Content uploaded by Marcello Costantini
Author content
All content in this area was uploaded by Marcello Costantini on Nov 15, 2017
Content may be subject to copyright.
The rubber hand illusion: Sensitivity and reference frame
for body ownership
Marcello Costantini
a,*
, Patrick Haggard
b
a
Department of Clinical Sciences and Bio-imaging, University of Chieti, Via Dei Vestini 33, 66013 Chieti, Italy
b
Institute of Cognitive Neuroscience and Department of Psychology, University College London, London, UK
Received 22 February 2006
Available online 20 February 2007
Abstract
When subjects view stimulation of a rubber hand while feeling congruent stimulation of their own hand, they may come to
feel that the rubber hand is part of their own body. This illusion of body ownership is termed ‘Rubber Hand Illusion’ (RHI).
We investigated sensitivity of RHI to spatial mismatches between visual and somatic experience. We compared the effects of
spatial mismatch between the stimulation of the two hands, and equivalent mismatches between the postures of the two hands.
We created the mismatch either by adjusting stimulation or posture of the subject’s hand, or, in a separate group of subjects, by
adjusting stimulation or posture of the rubber hand. The matching processes underlying body ownership were asymmetrical.
The illusion survived small changes in the subject’s hand posture, but disappeared when the same posture transformations
were applied to the rubber hand. Mismatch between the stimulation delivered to the subject’s hand and the rubber hand
abolished the illusion. The combination of these two situations is of particular interest. When the subject’s hand posture
was slightly different from the rubber hand posture, the RHI remained as long as stimulation of the two hands was congruent
in a hand-centred spatial reference frame, even though the altered posture of the subject’s hand meant that stimulation was
incongruent in external space. Conversely, the RHI was reduced when the stimulation was incongruent in hand-centred space
but congruent in external space. We conclude that the visual–tactile correlation that causes the RHI is computed within a
hand-centred frame of reference, which is updated with changes in body posture. Current sensory evidence about what is
‘me’ is interpreted with respect to a prior mental body representation.
!2007 Elsevier Inc. All rights reserved.
Keywords: Body image; Body schema; Rubber hand illusion; Body ownership; Proprioception
1. Introduction
1.1. Body ownership and body representations
The sense of one’s own body is a fundamental aspect of self-consciousness. In everyday life, we see, feel and
move our body, and have no doubt that it is our own. The term ‘body ownership’ has been given to this
1053-8100/$ - see front matter !2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.concog.2007.01.001
*
Corresponding author. Fax: +39 0871 3556930.
E-mail address: marcello.costantini@unich.it (M. Costantini).
Consciousness and Cognition 16 (2007) 229–240
Consciousness
and
Cognition
www.elsevier.com/locate/concog
experience (Gallagher, 2000). The sense of body ownership presumably depends on afferent sensations arising
within the body itself (Tsakiris, Hesse, Boy, Haggard, & Fink, 2006), but also on the coherence of current sen-
sory input with pre-existing cognitive representations of the body (Tsakiris & Haggard, 2005).
Psychological and neurological studies classically distinguish at least two internal representations of the
body, often called body schema and body image (for overviews of the distinction, see Bermudez, 2005; Cole
& Paillard, 1995; Eilan, Marcel, & Bermu
`dez, 1995; Gallagher, 1986; Gallagher, 1995; Paillard, 1999). Accord-
ing to Head’s (1920) classical description, the body schema is a model or representation of one’s own body that
provides a standard against which postures and body movements are judged. This model is considered to be
the result of past sensory experiences (primarily proprioceptive, but also tactile, and vestibular). This gives rise
to an unconscious representation of body configuration which allows us to move fluently through space, to
know where our body parts are, and to localise tactile stimuli on the body surface. In particular, the body
schema is described as intimately related to voluntary action: it is updated during action, and also supports
coordinated actions by providing a proprioceptive representation of the initial conditions for movement
(Ghez, Gordon, & Ghilardi, 1995; Sainburg, Poizner, & Ghez, 1993).
Body image, on the other hand is defined as a conscious idea or mental representation of one’s own body
(Adame, Radell, Johnson, & Cole, 1991; Schilder, 1935). It comprises body-specific perceptions, mental repre-
sentations, beliefs, attitudes and emotions (Cash & Brown, 1987; Gardner & Moncrieff, 1988; Powers, Schulman,
Gleghorn, & Prange, 1987). The appearance of one’s own body ‘from the outside’, is thought to be an important
component of body image. Therefore, we focus here on the visual perceptual component of body image.
In normal circumstances, body schema and body image together form a coherent basis for self-conscious-
ness. In abnormal circumstances, such as brain lesions or deafferentation, they may be dissociated (Gallagher
& Cole, 1995; Paillard, 1999; Rossetti, Rode, & Boisson, 1995).
