Professional Documents
Culture Documents
Cognition
journal homepage: www.elsevier.com/locate/COGNIT
a r t i c l e
i n f o
Article history:
Received 5 June 2013
Revised 4 February 2014
Accepted 12 February 2014
Available online 15 March 2014
Keywords:
Self-monitoring
Chimpanzees
Agency
Voluntary action
Imitation
Mirror system
a b s t r a c t
Humans are often unaware of how they control their limb motor movements. People pay
attention to their own motor movements only when their usual motor routines encounter
errors. Yet little is known about the extent to which voluntary actions rely on automatic
control and when automatic control shifts to deliberate control in nonhuman primates.
In this study, we demonstrate that chimpanzees and humans showed similar limb motor
adjustment in response to feedback error during reaching actions, whereas attentional allocation inferred from gaze behavior differed. We found that humans shifted attention to
their own motor kinematics as errors were induced in motor trajectory feedback regardless
of whether the errors actually disrupted their reaching their action goals. In contrast, chimpanzees shifted attention to motor execution only when errors actually interfered with
their achieving a planned action goal. These results indicate that the species differed in
their criteria for shifting from automatic to deliberate control of motor actions. It is widely
accepted that sophisticated motor repertoires have evolved in humans. Our results suggest
that the deliberate monitoring of ones own motor kinematics may have evolved in the
human lineage.
2014 Elsevier B.V. All rights reserved.
1. Introduction
To what extent do nonhuman primates intentionally
control their motor actions? This is an important question
for understanding the evolutionary origin of intentional
and deliberate control of motor actions. However, experimental studies comparing humans and nonhuman primates in this regard are scarce.
Humans automatically adjust their limb movements
according to the degree of discrepancy between the predicted results of actions and actual feedback, even when
they are unaware of the error (Fourneret & Jeannerod,
1998; Knoblich & Kircher, 2004; Musseler & Sutter,
2009). For example, when one moves a cursor to an icon
Corresponding author. Current Address: Department of Psychology,
Kyoto University, Yoshida-Honmachi, Kyoto 606-8501, Japan. Tel.: +81 75
753 2442.
E-mail address: kaneko.takaaki.22m@st.kyoto-u.ac.jp (T. Kaneko).
http://dx.doi.org/10.1016/j.cognition.2014.02.002
0010-0277/ 2014 Elsevier B.V. All rights reserved.
356
357
(a)
(b)
Goal Dissociation
Goal Congruent
Altered
action
Start
Angular
rotation
Default
action
Stimulus
onset
(c)
P
P
Hit
D
L
Fig. 1. Schematic diagram of the behavioral task and design. (a) Task. The trial required the subject to hit three different locations of a target consecutively.
Here, only one target is shown for simplicity. (b) Test conditions. Two types of perturbations were used. The typical time course of angular rotation is shown
for each condition. (c) Schematic illustration of the goal-congruent distortion algorithm. The rectangle and circle represent the target and the cursor,
respectively. The dotted circle at left represents the initial location of the cursor. The dotted circle in the middle shows the invisible cursor (see Section 2 for
details).
(a)
(b)
2
4
6
7
3
1
2.2. Apparatus
2. Methods
2.3. Task
2.1. Participants
Five chimpanzees living at the Primate Research Institute, Kyoto University, Japan, took part in the experiment
358
2.6. Analysis
As a measurement of limb motor adjustment in response to feedback errors, we dened the corrective movement as shown in Fig. 4a. The vector input of the
participants manipulation was divided into vertical and
horizontal components relative to the axis between the
target and the initial cursor location. At the baseline trial,
in which no distortion was imposed, the vertical component of the vector input was unnecessary but was induced
under the perturbation as a response to feedback error
during the test trial. Thus, we dened the total vertical
component as a corrective movement. Fixations were extracted from raw gaze-position data using a custom-made
xation lter based on Stampe (1993). A xation was
scored if the gaze remained stationary within a radius of
1 in visual angle for at least 50 ms. We calculated the
number of gaze shifts to the motor trajectory as an indicator of times when the feedback error triggered participants overt attention to motor corrections. Gaze shift
was dened as xation within a 1 visual angle from the
cursor location. We also recorded the movement time,
which was dened as time between the onset of stimulus
and hitting the target.
