Spatial Perspective Taking Mediated by Whole-Body Motor Simulation

Abstract

Humans can envision the world from other people's viewpoints. To explore the embodied process of such spatial perspective taking, we examined whether action related to a whole-body movement modulates performance on spatial perspective-taking tasks. Results showed that when participants responded by putting their left/right foot or left/right hand forward, actions congruent with a movement's direction (clockwise/counterclockwise) reduced RTs relative to incongruent actions. In contrast, actions irrelevant to a movement (a left/right hand index-finger response) did not affect performance. Furthermore, we demonstrated that this response congruency effect cannot be explained by either spatial stimulus-response compatibility or sensorimotor interference. These results support the involvement of simulated whole-body movement in spatial perspective taking. Moreover, the findings revealed faster foot responses than hand responses during spatial perspective taking, whereas the opposite result was obtained during a simple orientation judgment task without spatial perspective taking. Overall, our findings highlight the important role of motor simulation in spatial perspective taking.

Full text

1
Title:
Spatial perspective taking mediated by whole-body
motor simulation
Author names and affiliations:
Hiroyuki Muto*, Soyogu Matsushita, Kazunori Morikawa
School of Human Sciences, Osaka University, Japan (Address: School of Human Sciences,
Osaka University, 1-2 Yamadaoka, Suita-shi 565-0871, Japan)
Corresponding author:
Hiroyuki Muto. Address: School of Human Sciences, Osaka University, 1-2 Yamadaoka,
Suita-shi 565-0871, Japan. E-mail: h_muto@hus.osaka-u.ac.jp (H. Muto).
Word count: 13,083 words
© 2017, American Psychological Association. This paper is not the copy of record and may
not exactly replicate the final, authoritative version of the article. Please do not copy or cite
without authors permission. The final article will be available, upon publication, via its DOI:
10.1037/xhp0000464

2
Abstract Humans can envision the world from other people’s viewpoints. To explore the embodied
process of such spatial perspective taking, we examined whether action related to a whole-body
movement modulates performance on spatial perspective-taking tasks. Results showed that when
participants responded by putting their left/right foot or left/right hand forward, actions congruent
with a movement’s direction (clockwise/counterclockwise) reduced reaction times relative to
incongruent actions. In contrast, actions irrelevant to a movement (a left/right hand index-finger
response) did not affect performance. Furthermore, we demonstrated that this response congruency
effect cannot be explained by either spatial stimulus-response compatibility or sensorimotor
interference. These results support the involvement of simulated whole-body movement in spatial
perspective taking. Moreover, the findings revealed faster foot responses than hand responses during
spatial perspective taking, whereas the opposite result was obtained during a simple orientation
judgment task without spatial perspective taking. Overall, our findings highlight the important role of
motor simulation in spatial perspective taking.
Keywords: Spatial perspective taking, Embodied cognition, Spatial cognition, Mental rotation, Motor
simulation.
Statement of Public Significance Spatial perspective taking is a human ability to envision the world
from other people's viewpoints. Our five behavioral experiments show that during spatial perspective
taking, people mentally simulate whole-body movement as if they moved to a position from which
they took a new perspective. Specifically, we demonstrated that actions congruent with a movement's
direction facilitated spatial perspective taking compared to incongruent actions. This response
congruency effect was observed only when the action was relevant to whole-body movement.
Furthermore, we also demonstrated that foot responses were faster than hand responses for spatial
perspective taking although hand responses were faster than foot responses for a task for which spatial
perspective taking was unnecessary. These findings highlight the important role of motor processing
in spatial perspective taking, suggesting that spatial cognition is closely related to bodily movement.

3
1. Introduction
Humans are capable of understanding the world from other people’s viewpoints, e.g., you
can ask a friend to pass you a glass on his/her right side, even when the glass is not on the
right side from your perspective. This type of spatial problem can be solved readily or even
sometimes automatically (Tversky & Hard, 2009); however, other primates seem to be
incapable of such spatial perspective taking
1
(e.g., Tomasello, Carpenter, Call, Behne, &
Moll, 2005). Previous studies have shown that spatial perspective-taking ability relates
closely to a variety of other important abilities, such as navigation (Wolbers & Hegarty, 2010,
for a review), theory of mind (Hamilton, Brindley, & Frith, 2009), and empathic perspective
taking (Erle & Topolinski, 2015). However, cognitive processes underlying spatial
perspective taking have not yet been adequately elucidated. The present study addresses this
issue using an embodied cognition approach.
1.1. Object-based and perspective transformations
Pioneering studies on spatial perspective taking by Presson et al. focused on comparing
object-based and perspective transformations (e.g., Huttenlocher & Presson, 1973, 1979;
Presson, 1982). While object-based transformations refer to operating a mental image of an
object or an array, perspective transformations refer to operating a mental image of the self,
1
Previous studies (e.g., Surtees, Apperly, & Samson, 2013) have reported two forms of spatial perspective
taking; one is related to an understanding of whether another person can see a particular object (e.g.,
visibility or front/behind judgments) and the other is related to an understanding of where an object is
located from another person’s viewpoint (e.g., left/right judgments). Because the former can be performed
by drawing a line between another person and an object (Kessler & Rutherford, 2010; Michelon & Zacks,
2006; Surtees et al., 2013) and thus does not require perspective transformations, we focus only on the
latter form, referring to it as “spatial perspective taking” for convenience.

4
and spatial perspective taking has been assumed to be one form of perspective
transformations (see Zacks & Michelon, 2005, for a review). Presson et al. found that the two
transformations were processed differently.
The most studied object-based transformation is mental rotation of an object. In the initial
experiment of Shepard and Metzler (1971), participants were presented a pair of two pictorial
three-dimensional objects comprising cubes and were asked to respond as quickly as possible
as to whether the two objects were the same or different. Results showed that response times
(RTs) for same-different judgments increased linearly with the angular disparity between the
two objects. This suggested that mental imagery can be rotated just like a real object.
Analogous to the mental rotation of an object, perspective transformations have been
extensively studied in terms of mental rotation of the self or viewer rotation (e.g., Amorim &
Stucchi, 1997; Carpenter & Proffitt, 2001; Creem, Downs, Wraga, Harrington, Proffitt, &
Downs, 2001; Creem, Wraga, & Proffitt, 2001; Lambrey, Doeller, Berthoz, & Burgess, 2012;
Wraga, Creem, & Proffitt, 2000). Most previous research has shown that performance (i.e.,
speed or accuracy) on both kinds of mental transformation are impaired with increasing
angles of rotation; this implies the existence of mental spatial transformations analogous to
physical ones.
Regarding different mental spatial transformations, Zacks et al. proposed a multiple
systems framework (e.g., Zacks & Michelon, 2005; Zacks & Tversky, 2005). This framework
assumes that the two forms of mental spatial transformations are implemented to some degree
by distinct neural substrates, which are hypothesized to have been shaped by natural selection.
This means that unique neural and cognitive mechanisms underlie each form of
transformation, and they lead to unique physiological or behavioral consequences. Several
empirical studies have provided evidence for the multiple systems framework. For example,
some studies have shown that object rotation and viewer rotation depend on different neural

5
structures (Lambrey et al., 2012; Wraga, Shephard, Church, Inati, & Kosslyn, 2005; Zacks,
Vettel, & Michelon, 2003), and, in fact, viewer rotation can usually be performed more
efficiently than object or array rotation (e.g., Amorim & Stucchi, 1997; Presson, 1982; Wraga
et al., 2000), particularly when the rotational axis is perpendicular to the horizontal plane
(Carpenter & Proffitt, 2001; Creem, Wraga, & Proffitt, 2001). Furthermore, humans can
select an appropriate transformation for a given situation, and instructions to use an
inappropriate transformation adversely affect task performance (Zacks & Tverskey, 2005).
Other studies have shown that psychometric tests can measure abilities related to each
transformation as two separable factors (Hegarty & Waller, 2004; Kozhevnikov & Hegarty,
2001) and that the ability of object-based transformation develops earlier in childhood and
declines with age later than that of perspective transformation (Huttenlocher & Presson,
1973; Inagaki, Meguro, Shimada, Ishizaki, Okuzumi, & Yamadori, 2002). These findings are
all consistent with the multiple systems framework.
1.2. Spatial perspective taking as a perspective transformation
Thus far, spatial perspective taking has been naively (or perhaps implicitly) thought of as a
form of perspective transformation because the results of typical experiments on spatial
updating or perspective change have shown monotonic increases in RT or error with the
rotational angle (e.g., Easton & Sholl, 1995; Rieser, 1989). However, some researchers have
proposed a different interpretation of the angle effect in terms of sensorimotor interference
(e.g., Brockmole & Wang, 2003; May, 2004; Wang, 2005). According to this account,
impaired performance associated with an angle is attributed not to an additional cognitive
effort of mental transformations but to interference conflict between real and imagined
perspectives. For example, May (2004) provided empirical evidence favoring the
sensorimotor interference account. He compared angle effects of self-translation and

6
self-rotation while controlling for the amount of angular disparity between real and imagined
perspectives. In the self-translation condition, efforts of mental transformations were the
same regardless of angular disparities because the distance between real and imagined
positions was constant. Thus, if the transformation was needed, the angle effect would appear
only in the self-rotation condition. However, results showed monotonic increases of RT and
error as a function of angular disparity for both translation and rotation conditions.
Furthermore, the angle effect was observed even when extra time was given so that
participants could complete, if any, a mental transformation in advance (May, 2004; Wang,
2005). These findings seem to contradict the transformation account.
Nonetheless, these findings do not necessarily deny the transformation account. First, as
indicated by Kessler and Thomson (2010), tasks employed by May (2004) and Wang (2005)
imposed a heavy cognitive load on working memory. During their tasks, participants had to
maintain simultaneously a complicated array of four or five objects and the self’s updated
location. This might have motivated participants to use another strategy (e.g., simply wait and
do nothing during the extra time) against researchers’ expectations. Second, most previous
research on perspective change has employed a task that can be solved largely based on
knowledge from long-term memory, for example, a previously remembered array (e.g., May,
2004; Wang, 2005) or a familiar environment (e.g., Brockmole & Wang, 2003). Such
knowledge-based offline processes might be helpful in some situations, such as route
planning or giving navigational directions.
However, online processes are also important for real-life spatial problem solving. In fact,
many spatial problems in daily life are solved by real-time processing rather than a priori
knowledge because of limited time, lack of knowledge or cognitive tools for using it, or
difficulty in the knowledge-level solution (Freksa & Schultheis, 2014). In addition, most
studies on object-based transformations have employed tasks requiring real-time processes

