein NA (1967) The coordination and regulation of movements. New York: Pergamon
Bernstein NA (1988/ Original work published 1947) Die Koordination der bewegungen in der ontogenese. In:
Bewegungsphysiologie, edited by Bernstein NA. Leipzig, Germany: Johann Ambrosius Barth, p. 99-137
Bernstein NA (1996/ Original work published 1947) On Dexterity and its development. In: Dexterity and its
development, edited by Latash ML, Turvey MT. Mahwah, NJ, USA: Lawrence Erlbaum Associates
Berry DM, Rieser JJ (1999) Changes in rotational locomotion transfer from foot to hand. Tenth International
Conference on Perception and Action (5 ed.), edited by Grealy MA, Thomson JA, Edinburgh, Scottland
Bingham G, Romack JL (1999) The rate of adaptation to displacement prisms remains constant despite
acquisition of rapid calibration. Journal of Experimental Psychology: Human Perception & Performance 25:
1331-1346
Bourdin C, Gauthier G, Blouin J, Vercher J-L (2001) Visual feedback of the moving arm allows complete
adaptation of pointing movements to centrifugal and Coriolis forces in human subjects. Neuroscience Letters
301: 25-28
Bruggeman H, Eid CL, Rieser JJ, Pick HL Jr. (2003) Alteration of the direction of throwing; Multiple levels of
adaptation. Journal of Vision, 3(9), 678a, DOI:10.1167/3.9.678.
Conditt MA, Gandolfo F, Mussa-Ivaldi FA (1997) The motor system does not learn the dynamics of the arm by
rote memorization of past experience. Journal of Neurophysiology 78: 554-560
Debicki DB, Gribble PL, Watts S, Hore J (2004) Kinematics of wrist joint flexion in overarm throws made by
skilled subjects. Experimental Brain Research 154: 382-394
DiZio P, Lackner JR (1995) Motor adaptation to Coriolis force perturbations of reaching movements: Endpoint
but not trajectory adaptation transfers to the nonexposed arm. Journal of neurophysiology 74: 1787-1792
Flanagan JR, Nakano E, Imamizu H, Osu R, Yoshioka T, Kawato M (1999) Composition and decomposition of
internal models in motor learning under altered kinematic and dynamic environments. J Neurosci 19: 34RC
Flanagan JR, Rao AK (1995) Trajectory adaptation to a nonlinear visuomotor transformation: evidence of
motion planning in visually perceived space. J Neurophysiol 74: 2174-2178
Gandolfo F, Mussa-Ivaldi FA, Bizzi E (1996a) Motor learning by field approximation. Neurobiology 93: 3843-
3846
Gandolfo F, Mussa-Ivaldi FA, Bizzi E (1996b) Motor learning by field approximation. Proceedings of the
National Academy of Sciences of the United States of America 93: 3843-3846
Goodbody SJ, Wolpert DM (1998) Temporal and amplitude generalization in motor learning. Journal of
Neurophysiology 79: 1825-1838
Greene PH (1972) Problems of organization of motor systems. In: Progress In Theoretical Biology, edited by
Rosen R, Snell F, M. New York, p. 303-338.
Jakobson LS, Goodale MA (1998) Trajectories of reaches to prismatically-displaced targets: evidence for
"automatic" visuomotor recalibration. Experimental Brain Research 78: 575-587
REFERENCES
Bernstein NA (1967) The coordination and regulation of movements. New York: Pergamon
Bernstein NA (1988/ Original work published 1947) Die Koordination der bewegungen in der ontogenese. In:
Bewegungsphysiologie, edited by Bernstein NA. Leipzig, Germany: Johann Ambrosius Barth, p. 99-137
Bernstein NA (1996/ Original work published 1947) On Dexterity and its development. In: Dexterity and its
development, edited by Latash ML, Turvey MT. Mahwah, NJ, USA: Lawrence Erlbaum Associates
Berry DM, Rieser JJ (1999) Changes in rotational locomotion transfer from foot to hand. Tenth International
Conference on Perception and Action (5 ed.), edited by Grealy MA, Thomson JA, Edinburgh, Scottland
Bingham G, Romack JL (1999) The rate of adaptation to displacement prisms remains constant despite
acquisition of rapid calibration. Journal of Experimental Psychology: Human Perception & Performance 25:
1331-1346
Bourdin C, Gauthier G, Blouin J, Vercher J-L (2001) Visual feedback of the moving arm allows complete
adaptation of pointing movements to centrifugal and Coriolis forces in human subjects. Neuroscience Letters
301: 25-28
Bruggeman H, Eid CL, Rieser JJ, Pick HL Jr. (2003) Alteration of the direction of throwing; Multiple levels of
adaptation. Journal of Vision, 3(9), 678a, DOI:10.1167/3.9.678.
Conditt MA, Gandolfo F, Mussa-Ivaldi FA (1997) The motor system does not learn the dynamics of the arm by
rote memorization of past experience. Journal of Neurophysiology 78: 554-560
Debicki DB, Gribble PL, Watts S, Hore J (2004) Kinematics of wrist joint flexion in overarm throws made by
skilled subjects. Experimental Brain Research 154: 382-394
DiZio P, Lackner JR (1995) Motor adaptation to Coriolis force perturbations of reaching movements: Endpoint
but not trajectory adaptation transfers to the nonexposed arm. Journal of neurophysiology 74: 1787-1792
Flanagan JR, Nakano E, Imamizu H, Osu R, Yoshioka T, Kawato M (1999) Composition and decomposition of
internal models in motor learning under altered kinematic and dynamic environments. J Neurosci 19: 34RC
Flanagan JR, Rao AK (1995) Trajectory adaptation to a nonlinear visuomotor transformation: evidence of
motion planning in visually perceived space. J Neurophysiol 74: 2174-2178
Gandolfo F, Mussa-Ivaldi FA, Bizzi E (1996a) Motor learning by field approximation. Neurobiology 93: 3843-
3846
Gandolfo F, Mussa-Ivaldi FA, Bizzi E (1996b) Motor learning by field approximation. Proceedings of the
National Academy of Sciences of the United States of America 93: 3843-3846
Goodbody SJ, Wolpert DM (1998) Temporal and amplitude generalization in motor learning. Journal of
Neurophysiology 79: 1825-1838
Greene PH (1972) Problems of organization of motor systems. In: Progress In Theoretical Biology, edited by
Rosen R, Snell F, M. New York, p. 303-338.
