Are blind individuals immune to bodily illusions? Somatic rubber hand illusion in the blind revisited

Multisensory awareness of one ’ s own body relies on the integration of signals from various sensory modalities such as vision, touch, and proprioception. But how do blind individuals perceive their bodies without visual cues, and does the brain of a blind person integrate bodily senses differently from a sighted person? To address this question, we aimed to replicate the only two previous studies on this topic, which claimed that blind individuals do not experience the somatic rubber hand illusion, a bodily illusion triggered by the integration of correlated tactile and proprioceptive signals from the two hands. We used a larger sample size than the previous studies and added Bayesian analyses to examine statistical evidence in favor of the lack of an illusion effect. Moreover, we employed tests to investigate whether enhanced tactile acuity and cardiac interoceptive accuracy in blind individuals could also explain the weaker illusion. We tested 36 blind individuals and 36 age-and sex-matched sighted volunteers. The results show that blind individuals do not experience the somatic rubber hand illusion based on questionnaire ratings and behavioral measures that assessed changes in hand position sense toward the location of the rubber hand. This conclusion is supported by Bayesian evidence in favor of the null hypothesis. The findings confirm that blind individuals do not experience the somatic rubber hand illusion, indicating that lack of visual experience leads to permanent changes in multisensory bodily perception. In summary, our study suggests that changes in multisensory integration of tactile and proprioceptive signals may explain why blind individuals are “ immune ” to the nonvisual version of the rubber hand illusion.


Introduction
We experience ourselves as single connected wholes with the multitude of bodily sensations automatically blending into unitary and continuously sensed bodily selves.This perception of one's own bodyi.e., the feeling of body ownershipdepends on combining information from various sensory sources [31,33,34].Sensory information from vision, touch, and proprioception plays an important role in the multisensory integration process that supports own-body perception [58,31,22,95,47,33,34]; in addition, other senses also contribute to own-body perception, such as visceral interoception [29,67], skin-based interoception [27,25,100,28], vestibular inputs [38,73], and auditory feedback [95,75,68].Vision, however, is often thought to play a leading role in the integration of these body-related signals due to its high reliability and high spatial acuity under good viewing conditions, thereby providing detailed information about the location, movement, postural state, and identity of one's body parts [50,47,62,64,100].For example, visual feedback is important for accurately localizing our limbs in space (e.g., [98]).Furthermore, even when vision is noninformative, it can influence tactile perception [62,16].However, it remains unclear how blind individuals experience their own bodies, given the lack of sensory feedback from this important modality.Blind individuals who were born blind and therefore never had a visual experience still develop a coherent sense of their body, although very little research has investigated this topic.Do blind people perceive their own body in a similar way as sighted people, or are there differences in terms of how the remaining sensory modalities are combined into a coherent representation of their own bodies?
The first signs of the capacity of multisensory integration are already present as soon as a few hours after birth [85].Nevertheless, its further development depends on the quantity and quality of sensory experiences until this ability finally reaches the adult form [42,49].The integration of touch and proprioception has been found to be altered in blind individuals, indicating that (lack of) visual experience affects both behavior and neural response to tactile-proprioceptive stimulation [24].Similarly, multisensory space near the body (peripersonal space) has been shown to be influenced by (lack of) early visual experience [21]; but see also: [79].Thus, there are indications that multisensory integration mechanisms are altered in blind individuals, although very little is known about such changes in the integration process responsible for the perception of their own bodies.
Multisensory bodily awareness can be studied experimentally using bodily illusions [33].One nonvisual illusion that can be used to investigate tactile-proprioceptive integration in both blindfolded sighted (e. g., [73,74]) and blind individuals (e.g., [72]) is the "somatic rubber hand illusion" [35].In this illusion, the blindfolded participant touches a right rubber hand with their left index finger while synchronous tactile stimuli are being delivered to their real right hand by the experimenter at the corresponding site.After a short period of such repeated synchronous touches (typically approximately 10 s on average; [35]), most participants start to experience a bodily illusion that they are directly touching their right hand, although the model hand feels harder and colder than normal, and the participants know that they are touching a rubber hand.Asynchronous touches (on the order of 500-to 1000-ms delays) break the illusion due to violation of the temporal rule of multisensory integration [93] and serve as a control in otherwise equivalent conditions [35].The illusion is typically quantified using questionnaire rating scales and a behavioral indirect test, the proprioceptive drift task, which assesses the degree to which people perceive their (real) hand to be located closer to the rubber hand after a period of experiencing the illusion [12,96,35].The proprioceptive drift is greater after the synchronous condition than after asynchronous control [35,75,74].The somatic rubber hand illusion occurs due to the dynamic multisensory integration process wherein the brain automatically attributes a common cause to the tactile and proprioceptive feedback from the left and right hands, thereby updating the central multisensory representation of the right hand, leading to illusory sensation of self-touch and changes in body ownership [35,33].