Statistical analyses were performed based on the functions of the statistical toolbox (ttest for t-test and anovan
for analysis of variance [ANOVA]) in MATLAB (R2011a).
Post hoc comparisons were conducted with Bonferroni correction. Alpha was set at .05 for all analyses, and the Bonferroni correction adjusted the alpha level as necessary.
Means of multiple trials for each subject were computed
for each behavioral measurement, and those values were
used as dependent variable in the statistical tests.
2.7. Expertise effect
We addressed possible artifactual effects of the difference in expertise between species. There were two types
of asymmetry between the two species in the degree of
expertise. The rst was experience with our experimental
device. The chimpanzee participants had previously performed several experiments using the current device, but
the humans had not. The second was the number of distortion trials the participants experienced. The chimpanzee
participants performed 32 trials per condition, and the humans performed only ve trials. Two controls were used
for these expertise asymmetries. First, we asked two
human participants (one female and one male; age, 27
28 years) to perform more than 1,200 aiming actions over
50 days (Fig. S1) until their movement performance became very stable. Additionally, the same two participants
performed 32 trials per condition, like the chimpanzees.
3. Results
The descriptive results of the participants overall
behavior are shown in Fig. 3. The results show that both
the chimpanzee and the human participants mainly xated
on the goal of the action and occasionally looked at cursor
trajectory during aiming. This trend was consistent across
359
species and from the baseline trial to the test trials involving distortion. The xation probability maps (color maps in
the rst row of Fig. 3) showed that there were roughly
three different color regions, i.e., gray, yellow, and green
areas, suggesting that the participants predominantly
looked at the target location (corresponding to the green
areas), irrespective of distortion type, and that they occasionally looked at the trajectory of the cursor (corresponding to the yellow areas). This trend is also evident in the
second row of Fig. 3, which shows the xation probability
as a function of the target and the start location axis. Target
location was the most frequent target of xation for both
humans and chimpanzees. The characteristics of the corrective movements were also similar between the species
(bottom row in Fig. 3). Corrective movements appeared
gradually as the cursor moved under both the goal-dissociation and goal-congruent conditions. Under the goal-dissociation condition, such corrective movements further
accumulated at the latter part of the trajectory. The polarity of corrective movement was reversed at the mid-point
of the trajectory during the goal-congruent condition. This
is because corrective movement was not required at all under this condition, and the corrective movements performed until the mid-part of the trajectory must be
compensated by movement in the opposite direction.
These patterns in the corrective movements appeared
comparable between species. In the latter analysis, we further analyzed the details of the corrective movement and
how the needs for these motor corrections induced gaze
direction toward the motor trajectory.
Both species effectively corrected their limb movements when distortion was applied (Fig. 5). The results
show that both species increased their corrective movements as distortion increased, irrespective of distortion
type. We performed a two-way ANOVA with distortion
type and distortion amplitude as xed factors and participant as a random factor for each species. The overall statistical results for corrective movement were comparable
between species. The ANOVAs showed signicant interactions
of
distortion
type distortion
amplitude
(F(3, 12) = 58.7, p < .001, gp2 = .94 for chimpanzees;
F(3, 18) = 33.2, p < .001, gp2 = .85 for humans). Post hoc
comparisons of distortion type for each distortion amplitude with Bonferroni correction revealed that the corrective movements were the same between the distortion
types when the distortion amplitude was small
(t(4) = 1.8, p = .15, r = .67, for chimpanzees at 6;
t(6) = 1.6, p = .17, r = .54, t(6) = 1.6, p = .16, r = .55 for humans at 6 and 12, respectively) but were larger under
the goal-dissociation condition than the goal-congruent
condition when the distortion amplitudes were large
(t(4) = 3.1, p = .04, r = .84, t(4) = 6.1, p = .004, r = .95,
t(4) = 8.1, p = .001, r = .97 for chimpanzees at 12, 28, and
46, respectively; t(6) = 8.4, p = .0002, r = .96; t(6) = 7.2,
p = .0004, r = .95 for humans at 28 and 46, respectively).