1.2. Manipulating body ownership with the rubber hand illusion
Body image and body schema are putative internal representations, defined by their epistemic rather than phe-
nomenal properties. The link between these representations and the phenomenal sense of ownership has not been
explored. For example, does our experience of our own body arise primarily ‘from the inside’ through body sche-
ma, or ‘from the outside’ through body image? Illusions which manipulate sense of ownership are a powerful
experimental tool to investigate this question. In the rubber hand illusion (RHI; Botvinick & Cohen, 1998),
watching a rubber hand being stroked synchronously with one’s own unseen hand causes the rubber hand to ‘‘feel
like it’s my hand’’. A sense of ownership arises due to the matching visual and tactile stimulation. Because the
subject takes the viewed rubber hand as their own hand, one consequence of the illusion is that the viewed location
of the rubber hand adapts the proprioceptively perceived location of the subject’s own hand. The processes of this
adaptation are presumably the same as those in standard visual–proprioceptive conflicts such as prismatic adap-
tation (Welch, 1978). Therefore, the RHI can be measured indirectly but quantitatively as a drift of the perceived
position of the subject’s own hand toward the rubber hand, offering one of the few experimental measures of body
ownership (Botvinick & Cohen, 1998; Tsakiris & Haggard, 2005).
Several studies suggest that the RHI depends not only on matching the pattern of stimulation, but also on
the match between the rubber hand and pre-existing body representations. Ehrsson et al. (Ehrsson, Spence, &
Passingham, 2004), and Tsakiris and Haggard (Tsakiris & Haggard, 2005) both showed that orienting the rub-
ber hand at 180"and 90", respectively, to the subject’s own hand abolishes the illusion. This suggests that the
rubber hand must match the subject’s proprioceptive body schema for a sense of ownership to arise.
Tsakiris and Haggard (2005) showed that when subjects viewed a piece of wood, or a right rubber hand,
while being stroked on their left hand, the illusion was reduced. That is, the RHI requires that the to-be-in-
corporated object be visually similar to the actual body part that is stimulated. This suggests that the rubber
hand must also match the subject’s visual body image.
1.3. Two levels of multisensory matching
Ownership may therefore involves multiple levels of multisensory matching. First, the incorporation of the
rubber hand into the body depends on visual stimulation of the rubber hand matching the tactile stimulation
230 M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240
of the subject’s hand. This can be called a bottom–up matching process. Second, incorporation depends on the
visual image of the rubber hand matching the subject’s existing representations of their own body, including at
least a proprioceptive body schema and visual body image. These latter processes can be called top–down
matching process, since they depend on internal mental representation rather than current sensory input. Nei-
ther multisensory correlation nor coherence with pre-existing representations is alone sufficient for the sense of
ownership. For example, when I ride a bicycle over a bump in the road, I have correlated visual and tactile
experience of the bump, but do not feel that the bump is ‘me’. Similarly, looking at someone else’s hand does
not give me the sense of that hand being mine. Instead, the sense of ownership involves the combination of
these factors. When both stimulation and body representation are matched between the subject’s hand and
the rubber hand, there is supra-additive interaction in measures of the RHI (Tsakiris & Haggard, 2005).
1.4. Implications of matching processes for body ownership
Sensitivity is an important property of any matching process. What degree of mismatch is necessary for the
association between the subject’s hand and the rubber hand to be broken? Previous studies have shown that
the RHI is sensitive to the temporal pattern of stimulation (Armel & Ramachandran, 2003; Tsakiris &
Haggard, 2005). However, the spatial aspects of matching in the RHI have not been investigated in detail.
The RHI is known to depend on top–down factors such as body posture (Austen, Soto-Faraco, Enns, &
Kingstone, 2004; Ehrsson et al., 2004), visual appearance (Tsakiris & Haggard, 2005) and hand identity
(Tsakiris, Prabhu, & Haggard, 2006). Those studies show that large incongruities between the rubber hand
and the subject’s hand reduce the illusion, but the sensitivity has not been studied.
Here, we investigate two aspects of the matching process to clarify the mental operations underlying sense
of ownership. First, we investigate if ownership is equally sensitive to mismatches generated by manipulating
the subject’s hand or the rubber hand. Several studies suggest that large manipulations of proprioceptive rep-
resentation, up to 15"(Fourneret & Jeannerod, 1998; Paillard & Brouchon, 1974) can go undetected, while
visual acuity is much higher (Ernst & Banks, 2002). On this view, the illusion should be much more sensitive
to mismatches caused by visual manipulation, such as spatial manipulations of the rubber hand, than to mis-
matches induced by equal spatial manipulations of the subject’s hand.