Two previous studies have investigated the somatic rubber hand illusion in blind individuals [72,69].The remarkable finding in both studies was that congenitally blind individuals seemed "immune" to the illusion, showing no significant questionnaire or proprioceptive drift effects supporting its elicitation.Petkova and colleagues [72] tested 10 congenitally blind individuals and 12 sighted controls and concluded that the blind group experienced "no illusion at all".Similar findings were observed in the study of Nava and colleagues [69], who compared 10 congenitally blind participants, 12 late blind participants, and 13 sighted controls and concluded that congenitally and noncongenitally blind individuals did not report experiencing the illusion.Interestingly, noncongenitally blind participants did show significantly larger proprioceptive drift toward the rubber hand than congenitally blind participants; in fact, proprioceptive drift in the noncongenitally blind group was not significantly different from the drift observed in the sighted group [69].Notably, it has been shown that early visual experience is important for developing the external spatial reference frames that are centered on the body, which then facilitate the integration of visual and tactile signals in a shared coordinate system [63,4,14].Whether this body-centered external spatial reference frame (see also [15,45]) is present from birth or rather later acquired has been a topic of investigation in children, pointing to its postnatal development [6,13,79,7].In fact, based on the results of studies with tasks involving the interaction between hearing and touch, there is evidence to suggest that early sensory experiences are essential in shaping its formation [83,84,65,43,66,11] and that blind people rather rely on an anatomically based reference frame [81,102].Thus, the findings from Nava and colleagues suggest that noncongenitally blind individuals might have acquired enough visual information to develop a mechanism that allows the spatial recalibration of hand position sense in response to the correlated tactile events but not sufficient early visual experience to allow "sighted-like" multisensory processing of bodily signals to trigger the changes in bodily awareness and illusory self-touch associated with full elicitation of the somatic rubber hand illusion.This view is further consistent with the findings in a single-case study of a person who had their vision restored after the age of two years following congenital blindness [61].The participant was asked to perform a tactile temporal order judgment task with an uncrossed and crossed hands posture and did not show a crossing effect, similar to what has previously been observed in congenitally blind individuals (e.g., [81]).These observations all indicate that early vision is necessary for the acquisition of a mechanism allowing for the automatic spatial remapping of touch as well as for the integration of tactile and proprioceptive signals in a common external spatial reference frame for multisensory bodily awareness.
Thus, the first aim of the current study was to replicate the results of Petkova and colleagues [72] and Nava and colleagues [69] regarding the claim that blind individuals do not experience the somatic rubber hand illusion.We thought a replication would be important given the relatively small sample sizes used in the original studies and the significant conceptual implications of the negative finding.Moreover, since the previous studies' main findings were based on nonsignificant results using frequentist statistics, we reasoned that a replication study should use Bayesian analysis to investigate the statistical evidence in support of the lack of illusion effects, i.e., support for the null hypothesis, to clarify if the illusion is genuinely lacking in the blind group or merely significantly weaker.Note that one of the blind participants in Petkova and colleagues' [72] study and three participants in Nava and colleagues' [69] study affirmed experiencing the illusion in the questionnaire, indicating that even if the group results were nonsignificant, there could be some illusion effects.To address the above points, we have conducted the largest study to date and included 36 blind participants (out of which 31 were congenitally blind) and 36 well-matched controls to revisit the question of whether blind individuals can experience the somatic rubber hand illusion.