7
(e.g., Shepard & Metzler, 1971). Perhaps online processes require a more concrete strategy
(e.g., mental transformation) than offline processes that might prompt a more abstract
strategy (e.g., calculation or verbal thought). Consistent with this view, Kessler and Thomson
(2010) provided evidence that spatial perspective taking involves “embodied”
transformations using a task that emphasized real-time processing (described in detail in the
following section). To elucidate cognitive processes of spatial perspective taking as a mental
transformation, the present study also focuses on online processes.
1.3. Embodiment in spatial perspective taking
Given that human evolution covers less than 1% of the entire evolutionary history of life
on Earth, high-level cognitive functions unique to humans are likely based largely on
primitive functions such as motor processing (Waller, 2014). In other words, cognition is
embodied. Approaches based on such embodied cognition have thus far revealed that mental
object rotation is closely related to physical hand movements. For example, concurrent
rotational hand movements facilitate or inhibit mental object rotation when they are
congruent or incongruent with the direction of mental rotation, respectively (Wexler, Kosslyn,
& Berthoz, 1998; Wohlschläger & Wohlschläger, 1998); same-different judgments via mental
and physical rotations yield a similar RT pattern (Gardony, Taylor, & Brunyé, 2014;
Wohlschläger & Wohlschläger, 1998); and objects difficult to move physically by hand are
also difficult to move in mental imagery (Flusberg & Boroditsky, 2011). These findings
suggest shared processing between mental object rotation and motor simulation of hand
movements; this has been corroborated by neuroimaging studies’ reports of brain activities in
motor regions (Zacks, 2008, for a meta-analysis and review).
Less attention, however, has been paid to the kinesthetic aspects of spatial perspective
taking, but Kessler and Thomson (2010) introduced a promising new approach. They used a

8
round-table stimulus on which two objects (a gun and a flower) were laid in front of a sitting
avatar (Experiments 1 and 4) or an empty chair (Experiment 2). Participants were asked to
judge the position (left or right) of a target object indicated in advance from the avatar or the
chair’s perspective. Consistent with other studies, results showed monotonically increasing
RTs with angular disparity between participants’ actual and imagined perspectives.
Ingeniously, Kessler and Thomson (2010) also manipulated the actual orientation/posture of
participants’ bodies in a clockwise or counterclockwise direction. They found that body
posture congruent with an imagined movement’s direction facilitated spatial perspective
taking compared to straight body posture (baseline), and incongruent body posture hindered
spatial perspective taking compared to baseline. This posture congruency effect could not be
accounted for by the angle difference between a body orientation and an imagined
perspective; it thus contradicted the sensorimotor interference account. Instead, the posture
congruency effect depended on whether body posture was congruent or incongruent with the
imagined movement direction. Therefore, Kessler and Thomson (2010) concluded the
existence of embodied transformation. Interestingly, the posture congruency effect
disappeared in a comparable task that required object-based transformations instead of
perspective transformations, suggesting the involvement of a whole-body schema in spatial
perspective taking, not that of a specific body part (i.e., hand) as in mental object rotation
(Experiment 3 in Kessler & Thomson, 2010).
Although Kessler and Thomson (2010) elegantly demonstrated that spatial perspective
taking is embodied in simulated movements, its underlying mechanism remains unclear. For
example, they claimed that a whole-body schema was involved in spatial perspective taking,
which has yet to be proven because their manipulation of participants’ body posture could
affect representations of both a whole-body and specific body parts (i.e., turning the
whole-body orientation also altered the position of the arms and legs). To confirm the

9
involvement of the whole-body schema, we have to manipulate different body parts (e.g., feet
and hands) separately.
It also remains unclear whether actions related to a whole-body movement affect the
spatial perspective taking. A number of previous studies demonstrated the involvement of
motor simulation in various tasks such as mental object rotation (e.g., Schwartz & Holton,
2000; Wexler, Kosslyn, & Berthoz, 1998; Wohlschläger & Wohlschläger, 1998) and imagined
locomotion (e.g., Kunz, Creem-Regehr, & Thompson, 2009) by examining the effect of
concurrent physical action on the performance. If spatial perspective taking involves motor
simulation of a whole-body movement, it should be affected only by actions related to a
whole-body movement. Thus, the effect of actions would be a more direct evidence of
simulated whole-body movement than the posture effect (Kessler & Thomson, 2010).
Although some neuroimaging studies have reported activations of brain regions associated
with motor processing during perspective transformations (Creem, Downs, et al., 2001;
Wraga et al., 2005; Schwabe, Lenggenhager, & Blanke, 2009), very little behavioral data
exist to help interpret such neuroscientific findings. This has led to controversy regarding the
involvement of motor simulation in spatial perspective taking (e.g., Wraga et al., 2005). To
dissipate this controversy, we need behavioral studies that examine the effect of actions.
1.4. The present study
To determine whether simulated whole-body movement shares a common process with
spatial perspective taking, the present study manipulated a response method in which
participants indicated their judgments about the position (left or right) of a target object in a
task that resembled one used by Kessler and Thomson (2010). We assume that when
participants intend to move in a clockwise or counterclockwise direction along the edge of a
round table, they must put the left or right side of their bodies forward first, respectively (Fig.

10
1). Indeed, in our preliminary study of a real situation, we confirmed this assumption: A
majority of 10 participants tended to move their left foot to start walking in the clockwise
direction, but their right foot in the counterclockwise direction (for details, see the Appendix).
If spatial perspective taking is analogous to such whole-body movements, corresponding
motor simulation should facilitate the mental transformation process. Therefore, our
hypothesis predicts that responses congruent with the direction of an imagined movement
(e.g., moving the left foot forward during a clockwise transformation) would facilitate spatial
perspective taking compared to incongruent responses (e.g., moving the left foot forward
during a counterclockwise transformation).
The present study’s task employed 0°, 40°, 80°, 120°, and 160° angle conditions in
clockwise and counterclockwise directions (Fig. 1). To focus on the top-down processing of
spatial perspective taking, the viewpoints to be imagined were represented by a chair but not
by an avatar because the avatar’s existence triggers additional bottom-up processing (Kessler
& Thomson, 2010). If our hypothesis is correct, the response congruency effect would lead to
a result similar to the posture congruency effect observed in Experiment 2 in Kessler and
Thomson (2010). That is, the congruency effect would occur only in high angle conditions
(i.e., 120° and 160°) because low angle conditions might allow direct judgments without
perspective transformations (Kessler & Thomson, 2010).
Fig. 1. Stimuli used in the spatial perspective-taking task (Experiments 1, 2, 4, and 5). We
assumed that participants first imagined moving the left or right side of their bodies
forward depending on the stimuli presented.

11
2. Experiment 1
Experiment 1 examined whether performance on a spatial perspective-taking task is
influenced by putting the left/right foot forward to respond. Our hypothesis predicts that an
action congruent with the direction of an imagined movement would facilitate spatial
perspective taking relative to an incongruent action, especially in high angle conditions (120°
and 160°), in which spatial perspective taking is more involved than in low angle conditions
(Kessler & Thomson, 2010).
2.1. Method
2.1.1. Participants
Participants in Experiment 1 were 24 undergraduate and graduate students (mean age =
21.4 years; 12 female and 12 male; 23 right-footed and one left-footed
2
). All had normal or
corrected-to-normal vision, were naïve to the study’s purpose, and received either pre-paid
cards for purchasing books or course credit for their participation. We determined this
number of participants in advance, following Kessler and Thomson (2010) who chose the
same sample size of 24 in all their experiments. According to post hoc analyses, this sample
of 24 would give us more than .99 power to detect the main effect of congruency and the
interaction of angle and congruency for RT data at the .05 significance level if the response
congruency effect has as large effect sizes as the posture congruency effect in Kessler and
Thomson’s (2010) Experiment 3. For the same reason, we applied this sample size to
Experiments 2, 4, and 5 as well. All experiments reported in this article were approved by the
2
In all experiments reported here, we determined participants’ dominant hand and foot by asking “which is
your dominant hand?” and “which foot do you use to kick a ball?”, respectively.

12
ethics board of the School of Human Sciences of Osaka University.
2.1.2. Stimuli and apparatus
Visual stimuli were created using the 3D computer graphics software Blender 2.71
(Blender Foundation, Amsterdam). Stimuli showed a room with a circular table on which a
flower (a chrysanthemum) and a sword were lying in front of a chair. The chair was
positioned at 0°, 40°, 80°, 120°, or 160° angular disparity from the participants’ viewpoint,
clockwise or counterclockwise (Fig. 1). Our stimuli mimicked those used by Kessler and
Thompson (2010). The circular table was viewed from an angle of 65° from horizontal.
Although this kind of bird’s eye view is somewhat unnatural in daily life, we adopted this
angle for two reasons. First, we wanted to use stimuli comparable to those used by a number
of previous studies on spatial perspective taking (e.g., Dalecki, Hoffmann, & Bock, 2012;
Kessler & Rutherford, 2010; Kessler & Thomson, 2010; Michelon & Zacks, 2006; Surtees,
Apperly, & Samson, 2013). Second, if the table had been viewed from a lower angle, the two
target objects and their separation would have been foreshortened, so their appearance would
have varied too much depending on their location. This might have contaminated results
because people are notoriously poor at precisely estimating depth dimension from 2-D
pictures (e.g., Sugihara, 2015).

13
Stimuli were displayed on a 24.1-in-wide LCD monitor (NEC MultiSync LCD-PA241W;
resolution of 1,920 × 1,200 pixels) at a viewing distance of about 80 cm. As shown in Fig. 2A,
participants stood, without their shoes, on a mat in front of a white line marked on the floor. A
120 mm (width) × 67 mm (length) dual-foot switch (USB 2FOOT SWITCH, Scythe Co., Ltd.,
Tokyo) was fixed on the floor about 2 cm in front of the participants’ toes as a response
device.
2.1.3. Procedure
Fig. 3 illustrates the stimulus sequence. All participants completed the experiment
individually in a laboratory. Each trial was initiated with a “Ready?” visual cue, which
remained until participants stepped on either the right or left switch using one foot. During
this time, participants could check the number of remaining trials by pressing the “T” key on
a keyboard placed in front of the monitor. The participants’ step initiated a 1-s blank screen;
then a picture of the target object (flower or sword) appeared, with its noun (in Japanese
kanji) for 1 s. Then, following a 1-s blank screen again, the experimental stimulus was
Fig. 2. Overhead views of setups used in Experiments 1, 2, 3, and 5. (A) The foot
condition. Without shoes, participants stood on a mat in front of a white line marked
on the floor and responded by stepping on a foot switch. The position and tilt of a
display were adjusted per participant so the viewing distance was about 80 cm. (B)
The hand condition. Participants sat on a pipe chair with their hands placed in front
of a white line on a table and responded by pushing a foot switch. A washcloth
covered the foot switch for hygienic reasons, but it is not drawn here for the sake of
simplicity.

14
presented. Participants imagined the viewpoint from the chair and then judged whether the
target object would be on the chair’s left or right side. They responded by stepping on the
corresponding switch (left or right) with one foot as quickly and accurately as possible. The
response foot (left or right) was manipulated across two blocks. During a trial, participants
had to keep their eyes on the monitor. After the response, a 1-s blank screen appeared, and
then the initial cue (“Ready?”) was presented again for the next trial. Only in practice trials
was visual feedback given on the blank screen when the response was incorrect. After every
stepping response, participants moved the foot back to its original standing position.
The experiment consisted of two blocks of trials. Participants were instructed to keep using
the same foot (left or right) to respond throughout each block, regardless of whether the
response was left or right. The response foot (left or right) was switched between the two
blocks, with the order counterbalanced across participants. Each block consisted of 108 trials
in random order; each of nine angular disparities was repeated 12 times. The target object
(flower or sword) and its position (left/right or right/left) were counterbalanced across trials.
Hence, a correct response was left on half the trials and right on the other half. Before each
block, participants completed 20 warm-up trials, in which a blue square was presented on
either the left or right position on a gray background, and participants were required to step
on the corresponding switch (left or right) and then complete 27 practice trials (randomly
Fig. 3. Procedure of the spatial perspective taking task in Experiments 1, 2, and 4.
Participants memorized a target (flower or sword) and then judged its position (left or
right) on the round table from the viewpoint of the chair. In this example, the correct
answer is “right.” The rightmost figure depicts objects on the table in a larger scale.