Jakobson LS, Goodale MA (1998) Trajectories of reaches to prismatically-displaced targets: evidence for
"automatic" visuomotor recalibration. Experimental Brain Research 78: 575-587
Page 26
RECALIBRATION OF THROWING Page 26 of 26
Krakauer JW, Ghilardi M-F, Ghez C (1999) Independent learning of internal models for kinematic and dynamic
control of reaching. Nature Neuroscience 2: 1026-1031
Lackner JR, Dizio P (1998) Gravitoinertial force background level affects adaptation to Coriolis force
perturbations of reaching movements. J Neurophysiol 80: 546-553
Lackner JR, DiZio P (1994) Rapid adaptation to Coriolis force perturbations of arm trajectory. Journal of
neurophysiology 72: 299-313
Martin TA, Keating JG, Goodkin HP, Bastian AJ, Thach WT (1996) Throwing while looking through prisms. II.
Specificity and storage of multiple gaze-throw calibrations. Brain 119: 1199-1211
Reed ES (1982) An outline of a theory of action systems. Journal of Motor Behavior 14: 98-134
Reed ES (1996) Encountering the world: Toward an ecological psychology. Oxford: Oxford University Press
Rieser JJ, Pick HL, Ashmead DH, Garing AE (1995) Calibration of human locomotion and models of
perceptual-motor organization. Journal of Experimental Psychology: Human Perception & Performance 21:
480-497
Saltzman E (1997) Levels of sensorimotor representation. Journal of Mathematical Psychology 20: 91-163
Schöner G (1995) Recent developments and problems in human movement science and their conceptual
implications. Ecological Psychology 7: 291-314
Shadmehr R, Mussa-Ivaldi F (1994) Adaptive representation of dynamics during learning of a motor task.
Journal of Neuroscience 14: 3208-3224
Tong C, Wolpert DM, Flanagan JR (2002) Kinematics and dynamics are not represented independently in motor
working memory: Evidence from an interference study. Journal of Neuroscience 22: 1108-1113
Welch RB, Bridgeman B, Anand S, Browman KE (1993) Alternating prism exposure causes dual adaptation and
generalization to a novel displacement. Perception & Psychophysics 54: 195-204
Withagen R, Michaels CF (2002) The calibration of walking transfers to crawling: Are action systems
calibrated? Ecological Psychology 14: 223-234
Krakauer JW, Ghilardi M-F, Ghez C (1999) Independent learning of internal models for kinematic and dynamic
control of reaching. Nature Neuroscience 2: 1026-1031
Lackner JR, Dizio P (1998) Gravitoinertial force background level affects adaptation to Coriolis force
perturbations of reaching movements. J Neurophysiol 80: 546-553
Lackner JR, DiZio P (1994) Rapid adaptation to Coriolis force perturbations of arm trajectory. Journal of
neurophysiology 72: 299-313
Martin TA, Keating JG, Goodkin HP, Bastian AJ, Thach WT (1996) Throwing while looking through prisms. II.
Specificity and storage of multiple gaze-throw calibrations. Brain 119: 1199-1211
Reed ES (1982) An outline of a theory of action systems. Journal of Motor Behavior 14: 98-134
Reed ES (1996) Encountering the world: Toward an ecological psychology. Oxford: Oxford University Press
Rieser JJ, Pick HL, Ashmead DH, Garing AE (1995) Calibration of human locomotion and models of
perceptual-motor organization. Journal of Experimental Psychology: Human Perception & Performance 21:
480-497
Saltzman E (1997) Levels of sensorimotor representation. Journal of Mathematical Psychology 20: 91-163
Schöner G (1995) Recent developments and problems in human movement science and their conceptual
implications. Ecological Psychology 7: 291-314
Shadmehr R, Mussa-Ivaldi F (1994) Adaptive representation of dynamics during learning of a motor task.
Journal of Neuroscience 14: 3208-3224
Tong C, Wolpert DM, Flanagan JR (2002) Kinematics and dynamics are not represented independently in motor
working memory: Evidence from an interference study. Journal of Neuroscience 22: 1108-1113
Welch RB, Bridgeman B, Anand S, Browman KE (1993) Alternating prism exposure causes dual adaptation and
generalization to a novel displacement. Perception & Psychophysics 54: 195-204
Withagen R, Michaels CF (2002) The calibration of walking transfers to crawling: Are action systems
calibrated? Ecological Psychology 14: 223-234
Sign up today - FREE
Mendeley saves you time finding and organizing research. Learn more
- All your research in one place
- Add and import papers easily
- Access it anywhere, anytime
Start using Mendeley in seconds!
Readership Statistics
9 Readers on Mendeley
by Discipline
78% Psychology
by Academic Status
44% Post Doc
22% Associate Professor
11% Ph.D. Student
by Country
56% United States
22% Japan
11% Canada