The second aim of the study was to explore and possibly exclude two alternative factors that may contribute to the weaker or eliminated somatic rubber hand illusion in blind individuals.To conclude that changes in multisensory integration are the primary reason behind the lack of illusion in blind individuals, one needs to exclude the simpler explanation that this might be due to enhancements in unisensory sensory processing in blind individuals.It has been repeatedly shown that visual deprivation is associated with brain plasticity, which can involve enhancing other sensory modalities to compensate for the lack of vision, indicating an adaptive strategy (see [90]).Vision loss leads to enhanced tactile acuity [41], tactile estimation of object size [91], vibrotactile perception [103], cardiac interoception [76], and shorter somatosensory discrimination times [82], among others.The previous studies by Petkova and colleagues [72] and Nava and colleagues [69] ruled out the possibility that enhanced proprioceptive ability to localize one's real right hand (i.e., more precise and reliable proprioception) could explain the lack of somatic rubber hand illusion in blind individuals (more precise proprioception would work against the illusion; [19]) because behavioral testing revealed no significant differences in basic finger localization ability between blind and sighted individuals.However, it could be that enhanced tactile acuity in blind individuals would make them better at noticing subtle tactile incongruences between the tactile stimuli being manually delivered on the two hands, and increased sensitivity to detecting such tactile incongruence may work against the rubber hand illusion [104,18], which was not investigated in previous studies [72,69].Furthermore, we have recently found that blind individuals have higher accuracy in sensing their own heartbeats (cardiac interoception) than sighted individuals [76], which is noteworthy in the current context because a previous study reported a negative relationship between cardiac interoceptive accuracy and rubber hand illusion strength [97]; but see [51].It is thus possible that enhanced cardiac interoception in blind individuals may explain, at least partly, why they D. Radziun et al. report experiencing a weaker illusion.To address these points, we quantified tactile acuity and cardiac interoceptive accuracy in all our blind and sighted participants to explore to what degree enhancements in these unisensory channels may explain the lack of rubber hand illusion in blind individuals.
To summarize, this experiment aimed to re-examine the claims that blind individuals do not experience the somatic rubber hand illusion [72,69] and to investigate whether increased tactile acuity or cardiac interoceptive accuracy in blind individuals might be factors that contribute to the lack of illusion experience, in addition to reduced multisensory integration of tactile and proprioceptive information in a common external spatial reference frame.This study thus further explores the important but understudied topic of multisensory bodily awareness in blind individuals, which is informative not only for blindness research but also bears importance for studies on how the feeling of body ownership is developed and maintained in the absence of visual feedback.

Participants
Thirty-six blind and 36 sighted individuals (age range: 22-45 years, mean age: 33.42 years in the blind group, 33.19 in the sighted group; 19 men and 17 women per group; 1 left-handed individual in the sighted group) participated in the study.We did not conduct a formal a priori power analysis.All previous studies examining multisensory integration in blind individuals with the somatic rubber hand illusion paradigm involved smaller sample sizes (10 and 22 blind participants; [71,68], respectively).For each blind participant, a sighted, sex-and age-matched participant was recruited.All volunteers reported that they had no additional sensory or motor disabilities.The exclusion criteria included having a history of neurological or psychiatric conditions.
For all blind participants, blindness was attributed to a peripheral cause.The inclusion criteria were complete blindness or, at most, minimal light sensitivity with no ability to functionally use this sensation, as well as no pattern vision.Thirty-one participants were congenitally blind, 2 were early blind (early blindness is defined here as blindness acquired in childhood, 0-2 years after birth, as in [44,92]), and 3 were late blind.The results following the exclusion of late blind participants are described in Supplementary material for descriptive purposes; all main results remained the same.Blind participant characteristics are presented in Table 1.Handedness was determined using the Edinburgh Handedness Inventory [70] for the sighted participants and its adapted version for the blind participants [3].The same participants also took part in two other behavioral experiments that examined cardiac interoception and tactile perception [76,77].
The study was approved by the Jagiellonian University Ethics Committee.All participants gave written informed consent before the study and received compensation for their time; additionally, the travel costs of blind participants were covered.The experimenter read the documents to the blind participants and then used tactile markers to indicate the location where they were asked to provide a signature.

Experimental procedure
Before the start of the procedure, the participants were informed about the experimental setup.Participants from the sighted group were asked to wear a blindfold.Both groups were instructed to sit on a chair comfortably and place both their arms and hands on a table in front of them in a relaxed position, palms down, while a rubber hand (a life-sized cosmetic prosthetic right hand filled with hard plastic; Fillauer, Sollentuna, Sweden) was placed between their right and left hands, parallel with the participant's right hand (Fig. 1).The distance between the participant's right index finger and the index finger of the rubber hand was always 15 cm.Participants were allowed to make minor adjustments to the position of their hands on the tabletop to ensure that they were seated comfortably and any potential discomfort would not cause them to move their hands during experimental stimulation.The detailed instructions on how to place their hands on the table included information that they would be touching the rubber hand.Moreover, both groups were asked to manually explore the model hand before the experiment started so that they knew the object they touched was a rubber hand.Thus, all participants received identical knowledge about the experimental setup.