Second, the matching process underlying ownership appears to deliver a categorical decision (‘‘that’s me/
not me’’), rather than a graded output. Small visual–proprioceptive (Fourneret & Jeannerod, 1998), or tem-
poral mismatches (Blakemore, Frith, & Wolpert, 1999) typically do not enter awareness or influence the sense
of self. However, as mismatch increases beyond a critical spatial or temporal value, this perceptual binding
suddenly breaks down. A matching process should therefore produce floor and ceiling effects in the sense
of ownership. Partial sense of ownership would be an unusual but interesting case. We therefore investigated
if a partial ownership can be found using RHI, in a window intermediate between floor and ceiling levels.
If partial ownership were obtained, we were interested in the contributions of bottom–up stimulation and top–
down body representation to the RHI. In such cases, we can test whether breakdown of ownership is driven pri-
marily by the spatial mismatch between visual and tactile stimulation, or by the spatial mismatch between visual
and tactile hand position. In addition, the interaction between these factors is of particular interest. When the sub-
ject’s hand posture and the rubber hand posture are different, is the RHI felt more strongly when stimulation of
the two hands is congruent in a hand-centred spatial reference frame, or when it is congruent in external space?
We therefore investigated the sensitivity of the RHI to spatial manipulations applied either to the subject’s
hand or the rubber hand, in two groups of subjects. Within each group, we induced a spatial mismatch either
in the orientation of stimulation, or in the postural representation, by changing the orientation of the hand itself.
2. Materials and methods
Two groups of subjects participated in the experiment. In the first group, we manipulated the subjects’
hand, while in the second group we applied spatially identical manipulations to the rubber hand. Therefore,
in the terminology used above, the RHI was modulated by body schema factors in the first group, and by body
image factors in the second group. They are called the ‘proprioceptive group’ and the ‘visual group’,
respectively.
M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240 231
Each group experienced four experimental conditions, which differed in the method used to introduce a
mismatch between the subject’s hand and the rubber hand. In the baseline condition, the sensory stimulation
and the hand posture were identical on both the subject’s hand and the rubber hand (see Fig. 1, part a). This
Fig. 1. Experimental design. Experimental manipulations are shown exaggerated by a factor of 2 for display clarity only. In the figure, the
hands on the left side represent the rubber hand while the hands on the right side represent the subject’s right hand.
232 M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240
corresponds to the classic RHI situation of previous studies (Botvinick & Cohen, 1998; Tsakiris & Haggard,
2005).
First, in the stroking mismatch condition, the paintbrush stimulating the subjects’ or rubber hand (respec-
tively in the first or the second group) was rotated (see Fig. 1b and e). The posture of the subject’s hand and
the rubber hand were not changed. This condition thus did not involve any mismatch at the representational
level of body schema or body image. It did, however, involve mismatch at the level of tactile and visual
stimulation.
Next, in the postural mismatch condition, we rotated either the subjects’ or the rubber hand according to
each subject’s group. In this condition, the paintbrushes applying the stimulation were rotated with the hand,
so that the stimulation remained the same within hand-centred space (see Fig. 1c for the proprioceptive group
and 1f for the visual group). Thus, the postural-mismatch condition only involves mismatch at the represen-
tational level between the subject’s hand and the rubber hand. It does not involve any mismatch between visu-
al and tactile stimulation, if the stimulation is considered in hand-space and not in viewer-centred space.
Finally, we included a postural-plus-stroking mismatch condition. This involved exactly the same postural
manipulations of body schema and body image as the hand-posture condition, but a different relation between
the visual and tactile stimulation at the sensory level. In this condition, the paintbrushes delivering tactile and
visual stimulation to the subject’s hand and to the rubber hand remained in place while the posture of the sub-
ject’s hand or rubber hand was changed (see Fig. 1d and g, respectively). Therefore, the mismatch at the pos-
tural level was exactly as before. Now, however, there is no mismatch at the level of sensory stimulation, if the
stimulation is considered in an external, egocentric space. However, there is a clear mismatch if the stimulation
is considered in hand-centred space.
To summarise, two groups of subjects were tested in an RHI paradigm. In one (‘proprioceptive group’), the
sensory and postural experiences related to the subject’s hand were varied while those related to the rubber
hand were held constant. In another group (‘visual group’), equivalent variations were applied to the rubber
hand, while the experiences related to the subject’s hand were held constant. Each group experienced several
different mismatch conditions, obtained by rotating either the sensory stimulation, the posture of the subject’s
hand or rubber hand, or both. Finally, the degree of mismatch was systematically varied (10",20"or 30").
Thus, each subject performed 10 different conditions (three levels of mismatch in each of three sensory con-
ditions, plus one baseline condition). Each condition was tested once in a single block, and the order of blocks
was randomised anew for each subject.