15
selected from main trials). At the end of the experiment, the experimenter asked participants
for introspective reports (remarks, employed strategies, and troubles faced during the
experiment) via open questions.
In this study, participants were explicitly forbidden to infer the correct answer by
symmetrically reversing the position from their own viewpoints (i.e., their own “left” = “right”
at the table’s opposite side), especially at high angles (i.e., 120° and 160°), because such a
reversal strategy seems to require processing different from spatial transformation (Kessler &
Wang, 2012; Wraga et al., 2000). Otherwise, the experimenter did not imply any specific
strategy to be employed, such as internal movement simulation or blink transformations
(Wraga et al., 2000).
2.2. Results and discussion
For our analyses, we categorized trials into two conditions: congruency between a response
foot (left or right) and the imagined movement direction (clockwise or counterclockwise).
That is, clockwise trials were regarded as congruent in the left-foot block, but as incongruent
in the right-foot block and vice versa for counterclockwise trials. Because the 0° trials cannot
be classified in terms of congruency, they were not included in comprehensive analyses but
analyzed separately as necessary. Thus, there are two orthogonal experimental factors:
congruency (congruent or incongruent) and angle (40°, 80°, 120°, or 160°). We conducted
Fig. 4. Means and standard errors of RT
data in Experiments 1 and 2.
Fig. 5. Means and standard errors of error
data in Experiment 1.

16
repeated-measures ANOVAs
3
with these two factors on RT and error data. For RT analyses,
we excluded error trials (2.3% of data) and trials that took longer than 2.41 s (= M + 4 SD;
0.9% of data
4
) and then calculated the mean RTs per cell for each participant. The mean RTs
and errors across participants are shown in Fig. 4 and Fig. 5, respectively.
The 2 × 4 ANOVA for RT data revealed significant main effects of angle (F(3, 69) = 60.71,
ηp2 = .745, p < .001) and congruency (F(1, 23) = 13.10, ηp2 = .463, p = .001) and significant
interaction of angle and congruency (F(1, 69) = 6.17, ηp2 = .241, p = .002). Post hoc t-tests
5
revealed a monotonic increase of RT with increasing angle, showing significant differences
for any pair of two consecutive angles (40° vs. 80°, t(23) = 2.08, d = 0.10, p = .049; 80° vs.
120°, t(23) = 5.39, d = 0.63, p < .001; 120° vs. 160°, t(23) = 4.31, d = 0.55, p < .001). In
addition, a separate paired t-test confirmed a faster response at 0° than at 40° (t(23) = 2.88, d
= 0.13, p = .008). Post hoc t-tests also revealed that RTs in the congruent condition were
shorter than in the incongruent condition at 120° (t(23) = 2.91, d = 0.18, p = .023) and 160°
(t(23) = 3.35, d = 0.26, p = .011), but no congruency effects were detected at 40° (t(23) = 0.38,
d = 0.02, p = .710) and 80° (t(23) = 0.53, d = 0.03, p = .601).
For error data, the 2 × 4 ANOVA revealed a significant main effect of angle (F(3, 69) =
9.57, ηp2 = .294, p < .001), but neither of congruency (F(1, 23) = 0.27, ηp2 = .012, p = .607)
nor of angle by congruency (F(1, 69) = 1.24, ηp2 = .051, p = .294). Post hoc t-tests revealed
that more errors occurred at 160° than at 40° (t(23) = 3.81, d = 0.88, p = .004) and 80° (t(23)
3
For the repeated-measures ANOVAs conducted in this article, we reported p values corrected by
Chi-Muller’s ε (Chi, Gribbin, Lamers, Gregory, & Muller, 2012) without assuming sphericity.
4
We used this criterion so that omission rates fell around 1% throughout our experiments. Nonetheless,
application of another criterion of M + 3 SD did not affect results of significance tests.
5
For any multiple comparisons in this article, we reported p values corrected by Holm’s (1979)
sequentially rejective Bonferroni procedure.

17
= 4.51, d = 0.89, p = .001) and showed no other significant differences (all ps > .070). Given
that very few errors occurred (2.3% overall), we consider RT data the major index of task
performance.
2.2.1. Interpretation of the angle effect
Results showed a trend toward longer RTs and more errors with increasing angle, consistent
with a number of previous studies on perspective change and viewer rotation (e.g., Carpenter
& Proffitt, 2001; Creem, Wraga, & Proffitt, 2001; Easton & Sholl, 1995; Huttenlocher &
Presson, 1973, 1979; Kessler & Rutherford, 2010; Kessler & Thomson, 2010; Michelon &
Zacks, 2006; Presson, 1982; Rieser, 1989; Surtees et al., 2013; Wraga et al., 2000). This angle
effect should be interpreted with caution as explained in Section 1.2. According to
participants’ introspective reports, 66.7% (16 of 24) spontaneously reported adopting a
concrete perspective-taking strategy (e.g., “I imagined myself rotating around the table”; “I
imaginatively moved to and sat on the depicted chair and then reached for a target object
from the imagined position”). In other words, the majority consciously imagined placing
themselves in a position from which they took a new perspective. The remaining 33.3% did
not clearly describe what strategy they used. More importantly, none reported performing
mental object rotation or using a reversal strategy. These introspections, suggestive of the
angle effect, provided a rare glimpse into the mind because very few studies on spatial
perspective taking have so far reported participants’ introspections. Because the introspective
data were merely an auxiliary measure, not our main concern, they were not conclusive.
However, those introspections do suggest that perspective transformation is what most people
naturally perform in the present task.
2.2.2. The response congruency effect
As we predicted, results showed that RTs at high angle conditions (120° and 160°) were

18
shorter when a response method (putting a left or right foot forward) was congruent with an
imagined movement (clockwise or counterclockwise) than when it was incongruent and that
the response congruency effect was not detected at low angle conditions (40° and 80°). These
results exhibited the same pattern as those of Kessler and Thomson’s (2010) Experiment 2,
which manipulated body postures. The response congruency effect may indicate that our
participants internally engaged in whole-body movement simulation when they responded.
Thus, the foot response consistent with simulation was facilitated, compared with the
inconsistent response. This implies interdependence between spatial perspective taking and
action related to whole-body movement, suggesting involvement of motor simulation in
spatial perspective taking.
Since the present experiment could not include a baseline condition in which responses
were neither congruent nor incongruent, whether spatial perspective taking was facilitated by
congruent responses or hindered by incongruent responses remains unclear. On the other
hand, Kessler and Thomson (2010) demonstrated both facilitation and interference effects
caused by their posture manipulation, depending on whether the posture was congruent or
incongruent. Thus, if the response congruency effect shares processes with the posture
congruency effect, then the response congruency effect should also contain both facilitation
and interference processes.
The occurrence of the congruency effect only in high angle conditions can be attributed to
different processes at low and high angles because a position judgment at lower angles can be
achieved by direct visual judgments from participants’ perspectives and does not necessarily
require spatial transformation (Kessler & Thomson, 2010). The difference in the congruency
effect between high and low angles may reject another possible account, i.e., the spatial
stimulus-response (S-R) compatibility. This account predicts that a visual stimulus presented
on the participant’s right side can be processed faster by the right hand than by the left, even

19
when stimuli’s spatial layout is irrelevant to a given task (see Simon, 1990, for a review). If
S-R compatibility occurred in our experiment, the congruency effect could be observed at all
angle conditions because S-R compatibility occurs even in a very simple task (Simon, 1990).
However, this was not the case in our experiment. Thus, the results of Experiment 1 should
be interpreted as evidence that spatial perspective taking involves whole-body motor
simulation. The possibility of spatial S-R compatibility is further investigated in Experiment
5.
3. Experiment 2
We demonstrated in Experiment 1 that, compared to the incongruent response, the
congruent foot response facilitated spatial perspective taking. This raises the question of
whether the congruency effect is specific to a foot response. Kessler and Thomson (2010)
suggested that spatial perspective taking involves whole-body representations rather than
those of specific body parts, like hands, in mental object rotation (Gardony et al., 2014;
Wexler et al., 1998; Wohlschläger & Wohlschläger, 1998). If this is the case, spatial
perspective taking might be influenced by any response method related to a whole-body
movement, such as extending a left/right arm as well as putting a foot forward. Throughout
most of the human species’ biological evolution, before humans became bipedal, forelegs
were essential to locomotion. Therefore, arm movement might influence spatial perspective
taking as a proxy for foot movement when feet could not be used to respond. Thus,
Experiment 2 examined whether the results of Experiment 1 can be replicated even when a
hand, instead of a foot response, was employed.
3.1. Method

20
3.1.1. Participants
Participants in Experiment 2 were 25 undergraduate and graduate students. One male was
omitted from analysis because his mean RT was 3 SD longer than the mean RT across
participants, perhaps due to a lack of the instruction to respond as quickly as possible.
Therefore, analyses were based on data from 24 participants (mean age = 21.9 years; 12
female and 12 male; all right-handed). All participants had normal or corrected-to-normal
vision, were naïve to the study’s purpose, and received either pre-paid cards for purchasing
books or course credit for their participation. None had participated in the previous
experiment.
3.1.2. Stimuli, apparatus, and procedure
The same stimuli and procedure described in Experiment 1 were used in Experiment 2, but
the setup was modified for hand responses (Fig. 2B). Participants sat on a pipe chair at an
80-cm viewing distance to the monitor and placed their hands in front of a white line marked
on a table. The dual foot switch used in Experiment 1 was fixed on the table about 2 cm in
front of participants’ fingertips and covered with a washcloth for hygienic reasons. During the
spatial perspective-taking task, participants responded by pressing the left or right switch
with one hand. The response hand (left or right) was switched between two blocks, with the
order counterbalanced across participants. After each response, participants replaced the
responding hand in the original position.
3.2. Results and discussion
As in Experiment 1, we conducted repeated-measures ANOVAs with two factors
(congruency and angle) on RT and error data. For RT analyses, we excluded error trials (2.3%
of data) and trials that took longer than 2.93 s (= M + 4 SD; 0.9% of data) and then calculated
the mean RTs per cell for each participant. The mean RTs and errors across participants are

