Active self-touch restores bodily self-awareness 1 following disruption by “rubber hand illusion”

30 Bodily self-awareness relies on a constant integration of visual, tactile, proprioceptive, and motor


48
James' description of the "same old body, always there" (1), highlights that our own body is the 49 most familiar object in our mental life. However, it remains unclear how individual sensory 50 experiences related to the body give rise to a general sense of body awareness, and which types 51 of sensory signal dominate this process. An influential experimental approach to this question 52 involves the "Rubber Hand Illusion" (RHI) (2). During the RHI, the participant receives tactile 53 stimulation on her unseen hand, while seeing the same stimulation performed on a fake, rubber 54 hand. If the visual and tactile stimulations are synchronous, participants often report the feeling that 55 the rubber hand is theirs, and part of their own body: the fake hand is "embodied". Asynchronous 56 stimulation does not induce RHI and is commonly used as a control condition. The strength of the 57 illusion can be measured qualitatively through questionnaires (3). Some studies have used a 58 quantitative proxy measure by asking participants to report the position of their unseen hand (4).

59
Participants tend to perceive their hand as shifted towards the location where they saw the fake 60 hand. Crucially, this tendency is stronger in the synchronous than the asynchronous condition.

61
This "proprioceptive drift" measure quantifies the visual capture of proprioception in RHI.

62
The RHI demonstrates that the awareness of one's own body depends on integration of 63 multiple sensory input signals and shows remarkable plasticity when these signals change. Further, 64 the RHI paradigm has provided an important method for experimental studies of body awareness 65 more generally (3). The RHI seems to show that the visual representation of body parts is more 66 important for body awareness than truly somatic sensory inputs. Visual experience of one's body 67 is important for body part recognition and self-identification (5), but does not capture other key 68 features of body awareness, namely that one's body is sentient, and that the somatic sensations it 69 houses are intrinsically linked to one's sense of self.

70
In contrast, non-visual aspects of body awareness such as sentience and self-specificity 71 are strongly present in the act of self-touch. Self-touch is common in humans and many other 72 animals. It occurs in many contexts, such as grooming behaviours, affective self-stimulation, and 73 homeostatic thermoregulation. Self-touch also occurs incidentally in the process of many other 74 behaviours, notably feeding and bimanual object handling. A long tradition of phenomenological 75 interest in self-touch (6, 7) focusses on the integration of motor and tactile signals in self-touch (8-76 10) and speculates that the resulting sensorimotor contingencies may underpin the development 77 of a coherent, stable sense of a bodily self. This tradition emphasizes an intrinsic, or somatic 78 sentient basis for body awareness, in contrast with the external perspective on one's own body that 79 lies at the heart of the RHI. The relative importance of these two components in everyday body

91
We therefore investigated whether self-touch can restore a disturbance in bodily 92 awareness caused by RHI. This hypothesis is motivated by findings from clinical neuropsychology.

99
Therefore, we have combined these two key sources of information contributing to body 100 awareness and investigated their interaction. We first induced an alteration of body awareness 101 using the RHI, and then investigated whether we could mitigate the effects of this intervention, and 102 restore a normal sense of the body through self-touch. We hypothesized that active self-touch 103 immediately after inducing an RHI would reduce the strength of the rubber hand illusion, and act to 104 restore a 'normal', pre-existing awareness of the body, as previously shown in studies of 105 neurological patients.

106
We used two robotic arms in a leader-follower configuration to create an artificial self-touch 107 condition (22, 23) to test the influence of self-touch on body awareness. By moving the handle of 108 the leader robot with their right hand, participants were able to simultaneously feel a corresponding 109 tactile feedback on their left forearm (follower robot). Thanks to this mediated self-touch setup, and 110 contrary to everyday self-touch involving direct skin-skin contact, the right-hand's movement did 111 not provide any direct spatial information about the left-hand's position. We could therefore use the 112 perceived position of the left hand, and particularly the well-established proprioceptive drift 113 measure, to investigate the effects on body awareness of the RHI, of self-touch, and of the two in 114 combination.

115
We report a systematic series of three experiments based on power calculation, replication, 116 and preregistration (see Methods), to investigate the role of self-touch in bodily awareness. In 117 Experiment 1, we explored the hypothesis that self-touch has a restorative effect on an altered 118 body awareness caused by immediately-preceding RHI. In Experiment 2, we instead investigated 119 if self-touch has a protective effect on body self-awareness by asking the participant to perform a 120 self-touch stimulation immediately before inducing the RHI. Importantly, in both experiments we 121 also implemented passive self-touch conditions, in which the right hand of the participant was 122 passively moved by the experimenter while stroking the left forearm (22, 23). By comparing effects 123 of active and passive self-touch, we could therefore investigate whether motor signals play a 124 distinctive role in body awareness. Finally, Experiment 3 used a larger sample estimated a priori 125 by power analysis of results of Experiment 1 and included additional unimanual controls to 126 investigate the separate contributions of movement and touch alone to the combined effect of self-127 touch on body awareness.