Participants sat in front of a table. The subject’s right hand rested on the table in front of their right shoul-
der, while the rubber hand rested on the table in front of the subject’s midline at the same distance in front of
the subject’s chest as the subject’s hand was. The lateral distance between the middle finger of the subject’s
hand and the middle finger of the rubber hand was 30 cm. At the beginning of each block, the experimenter
adjusted the posture of the subject’s right hand or of the rubber hand (respectively in the first and second
group) while maintaining the same central position. The entire table was covered by a mirror, which prevented
the subject ever seeing their hand directly. A small half-mirror was set into the mirror just above the rubber
hand (Fig. 2). Lighting underneath the half-mirror was controlled by a computer to make the rubber hand
appear (during stimulation) and disappear (during position judgment, see later).
The RHI has been measured in a number of ways, including perceived position of the subject’s hand (Tsak-
iris & Haggard, 2005) reaching errors (Botvinick & Cohen, 1998), saccades toward the felt position of the hand
(Tsakiris, Costantini and Haggard, unpublished data) or skin conductance response (Armel & Ramachan-
dran, 2003). Here, we have used the proprioceptively perceived position of the subject’s hand as an implicit,
quantitative proxy for the illusion. Previous RHI studies have shown a shift of the perceived position of the
subject’s hand towards the rubber hand (Botvinick & Cohen, 1998; Tsakiris & Haggard, 2005), which corre-
lates with the strength of the illusion. We therefore presented a standard ruler above the table and reflected in
the mirror to appear at the same gaze depth as the rubber hand and subject’s hand, and spanning the spatial
range between them. Participants were asked, ‘‘Where is the knuckle of your middle finger?’’ They responded
by verbally reporting a number on the ruler. They were instructed to judge the position of their finger by pro-
jecting a parasagittal line from the centre of their knuckle to the ruler. During the judgments, there was no
tactile stimulation, and subjects were prevented from seeing the rubber hand or any other landmark on the
worksurface, by switching offthe lights under the half-mirror. After the judgment, the ruler was removed,
M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240 233
and the lights under the two-way mirror were turned on to make the rubber hand reappear, ready for the next
trial. The ruler was always placed with a different random offset for each judgement to prevent subjects from
memorising and repeating responses given on previous trials. Subjects made a baseline judgement before each
stimulation trial, and a further one afterwards. The difference between these two represents the change in per-
ceived hand position due to the stimulation, and was taken as our measure of RHI.
Stimulation was delivered using paintbrushes (1 mm width) mounted on computer-controlled stepper
motors. The visual stimulation applied to the rubber hand and the tactile stimulation applied to the subject’s
hand were delivered by identical brushes and identical motors. On each trial stimulation was delivered for
2 min, and consisted of a succession of several strokes over the dorsum of the hand, varying in direction
and speed. The stimulation was always exactly synchronous between the subject’s hand and the rubber hand.
Because we were interested in the sensitivity of the RHI to spatial mismatches, we focussed on the synchro-
nous condition only. Other studies have shown that the RHI does not occur following asynchronous stimu-
lation (Armel & Ramachandran, 2003; Ehrsson et al., 2004; Tsakiris & Haggard, 2005).
A brief rest period followed each block. To prevent transfer of the illusion across blocks, the subjects were
encouraged to move the hand and body during the rest period. The experimenter then passively replaced the
hands in the correct position ready for the next trial.
Twenty-six volunteers participated in the experiment on the basis of informed consent, and in accordance
with local ethical permissions and the declaration of Helsinki. Their mean age was 28.1 years. The first group
comprised 15 subjects (7 female), and the second 11 (6 female). Two of the subjects were left handed, and all
had normal or corrected-to-normal vision.
3. Results
The dependent variable was the change in perceived position of the subject’s hand between judgements tak-
en before and after each block. A positive value represents a mislocalization toward the rubber hand due to
Fig. 2. Experimental setup. The participant’s right hand was out of sight for the whole duration of the experiment. The rubber hand
appeared, aligned with the participant’s midline, only during the stimulation and disappeared during the judgment period.
234 M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240
the visual and tactile stimulation, and thus an RHI. Since there may be large individual differences in RHI, we
first compared the drift in the baseline condition of perfect visual–tactile matching between the two group of
subjects to ensure they were comparable. Unrelated samples t-test showed that the two groups of subjects had
similar baseline susceptibility to the illusion [t(24) = .34; p> .05]. The overall mean of both groups was, how-
ever, significantly different from zero, confirming a change in perceived hand position during the period of
stimulation, consistent with RHI (proprioceptive group = 1.87 cm; visual group 2.27 cm). The raw data are
represented in Table 1.
We next subtracted each subject’s RHI value in the baseline condition from their RHI values in each other
experimental condition. This isolated the change in the RHI due to the experimentally controlled mismatch
between the rubber hand and the subject’s hand. Subtracting the baseline condition also controls for minor
individual differences in susceptibility to the illusion, and in proprioceptive representation.