21
shown in Fig. 4 and Fig. 6, respectively.
The 2 × 4 ANOVA for RT data revealed significant main effects of angle (F(3, 69) = 64.55,
ηp2 = .737, p < .001) and congruency (F(1, 23) = 9.67, ηp2 = .297, p = .005) and significant
interaction of angle and congruency (F(1, 69) = 4.24, ηp2 = .156, p = .016). Post hoc t-tests
revealed the monotonic increase of RT with increasing angle, showing significant differences
for any pair of two consecutive angles (40° vs. 80°, t(23) = 4.49, d = 0.38, p < .001; 80° vs.
120°, t(23) = 5.03, d = 0.68, p < .001; 120° vs. 160°, t(23) = 8.68, d = 0.97, p < .001). In
addition, a separate paired t-test detected no difference between 0° and 40° (t(23) = 1.09, d =
0.94, p = . 285). Post hoc t-tests also revealed that RTs in the congruent condition were
shorter than in the incongruent condition at 120° (t(23) = 3.63, d = 0.24, p = .006) and 160°
(t(23) = 2.60, d = 0.24, p = .048), but no congruency effects were detected at 40° (t(23) =
1.18, d = 0.09, p = .252) and 80° (t(23) = 0.20, d = 0.02, p = .845).
For error data, the 2 × 4 ANOVA revealed a significant main effect of angle (F(3, 69) =
6.45, ηp2 = .219, p = .002) but neither of congruency (F(1, 23) = 1.23, ηp2 = .051, p = .279)
nor of angle by congruency (F(1, 69) = 0.71, ηp2 = .030, p = .538). Post hoc t-tests revealed
that more errors occurred at 160° than at 40° (t(23) = 3.67, d = 0.85, p = .008) and 80° (t(23)
= 2.99, d = 0.52, p = .033), but no other significant differences (all ps > .160).
In summary, the same result pattern as in Experiment 1, using a foot response, was
obtained in Experiment 2, using a hand
response. According to participants’
introspective reports, their main strategy was
also similar to that in Experiment 1: 62.5%
(15 of 24) reported that they employed a
concrete perspective-taking strategy, and none
reported using a reversal strategy. Although a
Fig. 6. Means and standard errors of error
data in Experiment 2.

22
few participants (2 of 24; 8.3%) reported that they performed object rotation in some trials,
this is not surprising because multiple solution strategies are commonly used for spatial
problems (Schultz, 1991). Hence, no matter which body part (foot or hand) was used for
responding, spatial perspective taking was facilitated or inhibited depending on congruency
between a response method and the direction of the imagined movement. This suggests
involvement not of a specific body part but a whole-body representation in spatial perspective
taking.
4. Comparison between Experiments 1 and 2
Although similar results were obtained in Experiments 1 and 2, whether the effects of foot
and hand movement on spatial perspective taking share a common mechanism is still
unknown. To examine this question, we directly compared the results of Experiments 1 and 2.
Since Experiments 1 and 2 used the same experimental design (two levels of congruency and
four levels of angle as within-participant factors), we can conduct a mixed-design ANOVA on
48 participants’ RT data by adding a two-level between-participants factor of the responding
body part
6
.
6
We can assume that samples in Experiments 1 and 2 were homogeneous for the following two reasons: (1)
Since Experiments 1 and 2 were simultaneously planned, their 48 participants were recruited from the
same class during the same period. In Japan, due to the strict entrance examination system and rigorous
university rankings, students at Japanese universities are much more intellectually homogenous than
students at Western universities. Therefore, we have no reason to suspect that a sample of 24 students
differs from another sample. (2) Neither experiment showed reliable linear trends of the individual’s mean
RT for the spatial perspective taking task as a function of participation order (for Experiment 1, r = ₋.281,
p = .184; for Experiment 2, r = .342, p = .102).

23
The 2 (responding body part) × 2 (congruency) × 4 (angle) mixed-design ANOVA revealed
a significant main effect of the responding body part (F(1, 46) = 7.57, ηp2 = .141, p = .008)
and significant interaction of the responding body part and angle (F(3, 138) = 7.51, ηp2 = .140,
p = .002). No other two-way and three-way interactions of the responding body part were
significant (responding body part and congruency, F(1, 46) = 0.96, ηp2 = .021, p = .331;
responding body part and congruency and angle, F(3, 138) = 0.03, ηp2 = .001, p = .981).
These results indicated that foot responses were faster than hand responses and that the
amount of this foot advantage varied between angles, being largest at 160° (see Fig. 4). In
addition, a separate Welch’s t-test revealed a marginally significant foot advantage even at 0°
(t(46) = 1.96, d = 0.57, p = .056).
To examine further the foot advantage and the congruency effect, we extracted only high
angle conditions (120° and 160°), which may require processing distinct from low angle
conditions (Kessler & Thomson, 2010), and then conducted a 2 (responding body part) × 2
(congruency) × 2 (angle) mixed-design ANOVA. The results showed that the main effects of
all factors were significant (responding body part, F(1, 46) = 8.27, ηp2 = .152, p = .006;
congruency, F(1, 46) = 28.44, ηp2 = .382, p < .001; angle, F(1, 46) = 85.59, ηp2 = .650, p
< .001). In addition, two-way interaction of the responding body part and angle was found to
be significant (F(1, 46) = 10.75, ηp2 = .189, p = .002), indicating that the foot advantage was
more salient at 160° than at 120°. Furthermore, interactions of congruency with any one or
two factors were not detected (congruency and responding body part, F(1, 46) = 0.43, ηp2
= .009, p = .515; congruency and angle, F(1, 46) = 1.21, ηp2 = .026, p = .278; congruency and
responding body part and angle, F(1, 46) = 0.01, ηp2 < .001, p = .941), implying that the
amounts of the congruency effects were equivalent (53 ms on average) regardless of angle
(120° or 160°) and responding body part (foot or hand).

24
4.1. Equivalence of the congruency effect
We first consider whether motor simulation of foot and hand movements modulates the
process of spatial perspective taking in the same way. If the embodied nature of spatial
perspective taking were more closely linked to one specific body part than another, the
congruency effect would vary depending on the responding body part. However, comparison
between experiments revealed that congruency effects in foot and hand conditions were
indistinguishable. In other words, foot movement contributed to spatial perspective taking as
much as hand movement, at least in the present study. Although not yet conclusive, this is
compatible with our hypothesis that spatial perspective taking is mediated by simulated
movement of not a specific body part (e.g., foot or hand) but a whole body.
Comparison between experiments also revealed that the RT difference between congruent
and incongruent responses at 120° was as large as that at 160°, regardless of the responding
body part. If simulation of a whole-body movement functioned throughout spatial perspective
taking, the congruency effect would be larger at 160° than at 120° because of the additional
demand of longer-distance movement, but this was not the case. Rather, our finding supports
the notion that congruent movement leads to a “head-start” effect at the beginning of a
perspective transformation, in accordance with Kessler and Thomson’s (2010) explanation of
the posture congruency effect.
4.2. Why did the foot advantage occur?
Surprisingly, our data showed that foot responses were faster than hand responses in all
angle conditions and that they were especially salient at 160°. This phenomenon seems
counterintuitive because “the hand is the human’s favorite tool and the training effect for
other extremities is limited due to physiological conditions” (Pfister, Lue, Stefanini, Falabella,
Dustin, Koss, & Humayun, 2014, p. 4). Actually, Pfister et al. (2014) demonstrated that the
mean RT for hands was shorter than for feet by strictly measuring simple RTs for a switch

25
release. One possible reason for our contradictory finding is that the foot advantage was
induced by a mechanism unique to spatial perspective taking. This unique mechanism, if any,
might reflect that feet are more closely related to locomotion than hands. Although this
explanation might seem incompatible with the involvement of a whole-body schema as
described in the preceding section, the foot advantage is possibly induced by a process
different from the response congruency effect. Another possibility is that the foot switch used
in our experiments was particularly conducive to foot responses because of its design. In the
next experiment, we examined whether the foot advantage was due to the use of the foot
switch and whether it is unique to spatial perspective taking.
5. Experiment 3
Experiment 3 was designed to determine whether the foot advantage observed in
Experiments 1 and 2 was unique to spatial perspective taking or if it could be ascribed to
other simple reasons (e.g., properties of the response device used and/or general human
abilities). For this purpose, Experiment 3 used a simple orientation judgment task, in which
spatial perspective taking was unnecessary, but it was otherwise the same as that in
Experiments 1 and 2.
5.1. Method
5.1.1. Participants
Participants in Experiment 3 were 16 undergraduate and graduate students (mean age =
21.9 years; eight female and eight male; all right-handed; 15 right-footed and one left-footed).
All participants had normal or corrected-to-normal vision, were naïve to the study’s purpose,
and received pre-paid cards for purchasing books for their participation. None had

26
participated in previous experiments. In advance, we determined 16 as a sample size because
a multiple of eight was needed for the three counterbalanced factors (gender, foot/hand order,
and left/right order).
5.1.2. Stimuli, apparatus, and procedure
The table set stimuli used in Experiments 1 and 2
were replaced with two pictures of a flower and a
sword presented side by side (Fig. 7). The trial
sequence was the same as that used in the previous
experiments: participants first memorized a target
object (flower or sword) and then judged its position
(left or right) in an arrangement. The target object
(flower or sword) and its position (left/right or
right/left) were counterbalanced across trials and presented in random order. In Experiment 3,
all participants completed both the foot and hand conditions. Setups and response methods
for both conditions were the same as those in Experiments 1 and 2, respectively (see Fig. 2).
Each condition contained left- and right-limb blocks, and each block consisted of 48 trials.
Half the participants started with the foot condition, and the other half started with the hand
condition. The order of the block (left → right or right → left) was also counterbalanced
across participants. Before each block, participants completed eight practice trials, in which
visual feedback was given for incorrect responses.
5.2. Results and discussion
We excluded error trials (0.2% of data) and then calculated the mean RTs of foot and hand
responses for each participant. Fig. 8 presents the aggregated results. A paired t-test showed
that the mean RT for hand responses was significantly shorter than that for foot responses by
Fig. 7. Stimuli presented in the
simple orientation judgement
task (Experiment 3).

27
58 ms (t(15) = 3.11, d = 0.54, p = .007), contrary to the results from the spatial
perspective-taking task in Experiments 1 and 2. This result suggests that the inherent process
of spatial perspective taking induces the foot advantage.
6. Experiment 4
The results of Experiments 1 and 2 showed that spatial perspective taking was facilitated
when participants responded using actions congruent with the direction of an imagined
movement, compared to incongruent actions. Although this response congruency effect
suggests the involvement of motor simulation of whole-body movement in spatial perspective
taking, another interpretation is possible. The congruency effect could simply be attributed to
which side of the body, left or right, participants used in responding, regardless of its
relevance to a whole-body movement. This interpretation is based on the possibility that the
left or right side of the body might function in the same way as body postures did in Kessler
and Thomson’s (2010) experiments. The left-or-right account predicts that the congruency
effect would occur even when a response method is irrelevant to a whole-body movement, as
long as a responding body part belongs to either the left or right side of the body. On the other
hand, the motor simulation account we hypothesized predicts that a response method
irrelevant to a whole-body movement would not cause
the congruency effect. To examine which account is
valid, Experiment 4 used the response of an index
finger, a response movement that is most likely
irrelevant to a whole-body movement.
6.1. Method
Fig. 8. Means and standard
errors of RT for the simple
orientation judgment task
(Experiment 3).