The participants, the experimenter, and the rubber hand all had on the same type of nitrile disposable gloves (Vileda, Weinheim, Germany) to keep the hands' surfaces as similar as possible.The experimenter held the participant's left hand in a steady grip throughout the procedure, with the index finger protruding and the remaining fingers loosely bent.The experimenter ensured the most consistent stimulation possible without touching any other part of the hand.The participant's right hand was comfortably resting on the table with the palm down.Then, the experimenter moved the participant's left index finger so that it touched the knuckle of the index finger of the rubber hand.The experimenter also touched the participant's knuckle of their real right hand either in temporal synchrony (the experimental condition to elicit the illusion) or temporal asynchrony (the control condition, which does not elicit the illusion; [35]).The taps were applied with gentle force.The duration of each tap was around 300 ms, and the interval between successive taps was approximately one second.The following pattern was used: three taps, one-second pause, two taps, one-second pause, and so on.The irregular pattern was chosen due to anecdotal reports of a stronger illusory experience following this type of stimulation [72].Each trial lasted approximately 60 s, so it consisted of approximately taps in both conditions.In the asynchronous (control) condition, stimulation was performed alternately so that when the participant's left index finger was touching the rubber hand, tactile stimulation of the participant's actual right hand was not performed, and vice versa.The conditions were counterbalanced and pseudorandomized for each participant to minimize possible order effects.Each trial was followed by a short 30-second pause during which participants were asked to move their hands.The pauses and instructed hand movements were also designed to eliminate any potential illusory experience carry-over between trials.The experimenter (DR) has been extensively trained to perform stimulation as consistently as possible regarding synchronicity, strength, duration, and frequency.The task was administered to all participants by the same experimenter.

Experimental design 2.3.1. Proprioceptive drift data
This part of the procedure consisted of 6 trials (3 synchronous and asynchronous) and was performed before the questionnaire was administered to ensure that the participants had as little explicit knowledge as possible of the expected illusion experiences.The procedures followed an established protocol [75,74].Immediately before and after stimulation, participants were asked to point with their left index finger toward the location of their right index finger.To do this, the experimenter placed the participant's left index finger on a ruler with a smooth surface glued to the table 10 centimeters above the rubber hand.The starting position was randomly selected between 40 and centimeters from the position of the right hand.The participant was asked to move his or her finger in one quick movement toward the perceived position of the right index finger.The pointing data were noted from the center of the index finger.This finger-pointing procedure was repeated twice in each trial: just before the trial started (pre-drift) and right after (post-drift).Proprioceptive drift was calculated as the difference in pointing before and after the stimulation (post-drift minus pre-drift)."Proprioceptive shift", which is a term we use in some of the analyses (see below), refers to the numerical difference in proprioceptive drift between the two conditions (proprioceptive drift in synchronous minus drift in asynchronous; [95,1]).

Questionnaire data
One trial of each condition (synchronous and asynchronous) was performed after the proprioceptive drift trials.Once the stimulation was completed, participants were instructed to answer a questionnaire in which they were asked to rate their experience during the stimulation using a 7-point Likert scale, ranging from '-3′ (I disagree very strongly) to '+ 3′ (I agree very strongly), with '0′ indicating "I am uncertain".The questionnaire consisted of five statements from the original article describing somatic rubber hand illusion [35]: one statement corresponded to the illusory feeling of self-touch (S1: "I felt as if I was touching my right hand with my left index finger"), and the remaining four statements aimed to control for any possible expectancy or task-compliance effects (S2: "It felt like I had more than one right hand", S3: "It felt like my right hand was larger than normal", S4: "It felt like my right hand was moving", S5: "It appeared to be I was not able to feel my own right hand").

Tactile acuity and cardiac interoception
To examine a potential relationship between the somatic rubber hand illusion and tactile and cardiac interoceptive abilities, we utilized a measure of tactile acuity, the grating orientation task [54], and one of cardiac interoceptive accuracy, the heartbeat counting task [88].The procedures followed the method described in Radziun et al. [77] and Radziun et al. [76], respectively, where they are described in detail.
In brief, the grating orientation task involved using eight plastic domes with parallel bars and grooves (JVP [Johnson-Van Boven-Phillips] Spatial Discrimination Domes, Stoelting, Inc. Wood Dale, IL) that were pressed on the distal pad of the right index finger for about 1.5 s.The gratings could be applied either horizontally or vertically, and participants had to identify their orientation using a two-alternative forced-choice method.The task was conducted in eight blocks, each with a different grating width, and each block included 20 randomized trials.The threshold for grating orientation was calculated using linear interpolation between grating widths where the correct responses were 75% (see [99]).Eight participants from the blind group and 12 from the sighted group were excluded from the grating orientation task data analyses because they could not perform the task beyond the expected level (75% accuracy).The results of the between-group comparison of the grating orientation threshold are described in Radziun et al. [77].