We performed two planned analyses, to address the hypotheses raised in the introduction. First, we inves-
tigated whether the RHI was more sensitive to postural manipulations of the subject’s hand or to equivalent
postural manipulations of the rubber hand. To do this, we performed a two-way factorial ANOVA with fac-
tors of degree of mismatch (10", 20", 30", within subjects), and origin of mismatch on the baseline-corrected
RHI values. Neither main effect was significant ([F(2, 48) = 2.56, p> .05]; [F(1, 24) = 0.46, p> .05]. The inter-
action between the two factors was, however, highly significant [F(2, 48) = 6.47, p< .01]. Follow-up simple
effects analysis revealed the following. In the proprioceptive group there was a clear modulation of the illusion
with the degree of mismatch. The change in the illusion induced by 10"of mismatch (!0.17 cm) was less than
at 20"(!1.6 cm, p< .01) and 30"(!1.4 cm, p< .05), which in turn did not differ from one another. Converse-
ly, in the visual group, even a 10"mismatch abolished the illusion, and there was no further modulation as the
degree of mismatch increased (Fig. 3). Further simple effects testing at each level of mismatch confirmed a dif-
ference between the groups at 10"of discrepancy (p< .01) but not beyond (p> .05). We conclude that the RHI
was abolished by mismatches due to changed posture of the rubber hand (visual group) but survived equiv-
alent small mismatches due to changes in posture of the subject’s hand (proprioceptive group). That is, small
alterations in the subject’s hand posture did not significantly affect the RHI.
Our second analysis compared the sensitivity of the illusion to mismatches in sensory stimulation, and in
postural representation. This analysis was necessarily focussed on the conditions where partial mismatch
occurred, namely the 10"mismatch for the proprioceptive group. The other mismatch conditions produced
floor and ceiling effects in the proprioceptive group, and the visual group showed such effects for all discrep-
ancies. Previous research showed that sense of ownership involves a supra-additive interaction between sen-
sory-level matching and postural-level matching (Tsakiris & Haggard, 2005). We therefore predicted that
mismatch in both stimulation and postural representation would reduce the illusion more than mismatches
in stimulation alone and in postural representation alone. However, describing our three conditions in terms
of stimulation mismatch or postural representation mismatch depends on the hypothesised frame of reference
used for matching. These hypotheses are outlined in Table 2. Specifically, the degree of matching between visu-
al and tactile stimulation could be computed in an external, viewer-centred reference frame, or in hand-centred
reference frame that is updated with changes in hand posture.
Table 2 shows that our stroking mismatch condition, in which the spatial orientation of stimulation is varied
while hand posture is unchanged, involves mismatch at the stimulation level alone, and no mismatch in pos-
tural representation, under both the viewer-centred spatial hypothesis, and the hand-centred hypothesis.
Table 1
Mean proprioceptive drift and standard errors across conditions
Degrees of mismatch Proprioceptive group Visual group
Baseline Stroking Postural Stroking plus
postural
Baseline Stroking Postural Stroking plus
postural
0 1.87 (0.67) 2.27 (1.06)
10 2.13 (0.67) 2.13 (0.59) 0.83 (0.49) !0.09 (0.41) 0.36 (0.83) 0.18 (0.78)
20 !0.23 (0.80) 0.93 (0.76) 0.13 (0.55) !0.05 (0.57) 0.55 (0.45) 0.32 (0.79)
30 0.97 (0.88) !0.17 (0.73) 0.63 (0.61) 1.00 (0.45) 1.09 (0.65) 1.09 (0.59)
M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240 235
Conversely, the postural mismatch condition, in which both the spatial orientation of the hand and the stim-
ulating apparatus were varied, involves a mismatch in postural representation under both hypotheses. It addi-
tionally involves a mismatch at the stimulation level, under the viewer-centred space hypothesis, but not under
the hand-centred hypothesis. The converse is true in the postural-plus-stroking mismatch condition. This con-
Fig. 3. Mean proprioceptive drifts toward the rubber hand. Error bars indicate standard errors. Zero represents the felt position of the
participant’s hand in the baseline condition (the classical rubber hand illusion). Negative numbers indicate a reduction in RHI relative to
this baseline.
Table 2
Hypothesised interactions between stimulus matching and postural representation matching underlying ownership, and contrasts
coefficients used to test them
Stroking mismatch Postural mismatch Postural plus stroking
mismatch
Spatial reference frame for
stimulus matching
Hand
space
Viewer-centred
space
Hand
space
Viewer-centred
space
Hand
space
Viewer-centred
space
Stimulation match Mismatch Mismatch Match Mismatch Mismatch Match
Postural representation match Match Match Mismatch Mismatch Mismatch Mismatch
Contrast for viewer-centred space
hypothesis
!12 !1
Contrast for hand-space
hypothesis
!1!12
236 M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240
dition again involves mismatch in postural representation under both hypotheses. However, the postural-plus-
stroking condition involves a mismatch at the stimulation level only under the hand-centred hypothesis, and
not under the viewer-centred space hypothesis, since the stimulator positions are exactly as in the baseline
condition.