28
6.1.1. Participants
The participants in Experiment 4 were 24 undergraduate, graduate, and research students
(mean age = 22.8 years; 12 female and 12 male; 22 right-handed and two left-handed). All
had normal or corrected-to-normal vision, were naïve to the study’s purpose, and received
pre-paid cards for purchasing books for their participation. None had participated in previous
experiments. According to post-hoc analyses of our Experiments 1 and 2, sample size 24
would give us more than .99 power to detect the main effect of congruency and the
interaction of angle and congruency for RT data at the .05 significance level. Thus, this
sample size is adequate to determine the congruency effect’s presence or absence.
6.1.2. Stimuli, apparatus, and procedure
All stimuli, apparatus, and basic procedures
in Experiment 4 were the same as those in
Experiment 2, except that a finger response was
employed. As described in Fig. 9, a keyboard
used as a response device was placed on a table
instead of the dual foot switch used in
Experiment 2. Participants sat on a pipe chair at
an 80-cm viewing distance to the monitor,
placed one hand on the table with the index
finger stretched and the thumb held by the other
fingers, laid the index finger on the “down
arrow (↓)” key, and kept the other hand on their laps. During the spatial perspective-taking
task, participants responded by pressing the “left arrow (←)” key or “right arrow (→)” key,
moving only the index finger. After each response, participants replaced the index finger in
the original position. The response hand (left or right) was switched between two blocks, with
Fig. 9. An overhead view of the setup in
the finger condition (Experiment 4).
Participants sat with the index finger of
the left or right hand placed on the
“down (↓)” key and responded by
pushing the “left arrow (←)” key or
“right arrow (→)” key. The other hand
was placed in their laps.

29
the order counterbalanced across participants.
6.2. Results and discussion
We conducted repeated-measures ANOVAs with two factors (congruency and angle) on RT
and error data. For the RT analyses, we excluded error trials (3.2% of data) and trials that
took longer than 2.58 s (= M + 4 SD; 0.9% of data) and then calculated the mean RTs per cell
for each participant. The mean RTs and errors across participants are shown in Fig. 10 and
Fig. 11, respectively.
The 2 × 4 ANOVA for RT data revealed a significant main effect of angle (F(3, 69) = 59.69,
ηp2 = .722, p < .001) but no main effect of congruency (F(1, 23) = 1.76, ηp2 = .071, p = .198)
and no interaction of angle and congruency (F(1, 69) = 1.22, ηp2 = .050, p = .309). Post hoc
t-tests revealed a monotonic increase of RT with an increasing angle, showing significant
differences for any pair of two consecutive angles (40° vs. 80°, t(23) = 4.22, d = 0.45, p
< .001; 80° vs. 120°, t(23) = 7.18, d = 0.84, p < .001; 12 0° vs. 160°, t(23) = 6.47, d = 0.89, p
< .001). In addition, a separate paired t-test found no significant difference between 0° and
40° (t(23) = 1.16, d = 0.10, p = .258).
To clarify further whether relevance to whole-body movement was critical to the response
congruency effect, we conducted a planned comparison of congruency-effect amounts in
relevant (Experiments 1 and 2) versus irrelevant (Experiment 4) conditions at 120° and 160°.
Fig. 10. Means and standard errors of RT
data in Experiment 4.
Fig. 11. Means and standard errors of
error data in Experiment 4.

30
The 2 (relevance) × 2 (congruency) × 2 (angle) mixed-design ANOVA revealed a marginally
significant interaction of relevance and congruency (F(1, 70) = 3.97, ηp2 = .054, p = .050).
This indicates that actions relevant to whole-body movement are necessary for the response
congruency effect.
For error data, the 2 × 4 ANOVA revealed a significant main effect of angle (F(3, 69) =
7.64, ηp2 = .249, p = .001) but no significant main effect of congruency (F(1, 23) = 0.03, ηp2
= .001, p = .867) and no interaction of angle and congruency (F(1, 69) = 0.11, ηp2 = .005, p
= .946). Post hoc t-tests revealed that more errors occurred at 160° than at 40° (t(23) = 3.46, d
= 0.71, p = .013) and 80° (t(23) = 3.05, d = 0.73, p = .028) and at 120° than at 40° (t(23) =
2.88, d = 0.50, p = .034) but no other significant differences (all ps > .095).
In summary, finger responses did not lead to a congruency effect, unlike movement-related
responses in Experiments 1 and 2. If participants in Experiment 4 used strategies other than
embodied transformations due to a finger response, the congruency effect’s absence might be
attributed to a qualitative strategy shift. However, that is unlikely because, according to
participants’ introspective reports, the dominant strategy employed by 62.5% (15 of 24) was
still a concrete perspective-taking strategy (66.7% in Experiment 1; 62.5% in Experiment 2);
a few participants (2 of 24; 8.3%) reported performing object rotation in some trials, just as in
Experiments 1 (0.0%) and 2 (8.3%). None reported using a reversal strategy. Overall, these
results support not the left-or-right account, but the motor simulation account as causing the
congruency effect.
7. Experiment 5
Experiments 1 and 2 found the response congruency effect. As mentioned in Section 2.2.2,
this effect might not be attributed to a kind of spatial S-R compatibility effect because the
response congruency effect was not observed at lower angles (i.e., 40° and 80°). Nonetheless,

31
Experiment 5 attempted to provide more direct evidence for ruling out this spatial
compatibility account for the response congruency effect. In this experiment, we manipulated
the presentation position (left or right) of a stimulus itself, as well as the movement’s
direction (clockwise or counterclockwise). If the spatial S-R compatibility account were true,
then RT would be shorter when the responding foot and the stimulus position were
compatible (e.g., the left foot for a left stimulus) than when they were incompatible (e.g., the
left foot for a right stimulus). Additionally, congruency between the rotational direction and
the responding foot would have no or less effect on RTs. On the other hand, if the response
congruency effect reflected the process of simulated whole-body movement during spatial
perspective taking, then congruency between the movement’s direction and the responding
foot would contribute to the response congruency effect regardless of spatial S-R
compatibility.
7.1. Method
7.1.1. Participants
Participants in Experiment 5 were 24 undergraduate and graduate students (mean age =
20.7 years; 12 female and 12 male; all right-footed). All had normal or corrected-to-normal
vision, were naïve to the study’s purpose, and received either pre-paid cards for purchasing
books or course credit for their participation. None had participated in previous experiments.
7.1.2. Stimuli, apparatus, and procedure
All stimuli, apparatus, and basic procedures in Experiment 5 were the same as those in
Experiment 1, except for the following three differences. First, we created new stimuli by
trimming both left and right edges of stimulus images in Experiments 1, 2, and 4, so new
stimuli could be presented within the display’s left or right half (Fig. 12). Second, stimuli’s

32
presentation position was
randomly varied left to right
from trial to trial. Stimuli
were presented at the center
of either the left or right
display half. Third, we
omitted the 0° condition to
secure an adequate number of
trials in limited experimental time. Thus, each block (for the left or right foot) consisted of
128 trials in random order: 8 angles × 2 presentation positions × 2 targets × 2 target positions
× 2 repetitions.
7.2. Results and discussion
The basic analytical procedure was the same as those for Experiments 1, 2, and 4, except
for addition of the new factor of spatial compatibility, defined by whether the presentation
position (left or right) was compatible or incompatible with the responding foot (left or right).
Thus, there were three orthogonal experimental factors: congruency (congruent or
incongruent), spatial compatibility (compatible or incompatible), and angle (40°, 80°, 120°,
or 160°). We conducted repeated-measures ANOVAs with these three factors on RT and error
data. For RT analyses, we excluded error trials (2.0% of data) and trials that took longer than
2.62 s (= M + 4 SD; 1.1% of data) and then calculated mean RTs per cell for each participant.
Mean RTs and errors across participants are shown in Fig. 13 and Fig. 14, respectively.
The 2 × 2 × 4 ANOVA for RT data revealed significant main effects of angle (F(3, 69) =
127.82, ηp2 = .848, p < .001) and congruency (F(1, 23) = 4.92, ηp2 = .176, p = .037). Post hoc
t-tests revealed monotonic increase of RT with increasing angle, showing significant
differences for any pair of two consecutive angles (40° vs. 80°, t(23) = 4.15, d = 0.14, p
Fig. 12. An example of a display showing a stimulus in
Experiment 5.

33
< .001; 80° vs. 120°, t(23) = 10. 31, d = 0.55, p < .001; 120° vs. 160°, t(23) = 9.23, d = 0.84,
p < .001). Importantly, there was no main effect of spatial compatibility (1139 ms and 1134
ms for the compatible and incompatible conditions, respectively; F (1, 23) = 1.93, ηp2 = .078,
p = .178), suggesting that a spatial compatibility effect did not work in this case. Interestingly,
there was a significant interaction of congruency and spatial compatibility (F(1, 23) = 10.60,
ηp2 = .316, p = .004). To unfold this interaction, we conducted separate ANOVAs for
compatible and incompatible conditions. Results revealed that the congruency effect was
significant when the stimulus was presented on the opposite side of the responding foot (F(1,
23) = 12.23, ηp2 = .347, p = .002), but not significant when the stimulus was presented on the
Fig. 13. Means and standard errors of RT data in Experiment 5.
Fig. 14. Means and standard errors of error data in Experiment 5.

34
same side of the responding foot (F(1, 23) = 0.10, ηp2 = .004, p = .759). There were no other
two-way and three-way interactions (congruency and angle, F(3, 69) = 2.28, ηp2 = .090, p
= .098; spatial compatibility and angle, F(3, 69) = 0.45, ηp2 = .019, p = .697; congruency and
spatial compatibility and angle, F(3, 69) = 0.08, ηp2 < .001, p = .945).
For error data, the 2 × 2 × 4 ANOVA revealed a significant main effect of angle (F(3, 69) =
11.57, ηp2 = .335, p < .001), but neither of congruency (F(1, 23) = 0.02, ηp2 < .001, p = .902),
nor of compatibility (F(1, 23) = 0.33, ηp2 = .014, p = .571). Post hoc t-tests revealed that more
errors occurred at 160° than at 40° (t(23) = 5.03, d = 1.46, p < .001) and 80° (t(23) = 4.30, d =
1.18, p = .001), at 120° than at 40° (t(23) = 2.83, d = 0.83, p = .038), and showed no other
significant differences (all ps > .050). No interactions were significant (congruency and angle,
F(3, 69) = 1.25, ηp2 = .051, p = .299; spatial compatibility and angle, F(3, 69) = 1.03, ηp2
= .043, p = .382; congruency and spatial compatibility, F(1, 23) = 1.23, ηp2 = .051, p = .279;
congruency and spatial compatibility and angle, F(3, 69) = 1.25, ηp2 < .051, p = .298).
In summary, spatial S-R compatibility had no effect on performance in spatial perspective
taking. In addition, the foot response congruent with the movement’s direction shortened
overall RT for spatial perspective taking only when the stimulus was presented on the
opposite side of the responding foot. Although these results contradict the spatial S-R
compatibility account, they also differ from our prediction of the motor simulation account in
some ways. In this experiment, the response congruency effect was limited to the stimulus on
the opposite side of the responding foot and was NOT limited to higher angle conditions.
These unpredicted findings could probably be explained by considering trajectories from the
participant’s position to the chair position. Unlike Experiments 1, 2, and 4, the distance
between the participant’s position and the chair position in Experiment 5 depended on the
rotational direction (clockwise or counterclockwise) even when the angle was the same. For
example, when the stimulus was presented on the display’s right side, the position of the