The main effects of distortion type (F(1, 12) = 57.8,
p = .002, gp2 = .83 for chimpanzees; F(1, 18) = 68.4,
p < .001, gp2 = .79 for humans) and distortion amplitude
(F(3, 12) = 510.7, p < .001, gp2 = .99 for chimpanzees;
F(3, 18) = 130.6, p < .001, gp2 = .96 for humans) were also
signicant. These results show that both species increased
360
(a) Chimpanzees
Fixation propertion
(b) Humans
Fixation propertion
Fig. 3. Cursor action, hand manipulation, and gaze behavior. The three columns represent the baseline, goal-congruent, and goal-dissociation conditions,
respectively. The rst row shows cursor trajectory and xation probability. The solid line represents cursor trajectory, the left terminus is the start position,
and the right terminus is the target position. The yellowgreen color represents xation probability. The second and third rows represent xation
probability and corrective movements, respectively, as a function of the horizontal distance from the start position on the axis and the target. The dotted
vertical lines represent the start positions (left, 0 mm) and target locations (right, 150 mm). Data are given as the mean across participants. Shaded areas
represent 95% condence intervals. Results for the goal-congruent and goal-dissociation conditions were derived from the maximum degree of distortion,
which shows the most salient distortion effect on the participants behavior. Note that the direction of distortion and the directions from the start position
to the target varied across trials; thus, the results were realigned so that relative geometry was constant across trials. (For interpretation of the references to
color in this gure legend, the reader is referred to the web version of this article.)
between chimpanzees and humans. We performed an ANOVA for each species with distortion type and distortion
amplitude as xed factors and participant as a random factor for each species. The ANOVAs showed a signicant
interaction of distortion type and distortion amplitude in
chimpanzees (F(3, 12) = 15.2, p < .001, gp2 = .79) but not
in humans (F(3, 18) = 0.17, p = .91, gp2 = .028). Post hoc
analyses of the simple main effects were conducted for
each distortion type for the chimpanzee results. The
chimpanzees increased their gaze-shifting response as
distortion increased under the goal-dissociation condition
(F(3, 12) = 10.96, p = .0009, gp2 = .73) but not under the
goal-congruent
condition
(F(3, 12) = 2.26,
p = .13,
gp2 = .36). The main effect of distortion amplitude was
361
(b)
Horizontal component
Vector input
Vertical component
=Corrective movement
Goal congruent
Chimpanzees
Humans
600
(a)
500
400
300
200
100
0
-100
12
28
600
500
400
300
200
100
0
-100
46
Distortion (degree)
12
28
46
Distortion (degree)
Fig. 4. Motor adjustment in response to feedback error. (a) Schematic illustration of the denition of corrective movements. (b) Mean corrective
movements across participants as a function of distortion amplitude. The gure shows values for each test condition subtracted from values for the baseline
condition. Error bars indicate standard errors. Dotted red lines represent the minimal corrective movements required for the goal-dissociation condition.
The minimal correction movement for the goal-congruent condition was zero regardless of distortion amplitude. (For interpretation of the references to
color in this gure legend, the reader is referred to the web version of this article.)
(a)
(b)
Goal dissociation
Chimpanzees
Altered action
Default action
2
1.5
1
0.5
0
-0.5
12
28
Humans
2.5
Gaze shift
2.5
46
Distortion (degree)
Goal congruent
2
1.5
1
0.5
0
-0.5
12
28
46
Distortion (degree)
Fig. 5. Overt attention to motor kinematics induced by feedback error. (a) Gaze shift was dened as xation within a 1 visual angle from the cursor
location. (b) Mean gaze shift across participants as a function of distortion amplitude. The gure shows the values for each test condition subtracted from
the values for the baseline condition. Error bars indicate standard errors.