We therefore constructed two planned contrasts to test whether the sense of body ownership depends on
spatial matching of stimulation and postural factors in external, viewer-centred space or in hand-centred
space. On the viewer-centred space hypothesis, the RHI should be more affected by mismatches in the postural
condition, than in either the stroking condition or postural-plus-stroking condition. This is because the pos-
tural condition involves external-space mismatches in both stroking and in hand posture. On the hand-space
hypothesis, the RHI should be more affected by mismatches in the postural-plus-stroking condition than in the
stroking condition and postural condition. This is because the postural-plus-stroking condition involves hand-
space mismatch in both stroking and hand posture. Since these hypotheses always predict stronger reductions
of the illusion in the double-mismatch conditions than in the single-mismatch conditions, one-tailed tests were
used to test each hypothesis. The contrast coefficients are shown in the lower rows of Table 2. The data used to
test the hypotheses are shown in Fig. 4, as baseline-corrected drifts for a 10"mismatch in each sensory con-
dition in the proprioceptive group.
The viewer-centred space hypothesis was not confirmed (t(14) = 1.08, p= .15), while the hand-space
hypothesis was confirmed (t(14) = 2.24, p= .02).
4. Discussion
Tsakiris & Haggard (2005) hypothesised that the sense of ownership involved a supra-additive interaction
between a bottom–up process of matching multisensory stimulation and a top–down process of assimilating
an external object, such as a rubber hand, to pre-existing representations of the body. Previous studies of the
RHI have confirmed the presence of this interaction at the behavioural (Pavani, Spence, & Driver, 2000;
Tsakiris & Haggard, 2005) and neural levels (Ehrsson et al., 2004; Tsakiris et al., 2006). However, those obser-
vations did not clarify how stimulation matching and body representation matching might interact. Our sen-
sitivity analysis sheds light on this point. We confirmed that an interaction is present: mismatch between the
subject’s hand and the rubber hand in both posture and stimulation reduced the illusion more the summed
effect of posture mismatch alone and stimulation mismatch alone. However, this was only true if the stimu-
lation mismatch is assumed to occur within an internal hand-centred space, which is updated with postural
Fig. 4. Using partial illusions to identify the reference frame for ownership. Mean baseline-corrected drift for 10"mismatch under different
sensory conditions in the proprioceptive group. More negative numbers indicate a reduction in RHI. Error bars indicate standard errors.
M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240 237
changes. Put another way, first a transformation aligns the rubber hand with the subject’s own hand and then
the correlation between visual and tactile stimulation is computed.
These results suggest that the RHI depends on a pre-existing body representation, with its own frame of
reference, distinct from external spatial representation. Several studies have emphasised the importance of
repetitive, correlated stimulation in inducing plasticity in sensory systems. For example, repeated paired stim-
ulation of two digits links the cortical representations of those digits by a process akin to Hebbian learning
(Armel & Ramachandran, 2003; Schweizer, Braun, Fromm, Wilms, & Birbaumer, 2001; Tegenthoffet al.,
2005). Multisensory plasticity may play a special role in learning to use tools (Iriki, Tanaka, & Iwamura,
1996). The plasticity associated with tool use could involve either updated representations of the body itself,
or of the external space around the body. To date, primate (Iriki et al., 1996) and human (Berti & Frassinetti,
2000; Holmes, Calvert, & Spence, 2004; Maravita, Husain, Clarke, & Driver, 2001; Maravita, Spence, Ken-
nett, & Driver, 2002) studies have focussed on plasticity of spatial representation rather than on plasticity
of body representation. Our result suggests a specific process of plasticity of body representation underlying
the sense of ownership.
The sense of owning one’s body is an important component of self-consciousness (Bermudez, 1998). Self-
consciousness in general, and the sense of ownership in particular, are characterised by unity. The body seems
a coherent whole ‘‘me’’, not merely a disorganised collection of inputs (Tsakiris et al., 2006). Our results sug-
gest that an internal body representation acts in a top–down fashion to constrain whether novel sensory stim-
uli are assimilated to the body or not. The unity of consciousness would come from this top–down influence.
Recent accounts of multisensory integration involve the feedforward combination of unimodal signals to cre-
ate a single percept (Ernst & Banks, 2002). Our result shows that perception of current sensory stimulation
involves assimilation to a pre-existing internal representation of the body, in addition to feedforward combi-
nation of current inputs. Specifically, the effects of matching visual and tactile stimulation can occur only after
the current visual and proprioceptive states of the body are first matched. The origin of the internal body rep-
resentation is not clear. It could be innate (Meltzoff& Moore, 1977) or it could be gradually constructed from
the history of sensory inputs, in a Bayesian manner (Jaynes, 1986; Kording & Wolpert, 2004; MacKay, 2003).