35
clockwise (i.e., inward) 40° was closer to the participant than that of the counterclockwise
(i.e., outward) 40°. This asymmetry of the imagined trajectory depending on the rotational
direction may explain Experiment 5 results.
Suppose that you imagine moving to the right outward side of the right table. In this case,
putting your right foot forward would make your body approach the table’s left rather than
right. Plus, rotating your body counterclockwise around the axis of your left leg would make
your back turn to the right table. Thus, responses to the stimulus on the responding foot’s
same side would not necessarily be congruent with the imagined movement. On the other
hand, putting your left foot forward would turn your whole body clockwise. Thus, when
responding to the right table opposite to the responding foot, your action is congruent with
movement to the left side of the table, but incongruent with movement to the right side. In
this case, the movement strategy may be preferred even for objects at lower angles because of
distance information that objects on the outward side are farther from you than objects on the
inward side.
These results could also be interpreted as new counterevidence against the sensorimotor
interference account. If putting the left or right foot forward mitigated interference between
real and imagined perspectives, then the response congruency effect should occur regardless
of the stimulus position because angular disparity between real and imagined perspectives
was invariant regardless of whether the stimulus position was left or right. However, results
of Experiment 5 showed the response congruency effect only for the stimulus presented on
the opposite side of the responding foot. Therefore, findings in Experiment 5 support the
motor simulation account for the response congruency effect, rejecting the sensorimotor
interference account as well as the spatial S-R compatibility account.

36
8. General discussion
8.1. Implication of the response congruency effect
In accordance with Kessler and Thomson (2010), we hypothesized that spatial perspective
taking is embodied as simulated whole-body movement. We found evidence that supported
this hypothesis and also provided new suggestions about embodied processes of spatial
perspective taking. Experiments 1 and 2 showed that spatial perspective taking at 120° and
160° was performed more efficiently when participants put forward a limb (left or right)
congruent with the direction of an imagined movement (counterclockwise or clockwise)
compared to incongruent movements. This finding conforms to Experiment 2 from Kessler
and Thomson (2010), in which a participant’s body posture was manipulated and the posture
congruency effect was observed at 120° and 160°.In addition, a comparison of Experiments 1
and 2 showed that the response congruency effects in foot and hand conditions were
indistinguishable, suggesting that simulated movement of a whole body, not of a specific
body part, mediates the process of spatial perspective taking. This was further confirmed by
Experiments 4 and 5. Experiment 4 used a response method irrelevant to a whole-body
movement (i.e., index finger movements of either hand) and resulted in no congruency effects.
Experiment 5 not only replicated the response congruency effect, but also rejected accounts
from spatial S-R compatibility and sensorimotor interference.
This response congruency effect suggests that a common neural basis underlies the
execution of spatial perspective taking and motor simulation of a whole-body movement.
This notion is evidenced by some previous research on brain activity during mental
perspective transformations. For example, participants in Creem, Downs, et al. (2001)
performed in an fMRI environment a viewer rotation task similar to that used by Wraga et al.

37
(2000); researchers found activation of the premotor area and other regions deemed to be
involved in motor processing. Likewise, Wraga et al. (2005) used fMRI and, during a
self-rotation task, observed that the left supplementary motor area was activated.
7
Additionally, using ERP mapping, Schwabe et al. (2009) reported activation of the posterior
frontal cortex corresponding to the premotor area during a perspective transformation task.
However, some fMRI studies on perspective transformations showed no motor-related
activations (e.g., Lambrey et al., 2012; Zacks et al., 2003). Such inconsistency might be
attributed to two reasons. First, the use of a whole-body schema in spatial perspective taking
does not seem obligatory but seems to be one possible strategy similar to motor strategies in
mental object rotation (see Zacks, 2008, for a review). This notion is consistent with Creem,
Downs et al.’s (2001) observation that some but not all participants showed premotor
activation. The likelihood of using a movement strategy is probably affected by a given task’s
properties. For example, stimuli used by Lambrey et al. (2012) had as many as four objects
on a table not aligned regularly; this seemed to impose somewhat-heavy cognitive demands
on participants and prompted the use of different strategies than movement simulation. This
issue is discussed in more detail in Section 8.4. Second, based on some limitations of fMRI
measurements indicated by Kunz et al. (2009), participants’ mobility is restricted in fMRI
environments. In addition, a recent study showed that the supine posture itself, required by
conventional fMRI studies, altered brain activities (Thibault, Lifshitz, & Raz, 2016). Such
fMRI features might have non-negligible effects on the strategy used in spatial perspective
taking. To support this, some participants in the present study informally reported that they
7
Although Wraga et al. (2005) supposed that activations of motor-related areas were not due to motor
simulations but to demands of their high-level cognitive task, their finding is also compatible with the
involvement of motor simulation.

38
sometimes moved their faces or shoulders a bit during the task to ease their position judgment.
This actual movement strategy is clearly impossible under the constrained fMRI condition.
Accordingly, neuroscientific data available so far must be interpreted with caution because
they are as yet inadequate for determining whether movement simulation is actually involved
in spatial perspective taking. The present behavioral study, however, has provided evidence
supporting involvement of motor simulation in spatial perspective taking by examining
effects of action related to whole-body movement. In addition, we suggested that movement
simulation plays a significant role only at the beginning, not throughout spatial perspective
taking (see also Section 4.1). Overall, our findings on the response congruency effect not
only extended Kessler and Thomson’s (2010) findings on the posture congruency effect, but
also unveiled cognitive and motor processes of spatial perspective taking. Nonetheless, the
present study is only the first step in investigating involvement of motor simulation, so our
conclusion is still premature. Further studies from a broader perspective (including both
behavioral and physiological viewpoints) are needed to draw a strong conclusion.
The response congruency effect also has implications for computational processes by
which people know the direction or trajectory of simulated movement. There are at least two
sources of information to determine the trajectory of simulated movement: One is the
rotational angle of target objects and the other is a path between themselves and the target
position on a stimulus image. This raises the question of whether people use information
about the rotational angle only or about both the rotational angle and the path to calculate the
trajectory of simulated movement. The present finding of the response congruency effect
supports that both the sources were used because Experiment 5 demonstrated that the
presentation position of stimuli modulated the response congruency effect in spite of the
same rotational angles. In our paradigm, participants probably executed mental self
translation and mental self rotation simultaneously by taking the smoothest and shortest path

39
computed based on the prior information about the rotational angle and the path. However,
what is the smoothest and shortest remains unclear. For example, does the layout of a scene
(e.g., the presence of obstacles or the shape of a table) affect the trajectory of simulated
movement? To clarify the nature of simulated whole-body movement, these issues should be
addressed in future studies.
8.2. Implication of the foot advantage
A comparison of Experiments 1 and 2 showed that spatial perspective taking was
processed more quickly when responses were made by a foot rather than by a hand. This is
contrary to our common-sense notion, and the hand advantage was observed in simple
orientation judgments (Experiment 3) and in a previous study (Pfister et al., 2014). Therefore,
the foot advantage discovered here must be considered unique to spatial perspective taking
and as evidence for movement simulation’s contribution to spatial perspective taking.
While one is walking forward, visual input is continually updated, and an expanding optic
flow occurs. Such a close link between locomotion and vision is well known. In this regard,
some evidence indicates that walking alters visual perception and cognition. For example,
Yabe et al. (Yabe & Taga, 2008; Yabe, Watanabe, & Taga, 2011) reported that a person
walking on a treadmill perceived an ambiguous, apparent motion presented on the floor as
moving backward, as if an optic flow actually existed, more frequently than did a person
standing still on a treadmill. In another example, Kunz et al. (2009) demonstrated that the
time for imagined walking without vision was closer to the time for real walking while
participants were stepping in place than while they moved their arms circularly (irrelevant to
walking) or merely standing still. Kunz et al. inferred that perceptual-motor conflict was
eliminated by actual stepping, whereby a mental simulation of imagined walking became
accurate. Consideration of these effects of foot movements on visuo-spatial representations,
together with our hypothesis that spatial perspective taking involves simulated whole-body

40
movement, leads to a prediction that spatial perspective taking would also likely be facilitated
by concurrent foot movement.
To the best of our knowledge, no other phenomena are comparable to the foot advantage.
Thus, we tentatively propose that the foot advantage in spatial perspective taking is due to the
link between feet underpinning whole-body movement and visuo-spatial information.
Investigations are underway to clarify the foot advantage’s detailed mechanism and the
conditions in which it occurs (partially reported in Muto, Matsushita, & Morikawa, 2016).
Although the foot advantage conforms to the notion that simulated whole-body movement
underlies spatial perspective taking, its mechanism seems somewhat different from the
response congruency effect. The major difference is that while the response congruency effect
occurred only at high angle conditions (i.e., 120° and 160°), the foot advantage was seen at
all angle conditions, including 0°, and it was most salient at 160°. At first glance, the foot
advantage’s ubiquity seems contradictory to the notion that low angle conditions required
fewer perspective transformations than high angle conditions; thus, the response congruency
effect was limited to the high angles (see Section 2.2.2). Furthermore, while the response
congruency effect that was independent of a responding body part (i.e., foot or hand) supports
the involvement of a whole-body representation, the foot advantage clearly suggests a
specific body part’s role. Future studies should reveal the interconnection or independence
between mechanisms of the response congruency effect and the foot advantage. Indeed, we
have already undertaken such studies (e.g., Muto, Matsushita, & Morikawa, 2016).
8.3. Embodied transformation as a strategy
The present study has succeeded in demonstrating movement simulation’s important role
in spatial perspective taking. In this section, we discuss the extent to which our findings can
be generalized to various situations, including real-life ones. Simulation of a whole-body

41
movement is likely executed in limited situations instead of all situations. As described in
Section 1.2, movement strategy seems more likely to be employed when a given task
emphasizes online rather than offline processing. This notion is supported by Gärling, Böök,
Lindberg, and Arce’s (1990) finding that estimations of elevation in a large-scale real
environment based on a cognitive map can be accomplished without movement simulation
such as “mental travel.” To further understand strategy differences, we consider an alternative
hypothesis postulated previously, i.e., the sensorimotor interference account (e.g., Brockmole
& Wang, 2003; May, 2004; Wang, 2005). According to the sensorimotor interference account,
the angle effect of a spatial perspective-taking task stems not from cognitive loads of mental
transformations but from the conflict between real and imagined perspectives. However, most
previous findings regarded as evidence for this account can be interpreted without assuming
sensorimotor interference. Rather, as described below, they exemplify strategy differences.
One of the most compelling pieces of evidence for the sensorimotor interference account is
that allowing participants time to complete transformations in advance did not attenuate the
angle effect (May, 2004; Wang, 2005). However, this is also accounted for by a strategy
change to avoid large demands on working memory (Kessler & Thomson, 2010; see also
Section 1.2 in this article). In another example, Brockmole and Wang (2003) found that
imagined perspective change required less effort when participants changed perspective
across environments (e.g., from facing west in the middle of a building to facing north in the
middle of their office in the building) than when they changed perspective within a single
environment (e.g., from facing north to facing east in the middle of their office). Although
Brockmole and Wang (2003) attributed the benefit in across-versus-within conditions to
reduced conflict between initial and updated perspectives, this finding can again be explained
from another viewpoint. In the across condition, the initial perspective seems completely
unnecessary for position judgment from the new perspective; thus, participants could directly