species (Fig. 6), and differed from that of gaze shifts. Both
species increased movement time as distortion amplitude
increased (F(3, 12) = 112.8, p < .001, gp2 = .97 for chimpanzees; F(3, 18) = 72.8, p < .001, gp2 = .92 for humans). The
degree of incentive was larger under the goal-dissociation
than under the goal-congruent condition, which is evident
in the signicant interaction between distortion type and
distortion amplitude (F(3, 12) = 66.2, p < .001, gp2 = .94 for
chimpanzees; F(3, 18) = 7.10, p = .002, gp2 = .54 for humans). A post hoc analysis showed that movement time
was not different between distortion types (t(4) = 1.07,
p = .34, r = .48, t(4) = 1.29, p = .27, r = .54 , t(4) = 3.38,
p = .028, r = .86 for chimpanzees; t(6) = 1.55, p = .17,
r = .53 , t(6) = .29, p = .78, r = .12 , t(6) = .85, p = .43, r = .33
for humans at 6, 12, and 28 respectively), except that
for the largest distortion amplitude. Under the largest distortion, the movement time was longer under the goal-dissociation than under the goal-congruent condition
362
Goal dissociation
Chimpanzees
Humans
2500
2500
1750
1000
250
-500
Goal congruent
12
28
46
1750
1000
250
-500
Distortion (degree)
12
28
46
Distortion (degree)
Fig. 6. Movement time under perturbation. Mean movement time across participants as a function of distortion amplitude. The gure shows the values for
each test condition subtracted from the values for the baseline condition. Error bars indicate standard errors.
4. Discussion
We examined the difference between chimpanzees and
humans in their reliance on automatic motor control and
investigated when and to what extent they shifted their
attention from the goal to their own motor kinematics.
We found that the two species were similar with respect
to online motor correction in response to feedback error,
whereas the attentional allocation for that motor adjustment differed. Humans were sensitive to both goal and trajectory information about gaze shift, whereas chimpanzees
were mainly sensitive to goal information. Our results suggest that chimpanzees have a more parsimonious cognitive
system in that they relied on an automatic process for online motor adjustment as long as they were able to reach
the planned action goal. The motor system recruited their
deliberate attention only when they could not reach the
goal. In contrast, humans appeared to have more complex
criteria for shifting to deliberate control of their action and
exerted excess attentional monitoring over their own motor kinematics.
One may argue that because the chimpanzees were
interested only in the food reward and thus lacked motivation to move the cursor precisely to hit the target, they did
not pay attention to the detailed motor kinematics required to reach the target. However, this is unlikely, because the chimpanzees showed a certain amount of
corrective movement under the goal-congruent condition.
No corrective movement was necessary to obtain a reward
363
364
365
Musseler, J., & Sutter, C. (2009). Perceiving ones own movements when
using a tool. Consciousness and Cognition, 18(2), 359365.
Myowa-Yamakoshi, M., Kawakita, Y., Okanda, M., & Takeshita, H. (2011).
Visual experience inuences 12-month-old infants perception of
goal-directed actions of others. Developmental Psychology, 47(4),
10421049.
Myowa-Yamakoshi, M., & Matsuzawa, T. (1999). Factors inuencing
imitation of manipulatory actions in chimpanzees (Pan troglodytes).
Journal of Comparative Psychology, 113(2), 128136.
Myowa-Yamakoshi, M., & Matsuzawa, T. (2000). Imitation of intentional
manipulatory actions in chimpanzees (Pan troglodytes). Journal of
Comparative Psychology, 114(4), 381391.
Nagell, K., Olguin, R. S., & Tomasello, M. (1993). Processes of sociallearning in the tool use of chimpanzees (Pan troglodytes) and human
children (Homo sapiens). Journal of Comparative Psychology, 107(2),
174186.
Nakajima, K., Maier, M. A., Kirkwood, P. A., & Lemon, R. N. (2000). Striking
differences in transmission of corticospinal excitation to upper limb
motoneurons in two primate species. Journal of Neurophysiology,
84(2), 698709.
Nakayama, Y., Yamagata, T., Tanji, J., & Hoshi, E. (2008). Transformation of
a virtual action plan into a motor plan in the premotor cortex. The
Journal of Neuroscience, 28(41), 1028710297.
Nelissen, K., Luppino, G., Vanduffel, W., Rizzolatti, G., & Orban, G. A.
(2005). Observing others: Multiple action representation in the
frontal lobe. Science, 310(5746), 332336.
Nielsen, M. (2006). Copying actions and copying outcomes: Social
learning through the second year. Developmental Psychology, 42(3),
555565.