Our results show only that such an internal representation exists, that it modulates the matching of visual and
tactile stimulation, and that it supplies an internal reference frame for matching sensory events.
To conclude, we have used the rubber hand illusion to investigate the sense of body ownership. We con-
firmed that the sense of ownership of an external object (a rubber hand) decreased as the degree of spatial
mismatch between visual and tactile information increased. The matching processes underlying the sense of
ownership are asymmetrical: a small transformation of what the subject sees reduces the illusion more than
the equivalent transformation of what they feel. By slightly altering the tactile stimulation and proprioceptive
state associated with the illusion, we were able to introduce a partial sense of ownership. In this case, the
strength of the RHI depended on an interaction between stimulation factors and postural factors. Most
importantly, an internal hand-centred spatial reference frame was used to determine the match between what
was seen and what was felt. Our results suggest that the brain maintains an internal body representation, with
its own characteristic spatial organisation based on proprioception. This representation uses a frame of refer-
ence based on the specific part of the body that is stimulated. Moreover, this internal body representation is
used to process novel sensory stimulation, and to attribute such stimuli either to the self or the external world.
Specifically, novel stimuli are assimilated or discriminated from the self on the basis of coherence with this pre-
existing internal body representation. In that sense, the sense of body ownership is a top–down effect, of a kind
familiar in classical perceptual psychology. An important element of the self in personality and social psychol-
ogy (Baumeister, 1999) is the continuity of the self through time, demonstrated by autobiographical memory,
and by the ability to integrate novel experiences into a narrative account of self. We suggest that the integra-
tion of current visual and tactile stimulation with respect to a pre-existing body representation, as demonstrat-
ed in this experiment, may be a precursor of this continuity of self-consciousness.
Acknowledgment
This work was funded by grants from: BBSRC, Royal Society ESEP, MIUR.
238 M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240
References
Adame, D. D., Radell, S. A., Johnson, T. C., & Cole, S. P. (1991). Physical fitness, body image, and locus of control in college women
dancers and non-dancers. Perceptual and Motor Skills, 72, 91–95.
Armel, K. C., & Ramachandran, V. S. (2003). Projecting sensations to external objects: evidence from skin conductance response.
Proceedings of the Royal Society B, 270, 1499–1506.
Austen, E. L., Soto-Faraco, S., Enns, J. T., & Kingstone, A. (2004). Mislocations of touch to a fake hand. Cognitive, Affective, &
Behavioural Neuroscience, 4, 170–181.
Baumeister, R. F. (1999). The self in social psychology. Philadelphia, PA: Psychology Press.
Bermudez, J. L. (1998). The paradox of self-consciousness. Cambridge, MA: MIT Press.
Bermudez, J. L. (2005). The phenomenology of bodily awareness. In D. W. Smith & A. L. Thomasson (Eds.), Phenomenology and
philosophy of mind (pp. 295–316). Oxford: Oxford University Press.
Berti, A., & Frassinetti, F. (2000). When far becomes near: remapping of space by tool use. Journal of Cognitive Neuroscience, 12, 415–420.
Blakemore, S. J., Frith, C. D., & Wolpert, D. M. (1999). Spatio-temporal prediction modulates the perception of self-produced stimuli.
Journal Of Cognitive Neuroscience, 11, 551–559.
Botvinick, M., & Cohen, J. (1998). Rubber hands ‘feel’ touch that eyes see. Nature, 391, 756.
Cash, T. F., & Brown, T. A. (1987). Body image in anorexia nervosa and bulimia nervosa: a review of the literature. Behavior Modification,
11, 487–521.
Cole, J., & Paillard, J. (1995). Living without touch and peripheral information about body position and movement: Studies with
deafferented Subjects. In J. L. Bermudez, A. Marcel, & N. Eilan (Eds.), The body and the self (pp. 245–266). Cambridge Mass: MIT
Press.
Ehrsson, H. H., Spence, C., & Passingham, R. E. (2004). That’s my hand! Activity in premotor cortex reflects feeling of ownership of a
limb. Science, 305, 875–877.
Eilan, N., Marcel, A., & Bermu
`dez, J. L. (1995). Self-Consciousness: An interdisciplinary introduction. In J. L. Bermudez, A. Marcel, &
N. Eilan (Eds.), The body and the self (pp. 1–28). Cambridge Mass: MIT Press.
Ernst, M. O., & Banks, M. S. (2002). Humans integrate visual and haptic information in a statistically optimal fashion. Nature, 415,
429–433.
Fourneret, P., & Jeannerod, M. (1998). Limited conscious monitoring of motor performance in normal subjects. Neuropsychologia, 36,
1133–1140.
Gallagher, S. (1986). Body image and body schema: a conceptual clarification. Journal of Mind and Behavior, 7, 541–554.