In the heartbeat counting task, the participant's heart rate was recorded using a Biopac MP150 BN-PPGED (Goleta, CA, United States) pulse oximeter attached to their left index finger and connected to a laptop with AcqKnowledge software (version 5.0).The participants were asked to silently count each heartbeat they felt in their body without manually checking or feeling their pulse (instructions adapted from the study of [39]).They were given six trials, with intervals of 25, 30, 35, 40, 45, and 50 s, followed by a 30-second break between intervals.After each trial, participants verbally reported the number of heartbeats counted.Each participant's accuracy score was computed using the following formula, yielding the overall interoceptive accuracy across all trials (see [88]): The interoceptive accuracy scores resulting from this transformation typically range from 0 to 1, with higher scores indicating more accurate counting of the heartbeats (i.e., a smaller difference between counted and actual heartbeats).Two blind participants who failed to perform the task (extremely low accuracy levels) were excluded from the analyses of the heartbeat counting task data.The results of the between-group comparison of the interoceptive accuracy are described in Radziun et al. [76].

Plan of statistical analysis
Our predetermined statistical criteria for determining illusion induction were a significantly stronger illusion rating (S1) in the synchronous condition than in the asynchronous condition; significantly stronger ratings on the illusion statements (S1) than on the control statements (S2-S5); a significantly stronger difference score between the illusion statement (S1) and the controls (average S2-S5) in the synchronous condition than in the asynchronous condition; and significantly greater proprioceptive drift in the synchronous condition than in the asynchronous condition.In addition, we expected an illusion to be associated with a positive median rating score on illusion statement S1, which would indicate that a majority of the participants affirmed the illusion.
A complete lack of illusion in the blind group would be indicated by nonsignificant results on the above tests, as well as Bayesian analyses supporting the null hypotheses.In addition, a lack of illusion would be indicated by a very low and negative median rating score on the illusion statement, suggesting that most participants deny experiencing the illusion.
We also planned to test the hypothesis of stronger illusion in the sighted group than in the blind group by comparing the illusion ratings (S1) in the synchronous condition, the S1 difference score between the synchronous and asynchronous conditions, and the difference in proprioceptive drift between the synchronous and asynchronous conditions between the two groups.

Data analysis
The proprioceptive drift data followed a normal distribution (Shapiro -Wilk test value always >0.05).The comparison of the two groups was conducted using a mixed analysis of variance (ANOVA), where the factor of "group" (blind vs. sighted) was treated as a between-subjects variable.Then, we used independent-samples t-tests to compare the groups.For within-group comparisons, we used paired-samples t-tests.The pre-drift pointing error and interoceptive accuracy data were not distributed normally (Shapiro-Wilk test value < 0.001 in the blind group), so these data were analyzed with nonparametric statistics.Due to being on an ordinal scale, the questionnaire data were tested nonparametrically with the Mann-Whitney U test or the Wilcoxon signedrank test.Consequently, Spearman's rho was used for correlation analyses with these data.The false discovery rate (FDR; [10]) was used to correct for multiple correlations.Two-tailed tests were used in all statistical analyses.Data exclusion criteria were established prior to data analysis.This study was not preregistered.
For the Bayesian analyses, the default Cauchy prior was used.BF 01 indicates support for the null hypothesis over the alternative hypothesis, while BF 10 indicates support for the alternative hypothesis over the null hypothesis (e.g., BF 01 = 6 means 6 times more support for the null hypothesis and BF 10 = 6 means 6 times more support for the alternative hypothesis).BFs within the range of 0.333 to 3 are generally regarded as inconclusive [53,59]).
The data were analyzed and visualized with RStudio software, version 1.4.1717, and the BayesFactor software package, version 0.9.12-4.2.The data are available at https://osf.io/qjrmu/.Raincloud plots were to visualize the data [2].

Questionnaire data
The sighted group's median ratings on the illusion statement (S1: "I felt as if I was touching my right hand with my left index finger") were positive in the synchronous condition and negative in the asynchronous condition, confirming that most participants experienced the illusion in the former condition and denied experiencing it in the latter (Fig. 2; Median Sync = 1, SD Sync = 2.282; Median Async = − 2, SD Async = 1.889, respectively).Looking at the same median rating for the illusion statement (S1) in the blind group, we observed that their ratings are clearly negative in both the synchronous condition and the asynchronous condition, which shows that blind participants denied experiencing the illusion in both conditions (Median Sync = − 3, SD Sync = 1.775;Median Async = − 3, SD Async = 1.977, respectively).When examining individual data points, the questionnaire data showed that 58.3% (21 out of 36) of the sighted participants and 11.1% (4 out of 36) of the blind participants affirmed the illusion, as quantified by a score ≥ +1 on questionnaire statement S1 in the synchronous condition.The experience of the illusion was denied by 30.5% (11 out of 36) of sighted participants and 75% (27 out of 36) of blind participants, as quantified by a score ≤ -1 on questionnaire statement S1.The control statements (S2-S5) in the synchronous condition were rated negatively by both groups (M Sighted = − 1.604, SD Sighted = 1.090;M Blind = − 2.063, SD Blind = 0.988).