42
recall the new perspective instead of changing perspective. Mou and McNamara (2002)
demonstrated that humans’ representations of spatial layouts can be abstractly encoded
regardless of their actual visual experiences (i.e., representations from a never-seen-before
viewpoint can be recalled). In summary, these previous findings related to the sensorimotor
interference account can be interpreted as evidence of diverse strategies for spatial problems
involving offline processing rather than as evidence of sensorimotor interference effects. The
whole-body movement strategy is likely not necessarily suitable for these situations.
Even in the task we used, simulation of whole-body movement might not be obligatory, but
one possible strategy. For example, reversal strategy (i.e., reversing the left/right position of
objects) could also be used for high-angle conditions even though we eschewed this strategy
in our experiments. Consistently, Kessler and Wang (2012) reported that female participants
were more likely than male participants to use embodiment strategy for spatial perspective
taking. Kessler & Wang (2012) inferred that this gender difference occurred because men
adopted “rule-based” strategies such as the reversal strategy more often than women. To
examine whether such a gender difference was also obtained in our results, we reanalyzed
data from Experiments 1, 2, and 5 by including the gender factor, following Kessler and
Wang (2012). The analysis included only conditions in which the congruency effect was
detected (i.e., 120° and 160° angles of Experiments 1 and 2, and the spatially incompatible
condition of Experiment 5). We subtracted mean RTs for incongruent conditions from those
for congruent conditions per participant and treated RT difference as the index of the
response congruency effect. A 2 (gender) × 3 (experiment) between-participants ANOVA
revealed that the congruency effect of male participants (66 ms on average) was equivalent to
that of female participants (62 ms on average; F(1, 66) = 0.47, ηp2 = .007, p = .497). The
interaction of gender and experiment (F(2, 66) = 1.07, ηp2 = .031, p = .348) and the main
effect of experiment (F(2, 66) = 0.51, ηp2 = .015, p = .602) were also insignificant. This

43
absence of gender difference was probably due to our eschewal of reversal strategy. Therefore,
there seem to be multiple strategies for spatial perspective taking, like object-based mental
rotation (Flusberg & Boroditsky, 2011; Kosslyn, Thompson, Wraga, & Alpert, 2001). Future
research must determine conditions in which a certain strategy is more likely to be employed.
Nonetheless, as Kessler and Wang (2012) stated, the embodiment strategy seems to be the
natural, default method of spatial perspective taking because the vast majority of participants
showed a posture or response congruency effect. Specifically, 81 of 96 participants (84%) in
Kessler and Thomson (2010) showed a posture congruency effect indicated by a positive
value of RT differences from incongruent and congruent conditions (Kessler & Wang, 2012).
With the same criterion, a comparable proportion of our participants (81%, 58 of 72)
exhibited the response congruency effect. A distinct feature of our task was the emphasis on
online rather than offline processing (i.e., minimal demands on long-term memory); such a
feature is common to everyday spatial problem solving (Freksa & Schultheis, 2014).
Therefore, such embodied transformations are likely to be performed in real-life situations as
well.
8.4. Evolutionary and developmental origins
We demonstrated that simulated whole-body movement subserves the online process of
spatial perspective taking, unlike mental object rotation related to hand movements (e.g.,
Gardony et al., 2014; Wexler et al., 1998; Wohlschläger & Wohlschläger, 1998). This
difference might reflect different evolutionary histories between perspective and object-based
transformations and supports the multiple-systems framework (e.g., Zacks & Michelon, 2005;
Zacks & Tversky, 2005). As discussed by Kessler and Thomson (2010), the embodied nature
of spatial perspective taking can be considered a stepping stone from actual to imaginary
movements. This notion is consistent with previous findings in comparative psychology, for

44
example, that great apes are incapable of spatial perspective taking (Tomasello et al., 2005)
but can physically move to a human’s position in order to know what the human is looking at
(Bräuer, Call, & Tomasello, 2005). However, currently available findings on non-human
species are too indirect and few to draw such a conclusion.
To determine whether spatial perspective taking is unique to humans, the role of language
should also be considered because judgment of spatial directions is closely linked to spatial
terms (e.g., Franklin & Tversky, 1990; Imai, Nakanishi, Miyashita, Kidachi, & Ishizaki,
1999; Kessler & Rutherford, 2010). Kessler and Rutherford (2010) demonstrated that the
posture congruency effect occurred whether judgment was made by key or verbal responses.
However, even when the response modality was nonverbal, people could rely internally on
linguistic processing for spatial perspective taking. Unfortunately, to the best of our
knowledge, there has been no research on spatial perspective taking of humans without
egocentric direction terms in their language. To reveal the evolutionary history of spatial
perspective taking, future studies should also focus on linguistic/cultural factors.
Because spatial perspective taking involves motor processing, we also must focus on how
the spatial perspective-taking ability develops in human children. Although mental object
rotation ability is known to develop with action experience (e.g., Frick & Möhring, 2013),
any developmental link between spatial perspective taking and motor skills is still unknown.
Huttenlocher and Presson (1973) reported that fourth-grade children who had difficulty
imagining the appearance of a hidden array from new perspectives showed better
performances when they were allowed to move physically to the new perspective’s position,
suggesting that actual movement precedes imagined movement developmentally. In addition,
the onset of self-produced locomotion (i.e., crawling and walking) helps children develop
non-egocentric representations of locations (Needham & Libertus, 2011). Given these reports,
the spatial perspective-taking ability might be related to walking experience. Consistent with

45
this view, Creem, Wraga, and Proffitt (2001) argued that the advantage of viewer rotation
over array rotation on the ground plane is due to the daily experience of walking under
gravity.
In summary, our findings on movement simulation’s role in spatial perspective taking are
informative in terms of its evolutionary and developmental origins. For example, the fact that
a hand response produced as much congruency effect as a foot response suggests that our
arms remain integrated into the human brain’s locomotor system even several million years
after our ancestors became bipedal. Spatial perspective taking should be explored from
interdisciplinary perspectives to understand these issues more comprehensively.
Acknowledgements
This work was supported by a JSPS Grant-in Aid for JSPS Fellows (Number 16J00012).

46
References
Amorim, M.-A., & Stucchi, N. (1997). Viewer- and object-centered mental explorations of an
imagined environment are not equivalent. Cognitive Brain Research, 5, 229–239.
Bräuer, J., Call, J., & Tomasello, M. (2005). All great ape species follow gaze to distant
locations and around barriers. Journal of Comparative Psychology, 119, 145–154.
Brockmole, J. R., & Wang, R. F. (2003). Changing perspective within and across
environments. Cognition, 87, B59–B67.
Carpenter, M., & Proffitt, D. R. (2001). Comparing viewer and array mental rotations in
different planes. Memory & Cognition, 29, 441–448.
Chi, Y. Y., Gribbin, M., Lamers, Y., Gregory, J. F., & Muller, K. E. (2012). Global
hypothesis testing for high-dimensional repeated measures outcomes. Statistics in
Medicine, 31, 724–742.
Creem, S. H., Downs, T. H., Wraga, M., Harrington, G. S., Proffitt, D. R., & Downs, J. H.
(2001). An fMRI study of imagined self-rotation. Cognitive, Affective & Behavioral
Neuroscience, 1, 239–249.
Creem, S. H., Wraga, M., & Proffitt, D. R. (2001). Imagining physically impossible
self-rotation: Geometry is more important than gravity. Cognition, 81, 41–64.
Dalecki, M., Hoffmann, U., & Bock, O. (2012). Mental rotation of letters, body parts and
complex scenes: Separate or common mechanisms? Human Movement Science, 31, 1151–
1160.
Easton, R. D., & Sholl, M. J. (1995). Object-array structure, frames of reference, and retrieval
of spatial knowledge. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 21, 483–500.

47
Erle, T. M., & Topolinski, S. (2015). Spatial and empathic perspective-taking correlate on a
dispositional level. Social Cognition, 33, 187–210.
Flusberg, S. J., & Boroditsky, L. (2011). Are things that are hard to physically move also
hard to imagine moving? Psychonomic Bulletin & Review, 18, 158–164.
Franklin, N., & Tversky, B. (1990). Searching imagined environments. Journal of
Experimental Psychology: General, 119, 63–76.
Freksa, C., & Schultheis, H. (2014). Three ways of using space. In D. R. Montello, K.
Grossner, & D. G. Janelle (Eds.), Space in Mind: Concepts for Spatial Learning and
Education (pp. 31-48). Cambridge, MA: MIT Press.
Frick, A., & Möhring, W. (2013). Mental object rotation and motor development in 8- and
10-month-old infants. Journal of Experimental Child Psychology, 115, 708–720.
Gardony, A. L., Taylor, H. A., & Brunyé, T. T. (2014). What does physical rotation reveal
about mental rotation? Psychological Science, 25, 605–612.
Gärling, T., Böök, A., Lindberg, E., & Arce, C. (1990). Is elevation encoded in cognitive
maps? Journal of Environmental Psychology, 10, 341–351.
Hamilton, A. F., Brindley, R., & Frith, U. (2009). Visual perspective taking impairment in
children with autistic spectrum disorder. Cognition, 113, 37–44.
Hegarty, M., & Waller, D. (2004). A dissociation between mental rotation and
perspective-taking spatial abilities. Intelligence, 32, 175–191.
Huttenlocher, J., & Presson, C. C. (1973). Mental rotation and the perspective problem.
Cognitive Psychology, 4, 277–299.
Huttenlocher, J., & Presson, C. C. (1979). The coding and transformation of spatial
information. Cognitive Psychology, 11, 375–394.
Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian
Journal of Statistics, 6, 65–70.