Gallagher, S. (1995). Body schema and intentionality. In J. L. Bermudez, A. Marcel, & N. Eilan (Eds.), The body and the self
(pp. 225–244). Cambridge Mass: MIT Press.
Gallagher, S. (2000). Philosophical conceptions of the self: implications for cognitive science. Trends in Cognitive Sciences, 4, 14–21.
Gallagher, S., & Cole, J. (1995). Body schema and body image in a deafferented subject. Journal of Mind and Behaviour, 16, 369–390.
Gardner, R. M., & Moncrieff, C. (1988). Body image distortion in anorexics as a non-sensory phenomenon: a signal detection approach.
Journal of Clinical Psychology, 44, 101–107.
Ghez, C., Gordon, J., & Ghilardi, M. F. (1995). Impairments of reaching movements in patients without proprioception. II. Effects of
visual information on accuracy. Journal of Neurophysiology, 73, 361–372.
Head, H. (1920). Studies in neurology. London: Oxford University Press, Vol. 2.
Holmes, N. P., Calvert, G. A., & Spence, C. (2004). Extending or projecting peripersonal space with tools? Multisensory interactions
highlight only the distal and proximal ends of tools. Neuroscience Letters, 372, 62–67.
Iriki, A., Tanaka, M., & Iwamura, Y. (1996). Coding of modified body schema during tool use by macaque postcentral neurones.
Neuroreport, 7, 2325–2330.
Jaynes, E. T. (1986). Bayesian methods: General background. In J. H. Justice (Ed.), Maximum-entropy and bayesian methods in applied
statistics (pp. 1). Cambridge, UK: Cambridge University Press.
Kording, K. P., & Wolpert, D. M. (2004). Bayesian integration in sensorimotor learning. Nature, 427, 244–247.
MacKay, D. J. C. (2003). Information theory, inference, and learning algorithms. Cambridge, UK: Cambridge University Press.
Maravita, A., Husain, M., Clarke, K., & Driver, J. (2001). Reaching with a tool extends visual-tactile interactions into far space: evidence
from cross-modal extinction. Neuropsychologia, 39, 580–585.
Maravita, A., Spence, C., Kennett, S., & Driver, J. (2002). Tool-use changes multimodal spatial interactions between vision and touch in
normal humans. Cognition, 83, B25–B34.
Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 74–78.
Paillard, J. (1999). Body schema and body image - a double dissociation in deafferented patients. In G. N. Gantchev, S. Mori, & J.
Massion (Eds.), Motor Control, Today and Tomorrow (pp. 197–214).
Paillard, J., & Brouchon, M. (1974). A proprioceptive contribution to the spatial encoding of position cues for ballistic movements. Brain
Research, 71, 273–284.
Pavani, F., Spence, C., & Driver, J. (2000). Visual capture of touch: out-of-the-body experiences with rubber gloves. Psychological Science,
11, 353–359.
Powers, P. S., Schulman, R. G., Gleghorn, A. A., & Prange, M. E. (1987). Perceptual and cognitive abnormalities in bulimia. American
Journal of Psychiatry, 144, 1456–1460.
Rossetti, Y., Rode, G., & Boisson, D. (1995). Implicit processing of somaesthetic information: a dissociation between where and how?
Neuroreport, 15, 506–510.
M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240 239
Sainburg, R. L., Poizner, H., & Ghez, C. (1993). Loss of proprioception produces deficits in interjoint coordination. Journal of
Neurophysiology, 70, 2136–2147.
Schilder, P. (1935). The image and appearance of the human body. New York: International Universities Press.
Schweizer, R., Braun, C., Fromm, C., Wilms, A., & Birbaumer, N. (2001). The distribution of mislocalizations across fingers demonstrates
training-induced neuroplastic changes in somatosensory cortex. Experimental Brain Research, 139, 435–442.
Tegenthoff, M., Ragert, P., Pleger, B., Schwenkreis, P., Fo
¨rster, A. F., Volker, N., et al. (2005). Improvement of tactile discrimination
performance and enlargement of cortical somatosensory maps after 5 Hz rTMS. PLoS Biology, 3, e362.
Tsakiris, M., & Haggard, P. (2005). The rubber hand illusion revisited: visuotactile integration and self-attribution. Journal of experimental
psychology. Human perception and performance, 31, 80–91.
Tsakiris, M., Hesse, M. D., Boy, C., Haggard, P., & Fink, G. R. (2006). Neural signatures of body ownership: a sensory network for
bodily self-consciousness. Cerebral Cortex.
Tsakiris, M., Prabhu, G., & Haggard, P. (2006). Having a body versus moving your body: how agency structures body-ownership.
Consciousness and Cognition, 15, 423–432.
Welch, R. B. (1978). Perceptual modification: Adapting to altered sensory environments. New York: Academic Press.
240 M. Costantini, P. Haggard / Consciousness and Cognition 16 (2007) 229–240