Turning to the planned statistical analyses, first, we compared the illusion ratings on S1 between the synchronous and asynchronous conditions within each group.We found a significant difference between the conditions in the sighted group (V = 336, p < 0.001, 95% CI = Fig. 2. Questionnaire results for the synchronous and asynchronous conditions.The boxplots show the data using the median (represented by a thick black line) and quartiles (upper and lower ends of boxes).The vertical lines, i.e., the whiskers, indicate the minimum or maximum values that are within 1.5 times the interquartile range above and below the upper and lower quartiles.The observations that fall outside the vertical lines are considered outlier data points, with the furthest values being the minimum or maximum values in the data.1.500-4.000,BF 10 = 203.922)but, importantly, not in the blind group (V = 45, p = 0.972, 95% CI = − 2.000-2.500,BF 01 = 5.565).Moreover, when we compared the response on S1 with mean responses on control statements S2-S5 in the synchronous condition, we found a significant difference in the sighted group (V = 458, p < 0.001, 95% CI = 1.500-3.250,BF 10 = 6775.648)but not in the blind group (V = 153, p = 0.389, 95% CI = − 0.625-1.500,BF 01 = 3.407).These results show that a significant somatic rubber hand illusion was only elicited in the sighted group and not in the blind group, and Bayesian evidence indicated a lack of illusion in the blind group.
Finally, we examined whether the pre-drift pointing accuracy before the experimental manipulation, i.e., the baseline ability to localize one's own finger before the stimulation trials commenced, was different between the blind and sighted participants.Importantly, the groups did not differ significantly (M Sighted = 0.340, SD Sighted = 3.888; M Blind = − 0.688, SD Blind = 4.936; W = 680.5,p = 0.714, 95% CI = − 1.250-1.917,BF 01 = 2.722), which suggests a similar basic proprioceptive ability to localize the finger in the two groups (Fig. 4).We also conducted the Brown-Forsythe test to examine if there was a significant difference in the inter-individual variability of pre-drift pointing accuracy values between the two groups, and our analysis did not reveal a significant effect (p = 0.330).
Finally, to examine whether the questionnaire ratings of the illusion and proprioceptive drift are related, we examined the correlation between the illusion statement S1 in the synchronous condition and proprioceptive drift in the synchronous condition.We found a significant correlation in the sighted group (ϱ = 0.430, p = 0.008, 95% CI = 0.118-0.665,BF 10 = 3.794) but not in the blind group (ϱ = 0.074, p = 0.667, 95% CI = − 0.261-0.393,BF 01 = 2.429).We did not find a significant correlation between the difference in the illusion statement rating in the synchronous and asynchronous conditions and proprioceptive shift in synchronous compared to asynchronous conditions either in the sighted (ϱ = 0.206, p = 0.227, 95% CI = − 0.131-0.501,BF 01 = 1.030) or in the blind group (ϱ = 0.236, p = 0.166, 95% CI = − 0.100-0.524,BF 01 = 2.183).

Discussion
Very little is known about how blindness influences the experience of one's physical self.This study expands the previous attempts to investigate this issue using the somatic rubber hand illusion with the largest sample size to date (36 participants compared with 10 in the study by Petkova et al. [72] and 22 in the study by Nava et al. [69]) and Bayesian statistics providing support for null and alternative hypotheses.We found little evidence that blind individuals can experience the somatic rubber illusion but replicated the effect in the sighted group.Notably, Bayesian analysis indicated that the statistical evidence in favor of the blind group not subjectively experiencing the illusion was 5.56 times more likely than that they did (based on the illusion statement in the questionnaire ratings).Furthermore, the proprioceptive drift data suggested a 3.9 times greater likelihood of lack of an illusion-related change in hand position sense toward the rubber hand.These observations support the hypothesis of altered multisensory processing of bodily sensory signals in blind individuals and suggest that they are "immune" to the somatic rubber hand illusion.