130
In Experiment 1, we tested whether brief self-touch stimulation immediately after the induction of 131 the RHI could mitigate the effects of a previous RHI on body awareness. Participants (n = 16, based 132 on a power analysis on a pilot experiment, see Methods) made a baseline proprioceptive judgement 133 at the beginning of each trial by closing their eyes and pointing with their right hand to the location 134 of their left hand ( Figure 1A and Methods). Next, they received one of three visuo-tactile stimulation 135 conditions: synchronous, asynchronous, and no stimulation ( Figure 1B and Methods). The RHI 136 induction phase lasted for 60 s and was followed by one of three self-touch stimulations: active 137 self-touch, passive self-touch, and no self-touch. The self-touch stimulation was performed using

166
Pointing performance was measured using a webcam suspended above the workspace. B. RHI induction.

167
The experimenter sat opposite to the participant and used the two identical brushes to stroke homologous 168 points of the participants' left hand and the cosmetic glove either synchronously or asynchronously for one 169 minute. C. Self-Touch stimulation. Self-touch was performed immediately after (Experiments 1-3) or before 170 (Experiment 2) the RHI stimulation using two six-degrees-of-freedom robotic arms coupled in a leader-follower

182
In Experiment 2, we tested in a new set of participants (n = 16) whether a brief self-touch stimulation 183 performed before the induction of the RHI had a protective effect on the participants' bodily self-184 awareness, by reducing the susceptibility to the RHI. The experimental design was identical to 185 Experiment 1, except that the order of the self-touch stimulation and the RHI induction phases was 186 inverted. A protective effect on bodily self-awareness would be shown by a significant reduction of 187 the participants' proprioceptive drift after active or passive self-touch, but not after the control no

323
We also showed that the "restorative" effect of self-touch on body awareness was not due 324 to movement alone, nor to touch alone, but to the unique combination between these individual 325 signals that arises in self-touch. Ruling out 'unimodal' explanations of self-touch effects is important

394
One limitation of the present study is that we did not provide the skin-to-skin self-touch of everyday The final sample size for Experiments 1-2 (n = 16) was decided a priori based on a power analysis 405 on the results of a pilot study (n = 8) with a similar design. In the pilot study, the effect size for the 406 interaction of Self-touch Condition (active, passive) and Visuo-tactile Stimulation (Synchronous,

407
Asynchronous) was dz = -0,848, considered to be very large using Cohen's criteria (52). With an 408 alpha = .05 and power = .8, the projected sample size indicated to demonstrate a main effect of 409 type of movement was 13 participants (G*Power 3.1.9.2 software; (53). We nevertheless set a 410 sample size of n = 16 to counterbalance the blocks order between participants.

411
The final sample size for Experiment 3 (n = 28) was determined through a power analysis

433
The RHI was elicited using two identical brushes, following the classical RHI procedure (2, 434 55). The experimenter sat opposite to the participant and used the two brushes to stroke 435 homologous points of the participants' left hand and the cosmetic glove (between the middle and 436 the index finger) either synchronously (~1 Hz) or asynchronously (~1 Hz, 180° out of phase). The

437
RHI stimulation lasted for 60 s (55). To obtain an estimate of the participants' proprioceptive drift, 438 a webcam was mounted on the ceiling above the setup. The webcam provided a top view of the 439 participants' arms and was used to take accurate measurements of the pointing movements made 440 by the participants at the end of the RHI induction (see Figure 1). The coordinates of each 441 proprioceptive judgement were extracted by each picture through the ImageJ software 442 (http://rsbweb.nih.gov/ij/) and then converted from pixels to centimetres.

443
The sensorimotor self-touch stimulation was implemented using two six-degrees-of-444 freedom robotic arms (3D Systems, Geomagic Touch X, South Carolina, USA) linked in a 445 computer-controlled leader-follower system (see Figure 1). In this system, any 3D-movement of 446 the right-hand leader robot is reproduced by the follower robot with an estimated lag of ~2.5 ms finger. This setup allowed us to create a laboratory-equivalent of ordinary tool-mediated self-touch.

451
Importantly, the use of tool-mediated self-touch as opposed to skin-to-skin self-touch allowed us to 452 make sure that the right-hand movement did not provide any spatial information about the position 453 of the left-hand.

Experimental design
456 Experiments 1 and 2 tested, respectively, whether self-touch has a restorative vs. a protective effect 457 on bodily self-awareness. We reasoned that if self-touch has a restorative effect on bodily self-

539
To test our hypothesis that active self-touch has a restorative (Experiment 1) or protective 540 (Experiment 2) effect on bodily self-awareness, we ran two separate 3 (Visuo-tactile Stimulation:

549
To assess that the assumptions of the linear model were not violated, we checked that the 550 residuals of the models were normally distributed by visually examining Q-Q plots (see

697
The experimenter sat opposite to the participant and used the two identical brushes to stroke 698 homologous points of the participants' left hand and the cosmetic glove either synchronously or 699 asynchronously for one minute. C. Self-Touch stimulation.
Self-touch was performed 700 immediately after (Experiments 1-3) or before (Experiment 2) the RHI stimulation using two six-