Nissen, M. J., & Bullemer, P. (1987). Attentional requirements of learning
Evidence from performance-measures. Cognitive Psychology, 19(1),
132.
Norman, D., & Shallice, T. (1986). Attention to action: Willed and
automatic control of behavior. In R. Davidson, R. Swartz, & D.
Shapiro (Eds.), Consciousness and self-regulation: Advances in research
and theory IV. Plenum Press.
Peeters, R., Simone, L., Nelissen, K., Fabbri-Destro, M., Vanduffel, W.,
Rizzolatti, G., et al. (2009). The representation of tool use in humans
and monkeys: Common and uniquely human features. The Journal of
Neuroscience, 29(37), 1152311539.
Pisella, L., Grea, H., Tilikete, C., Vighetto, A., Desmurget, M., Rode, G., et al.
(2000). An automatic pilot for the hand in human posterior parietal
cortex: Toward reinterpreting optic ataxia. Nature Neuroscience, 3(7),
729736.
Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron system. Annual
Review of Neuroscience, 27, 169192.
Rizzolatti, G., Fogassi, L., & Gallese, V. (2001). Neurophysiological
mechanisms underlying the understanding and imitation of action.
Nature Reviews Neuroscience, 2(9), 661670.
Sailer, U., Flanagan, J. R., & Johansson, R. S. (2005). Eye-hand coordination
during learning of a novel visuomotor task. Journal of Neuroscience,
25(39), 88338842.
Sommerville, J. A., Woodward, A. L., & Needham, A. (2005). Action
experience alters 3-month-old infants perception of others actions.
Cognition, 96(1), B111.
Stampe, D. (1993). Heuristic ltering and reliable calibration methods for
video-based pupil-tracking systems. Behavior Research Methods, 25(2),
137142.
Tabachnick, B. G., & Fidell, L. S. (2001). Computer-assisted research design
and analysis. Boston: Allyn and Bacon.
Tanji, J., & Hoshi, E. (2008). Role of the lateral prefrontal cortex in
executive behavioral control. Physiological Reviews, 88(1), 3757.
Taylor, J. A., & Thoroughman, K. A. (2007). Divided attention impairs
human motor adaptation but not feedback control. Journal of
Neurophysiology, 98(1), 317326.
Tennie, C., Call, J., & Tomasello, M. (2006). Push or pull: Imitation vs.
emulation in great apes and human children. Ethology, 112(12),
11591169.
Tomasello, M., Davis-Dasilva, M., Camak, L., & Bard, K. (1987).
Observational learning of tool-use by young chimpanzees. Human
Evolution, 2(2), 175183.
Tomonaga, M. (2001). Investigating visual perception and cognition in
chimpanzees (Pan troglodytes) through visual search and related
tasks: From basic to complex processes. In T. Matsuzawa (Ed.),
Primate origins of human cognition and behavior (pp. 5586). Tokyo,
Japan: Springer.
Vanduffel, W., Fize, D., Peuskens, H., Denys, K., Sunaert, S., Todd, J. T., et al.
(2002). Extracting 3D from motion: Differences in human and
monkey intraparietal cortex. Science, 298(5592), 413415.
366
Whiten, A., Custance, D. M., Gomez, J. C., Teixidor, P., & Bard, K. A. (1996).
Imitative learning of articial fruit processing in children (Homo
sapiens) and chimpanzees (Pan troglodytes). Journal of Comparative
Psychology, 110(1), 314.
Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V., Sugiyama,
Y., et al. (1999). Cultures in chimpanzees. Nature, 399(6737),
682685.
Willingham, D. B. (1998). A neuropsychological theory of motor skill
learning. Psychological Review, 105(3), 558584.
Wolpert, D. M., Ghahramani, Z., & Jordan, M. I. (1995). Are arm trajectories
planned in kinematic or dynamic coordinates? An adaptive study.
Experimental Brain Research, 103(3), 460470.
Yamagata, T., Nakayama, Y., Tanji, J., & Hoshi, E. (2009). Processing
of visual signals for direct specication of motor targets and
for conceptual representation of action targets in the dorsal and
ventral premotor cortex. Journal of Neurophysiology, 102(6),
32803294.