The current findings thus replicate the results reported by Petkova and colleagues [72] and Nava and colleagues [69] and underscore fundamental differences in somatic rubber hand illusion susceptibility between blind and sighted individuals.Collectively, the data from the current study and these earlier investigations suggest a profound distinction in how tactile and proprioceptive signals from the two hands are processed in these two groups.In the blind group, the synchronous

Table 2
Correlations between the questionnaire statements and interoceptive accuracy across groups.tactile events experienced on the pulp of their left index finger touching the rubber hand and on their relaxed real right hand (applied by the experimenter) remain segregated and are perceived as two spatially distinct events even after one-minute exposure to repeated correlated stimulations.In contrast, in the sighted group, such correlated tactile stimuli elicit the somatic rubber hand illusion in approximately 10 s on average [35], during which people experience a single event of touching one's own right hand directly with the left index finger, while the object they actually touch is a rubber hand.The illusion emerges from an automatic perceptual decision process where tactile and proprioceptive information from the two hands are combined into a coherent multisensory experience of direct self-touch [86,20].Critically, such a dynamic multisensory integration process requires that tactile information and proprioceptive information are processed in a common external spatial reference frame [35,63,4].The rubber hand illusion involves such an external spatial reference frame centered on the hand [46,14].However, in blind individuals, and congenitally blind individuals in particular, who made up the vast majority of the current sample (31 out of 36), the remapping of tactile signals into an external spatial reference frame does not occur (or at least not to the same degree); instead, they rely more on an anatomical reference frame for localizing tactile events (see [52,83,84]).The limited remapping mechanism to quickly compare and potentially integrate tactile, proprioceptive, and other bodily-related sensory signals from limbs in the peripersonal space may thus prevent the (incorrect) combination of tactile and proprioceptive signals that correspond to the illusion.We think this is the most plausible interpretation of the current findings, consistent with the previous literature and well-supported theoretically; however, other possible explanations deserve further consideration.
In probabilistic models of multisensory integration, the automatic decision to fuse sensory signals from different modalities is solved by a mechanism known as "causal inference".In this process, the brain deduces the probability that sensory signals originate from a shared source, considering factors such as simultaneity, spatial proximity, temporal correlation, sensory uncertainty, and prior perceptual experiences [57,87]; this model has been extended to the rubber hand illusion [86,36,20].It is thus relevant to consider that stronger prior expectations based on previous sensory experience influence multisensory integration and the rubber hand illusion [57,87,86,20].However, both blind and sighted individuals have a lifetime of experiencing self-touch, so it is not obvious why blind individuals would have weaker prior expectations than sighted individuals who correlated tactile signals indicate self-touch based on previous experience.The somatic rubber hand illusion uses passive finger movement, so we can exclude that differences in motor intentions or sensorimotor predictions related to self-generated movements influenced our findings, which can otherwise affect the somatic perception of self-generated touches (e.g., [56]).Furthermore, in the Bayesian causal inference framework, greater sensory uncertainty, e.g., less precise and reliable proprioceptive signals, promotes the rubber hand illusion [86,20].However, there was no evidence suggesting more accurate hand proprioception in blind participants in our data (pre-drift pointing) or previous studies [72,69].Sensory uncertainty was shown to influence the multisensory combination process in cases of small temporal [20] or spatial incongruences [36] but does not fundamentally change the illusion strength in the case of a fully synchronous multisensory stimulation ("widening" or "sharpening" the temporal or spatial windows of integration somewhat); thus, even if differences in tactile or proprioceptive sensory signal reliability (accuracy) exist between sighted and blind individuals, it is unclear if this could explain the profound difference in illusion elicitation between the two groups that we observed in the current data.