48
Imai, M., Nakanishi, T., Miyashita, H., Kidachi, Y., & Ishizaki, S. (1999). The meanings of
FRONT/BACK/LEFT/RIGHT. Cognitive Studies, 6, 207–225.
Inagaki, H., Meguro, K., Shimada, M., Ishizaki, J., Okuzumi, H., & Yamadori, A. (2002).
Discrepancy between mental rotation and perspective-taking abilities in normal aging
assessed by Piaget’s three-mountain task. Journal of Clinical and Experimental
Neuropsychology, 24, 18–25.
Kessler, K., & Rutherford, H. (2010). The two forms of visuo-spatial perspective taking are
differently embodied and subserve different spatial prepositions. Frontiers in Psychology,
1, 213.
Kessler, K., & Thomson, L. A. (2010). The embodied nature of spatial perspective taking:
Embodied transformation versus sensorimotor interference. Cognition, 114, 72–88.
Kessler, K., & Wang, H. (2012). Spatial perspective taking is an embodied process, but not
for everyone in the same way: Differences predicted by sex and social skills score. Spatial
Cognition & Computation, 12, 133–158.
Kosslyn, S. M., Thompson, W. L., Wraga, M., & Alpert, N. M. (2001). Imagining rotation by
endogenous versus exogenous forces: Distinct neural mechanisms. Neuroreport, 12, 2519–
2525.
Kozhevnikov, M., & Hegarty, M. (2001). A dissociation between object manipulation spatial
ability and spatial orientation ability. Memory & Cognition, 29, 745–756.
Kunz, B. R., Creem-Regehr, S. H., & Thompson, W. B. (2009). Evidence for motor
simulation in imagined locomotion. Journal of Experimental Psychology. Human
Perception and Performance, 35, 1458–1471.
Lambrey, S., Doeller, C., Berthoz, A., & Burgess, N. (2012). Imagining being somewhere
else: neural basis of changing perspective in space. Cerebral Cortex, 22, 166–174.

49
May, M. (2004). Imaginal perspective switches in remembered environments:
Transformation versus interference accounts. Cognitive Psychology, 48, 163–206.
Michelon, P., & Zacks, J. M. (2006). Two kinds of visual perspective taking. Perception &
Psychophysics, 68, 327–337.
Mou, W., & McNamara, T. P. (2002). Intrinsic frames of reference in spatial memory.
Journal of Experimental Psychology. Learning, Memory, and Cognition, 28, 162–170.
Muto, H., Matsushita, S., & Morikawa, K. (2016). Comparison between foot and hand
responses for a spatial perspective-taking task. International Journal of Psychology, 51,
201.
Needham, A., & Libertus, K. (2011). Embodiment in early development. Wiley
Interdisciplinary Reviews: Cognitive Science, 2, 117–123.
Pfister, M., Lue, J.-C. L., Stefanini, F. R., Falabella, P., Dustin, L., Koss, M. J., & Humayun,
M. S. (2014). Comparison of reaction response time between hand and foot controlled
devices in simulated microsurgical testing. BioMed Research International, 2014, 769296.
Presson, C. C. (1982). Strategies in spatial reasoning. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 8, 243–251.
Rieser, J. J. (1989). Access to knowledge of spatial structure at novel points of observation.
Journal of Experimental Psychology. Learning, Memory, and Cognition, 15, 1157–1165.
Schwartz, D. L., & Holton, D. L. (2000). Tool use and the effect of action on the imagination.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 1655–1665.
Schultz, K. (1991). The contribution of solution strategy to spatial performance. Canadian
Journal of Psychology, 45, 474–491.
Shepard, R. N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science,
171, 701–703.

50
Simon, J. R. (1990). The effects of an irrelevant directional cue on human information
processing. In R. W. Proctor & T. G. Reeve (Eds.), Stimulus-response compatibility: An
integrated perspective (pp. 31-86). Amsterdam: Elsevier.
Schwabe, L., Lenggenhager, B., & Blanke, O. (2009). The timing of temporoparietal and
frontal activations during mental own body transformations from different visuospatial
perspectives. Human Brain Mapping, 30, 1801–1812
Sugihara, K. (2015). Height reversal generated by rotation around a vertical axis. Journal of
Mathematical Psychology, 68–69, 7–12.
Surtees, A., Apperly, I., & Samson, D. (2013). Similarities and differences in visual and
spatial perspective-taking processes. Cognition, 129, 426–438.
Thibault, R. T., Lifshitz, M., & Raz, A. (2016). Body position alters human resting-state:
Insights from multi-postural magnetoencephalography. Brain Imaging and Behavior, 10,
772–780.
Tomasello, M., Carpenter, M., Call, J., Behne, T., & Moll, H. (2005). Understanding and
sharing intentions: the origins of cultural cognition. The Behavioral and Brain Sciences, 28,
675–691; discussion 691–735.
Tversky, B., & Hard, B. M. (2009). Embodied and disembodied cognition: Spatial
perspective-taking. Cognition, 110, 124–129.
Waller, D. (2014). Embodiment as a framework for understanding environmental cognition.
In D. R. Montello, K. Grossner, & D. G. Janelle (Eds.), Space in Mind: Concepts for
Spatial Learning and Education (pp. 139-157). Cambridge, MA: MIT Press.
Wang, R. F. (2005). Beyond imagination: Perspective change problems revisited. Psicológica,
26, 25–38.
Wexler, M., Kosslyn, S. M., & Berthoz, A. (1998). Motor processes in mental rotation.
Cognition, 68, 77–94.

51
Wohlschläger, A., & Wohlschläger, A. (1998). Mental and manual rotation. Journal of
Experimental Psychology: Human Perception and Performance, 24, 397–412.
Wolbers, T., & Hegarty, M. (2010). What determines our navigational abilities? Trends in
Cognitive Sciences, 14, 138–146.
Wraga, M., Creem, S. H., & Proffitt, D. R. (2000). Updating displays after imagined object
and viewer rotations. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 26, 151–168.
Wraga, M., Shephard, J. M., Church, J. A., Inati, S., & Kosslyn, S. M. (2005). Imagined
rotations of self versus objects: An fMRI study. Neuropsychologia, 43, 1351–1361.
Yabe, Y., & Taga, G. (2008). Treadmill locomotion captures visual perception of apparent
motion. Experimental Brain Research, 191, 487–494.
Yabe, Y., Watanabe, H., & Taga, G. (2011). Treadmill experience alters treadmill effects on
perceived visual motion. PloS One, 6, e21642.
Zacks, J. M. (2008). Neuroimaging studies of mental rotation: A meta-analysis and review.
Journal of Cognitive Neuroscience, 20, 1–19.
Zacks, J. M., & Michelon, P. (2005). Transformations of visuospatial images. Behavioral and
Cognitive Neuroscience Reviews, 4, 96–118.
Zacks, J. M., & Tversky, B. (2005). Multiple systems for spatial imagery: transformations of
objects and bodies. Spatial Cognition & Computation, 5, 271–306.
Zacks, J. M., Vettel, J. M., & Michelon, P. (2003). Imagined viewer and object rotations
dissociated with event-related fMRI. Journal of Cognitive Neuroscience, 15, 1002–1018.

52
Appendix
In the Introduction section, we assumed that for a clockwise movement, participants would
put the left side of their bodies (e.g., left foot) forward first and for a counterclockwise
movement, the right side (e.g., right foot) first (see Fig. 1). To confirm this assumption and to
corroborate the finding of Experiment 1, we conducted the following preliminary experiment
in a real situation. In this experiment, participants were asked to physically move along the
edge of a round table in a real situation. We manipulated a moving direction (clockwise or
counterclockwise) and rotational angles (40°, 80°, 120°, or 160°). Since participants’ initial
position was unclear in typical computerized spatial perspective taking tasks (e.g., Kessler &
Thomson, 2010; Michelon & Zacks, 2006), we also manipulated the participants’ initial
positions (near or far). If our assumption is true, participants would tend to initially move
their left foot in the clockwise condition and their right foot in the counterclockwise
condition.
Participants were ten undergraduate and graduate students (mean age = 24.0 years; five
female and five male; nine right-footed and one left-footed). All had normal or
corrected-to-normal vision, were naïve to the study’s purpose and received pre-paid cards for
purchasing books. None had participated in the Experiments 1–5.
This experiment was conducted in a lecture room (610 × 706 cm). Fig. A1 shows the
configuration of the room. A round table (70 cm high and 180 cm in diameter) was positioned
at the center of the room. A pipe chair was set at one of eight positions around the table
according to the angle condition (40°, 80°, 120°, or 160° in the clockwise or
counterclockwise direction). The distance between the circumference of the table and the
front side of the chair was 40 cm. Two white starting lines were drawn on the floor, 50 cm
(near condition) or 100 cm (far condition) away from the table. Participants’ movements were

53
recorded by a fixed video camera right
behind their initial positions. The
experiment was guided by tones from
two speakers. The experimenter
controlled the procedure by using a
personal computer behind participants.
At the beginning of each trial,
participants stood in front of one of the
two starting lines (near or far) and
closed their eyes. During this, the
experimenter set a chair at one of the
eight angle positions (40°, 80°, 120°, or
160° in the clockwise or
counterclockwise direction). Then,
participants heard a 440-Hz tone and
opened their eyes to confirm the
position of a chair but remained standing still. Three seconds later, a 494-Hz tone was
presented and participants had to quickly walk to and sit on the chair along the shortest path.
After that, participants returned to the initial position and the next trial started.
Trials were blocked into two conditions of the initial positions (near or far) with the order
counterbalanced across participants. Each block consisted of eight trials (for eight chair
positions) in random order. Before the first block, participants completed two practice trials
randomly selected from the first block to understand the experimental procedure.
By watching the recorded video, we judged whether each participant moved his/her left or
right foot first away from the ground for each condition. Fig. A2 shows rates of participants
Fig. A1. Overhead view of the apparatus used
in the preliminary experiment. Participants
stood in front of a starting line (near or far) and
then walked down the shortest path to a chair
and sat on it. Gray diamonds represent possible
positions of a chair.

54
who moved their left (or right) foot first per condition. The results exhibited a clear pattern
consistent with our assumption: Participants tended to move their left foot to start walking in
the clockwise direction but right foot for the counterclockwise direction. To validate this, we
conducted a 4 (angle; 40°, 80°, 120°, or 160°) × 2 (direction; clockwise or
counterclockwise) × 2 (initial position; near or far) repeated-measures ANOVA on
first-moved foot (left foot = 0, right foot = 1). Consistent with our visual inspection, results
showed that participants initially moved their right foot in the counterclockwise condition
more frequently than in the clockwise condition (F(1, 9) = 74.68, ηp2 = .892, p < .001).
Results also showed a significant main effect of angle (F(3, 27) = 4.45, ηp2 = .331, p
= .036), a significant interaction of angle and initial position (F(3, 27) = 4.45, ηp2 = .331, p
= .036), and a marginally significant interaction of direction and initial position (F(1, 9) =
5.00, ηp2 = .357, p = .052). There were no main effect of initial position (F(1, 9) = 0.13, ηp2
= .014, p = .726) and other two-way (angle and direction, F(3, 27) = 1.54, ηp2 = .146, p
= .247) and three-way interactions (angle, direction and initial position, F(3, 27) = 1.54, ηp2
= .146, p = .247). These unpredicted significant patterns probably stemmed from an
exceptional trend observed in the 40°-clockwise-far condition, in which as much as 40%
participants started with their right foot even for clockwise movement. In the far condition,
Fig. A2. Rates of participants who moved their left (or right) foot first for each condition
in the preliminary experiment. The rates of left and right feet are represented by white and
black areas, respectively.

55
pathways to the 40° positions were straight rather than curved (see Fig. A1) and this might
prompt participants to move their dominant foot (right foot for the majority) first in the same
way as when they walked straight.
In summary, this experiment demonstrated our assumption that the first body movement
depends on the movement direction. This finding is consistent with the response congruency
effect found in Experiment 1, supporting the notion that spatial perspective taking involves
whole-body motor simulation that corresponds to actual whole-body movement.