Importantly, not only the experience of somatic rubber hand illusion but also other multisensory illusions seem to be altered following blindness.Sounds do not affect the performance of congenitally blind individuals as much as sighted individuals in the "double-tap illusion", an auditory-tactile equivalent of a well-known double-flash illusion [89].Specifically, they were shown to be more accurate in counting the number of touches when task-irrelevant tones were presented [52].As hypothesized by the authors, the reasons for this can be twofold: it could be the case that the participants, all professional Braille readers, were more accurate in counting the taps thanks to their superior tactile abilities.On the other hand, their performance could also be a result of  altered multisensory integration, which would lead to less interplay between remaining sensory modalities (see also [60]).Another interesting example of altered performance of blind individuals in a multisensory illusion comes from the study on the so-called parchment-skin illusion, an experimental paradigm in which the sound that accompanies rubbing one's palms together is manipulated, leading to a feeling of the skin being rough and dry [55].Strikingly, in stark contrast with sighted participants, who experienced the illusion constantly, blind individuals did not report any change in their experience following the auditory manipulation [17].In another experiment, using a sensory substitution device (see [5]), here translating vision to audition, it was found that blind participants are not susceptible to the Ponzo illusion, a geometrical illusion in which a pair of converging lines alters the perception of two other horizontally placed bars.Interestingly, in the same setup, the illusion occurred in sighted individuals, which suggests that developmental vision is needed to experience the illusion, even when visual content is translated to another sensory modality [78].Importantly, however, in the case of a unisensory tactile illusion, the Aristotle Illusion (see [8,9]), in which the participants are asked to touch a single sphere with crossed fingers and commonly report perceiving not one but two objects, no differences between blind and sighted individuals have been observed [69].Taken together, the previous literature and the current findings suggest that lack of visual experience leads to changes in multisensory processing of bodily-related information and reduced illusion susceptibility in different types of multisensory illusion phenomena, including the bodily illusion studied herein.This may suggest more profound differences in multisensory processing in blind and sighted individuals that go beyond a selective "impairment" of remapping touch into an external spatial reference and speak in favor of a more general processing policy to segregate rather than combine different sensory signals.
It has been reported that participants with higher interoceptive accuracy, as measured with the heartbeat counting task, reported a weaker sense of limb ownership and showed reduced proprioceptive drift in the classic visual rubber hand illusion paradigm [97].However, other studies have failed to replicate this finding [29,51,23].Another study found that interoceptive accuracy positively correlates with proprioceptive drift in a virtual hand illusion experiment exploring the interaction between internal and external sensory signals and body ownership [94].Thus, while the literature on this topic presents mixed findings, and we did not replicate Tsakiris and colleagues [97] observation of a negative correlation between cardiac interoceptive accuracy and proprioceptive drift, we did observe a negative relationship with the illusion questionnaire ratings in the sighted group, although it did not remain significant after the FDR correction.The hypothesized mechanism behind this observation, as per Tsakiris and colleagues [97], is that given the importance of interoception for bodily awareness [30], individuals with lower interoceptive accuracy "compensate" for the lack of this ability with more processing resources allocated to the exteroceptive aspects of multisensory integration (see [97]).In the blind group, no relationship between interoceptive accuracy and either illusion measure was observed.
Interestingly, however, blind individuals are significantly better than sighted individuals at sensing their own heartbeats [76].It might be the case that in the permanent absence of 'leading' exteroceptive, visual information, blind individuals may rely more on interoceptive signals for their sense of body ownership (see [48,70]) than sighted individuals.However, it is unclear how sensory signals from the heart would directly contribute to the current illusion, which primarily involves tactile and proprioceptive signals from the two hands.Moreover, recent evidence indicates that cardiac interoceptive accuracy does not correlate with accuracy measures of other visceral and skin-based interoceptive submodalities [37,40,29].This suggests that cardiac interoceptive accuracy is probably not an indicator of general sensitivity to all bodily signals.
An alternative interpretation, consistent with the data, proposes that individual differences in cardiac interoceptive accuracy modulate the reported strength of the illusion once experienced, as in the sighted group.However, in cases where the illusion is completely absent, such as in the blind group (due to another mechanism, i.e., reduced multisensory integration of tactile and proprioception information in a common external spatial reference frame as discussed above), this modulation becomes irrelevant.In other words, neuroplasticity after blindness may impact both the multisensory mechanisms mediating somatic rubber hand illusion and cardiac interception, albeit independently.Furthermore, as stated, we did not observe a significant correlation between proprioceptive drift and interoceptive accuracy in the sighted group, which was a critical observation in the study by Tsakiris and colleagues [97].This indicates that the association between the somatic rubber hand illusion and cardiac interoception in sighted individuals pertains only to subjective reports of the illusion.Therefore, we think that it is unlikely that enhanced cardiac interoceptive accuracy is the main reason why blind individuals do not experience the somatic rubber hand illusion, but it may be a relevant contributing individual differences factor that influences subjective illusion strength in sighted individuals.
Notably, the contribution of tactile acuity in this tactileproprioceptive illusion seems limited, as we did not observe correlations between illusion measures and the grating orientation task in either the blind or sighted groups.This makes it unlikely that the increased tactile acuity of blind individuals observed among the participants of this study (see [77]) could explain the lack of illusory experience in this group.Furthermore, as stated, we replicated the results from Petkova and colleagues [72] and Nava and colleagues [69], showing no difference in the basic ability of blind and sighted participants to localize their index finger, which speaks against the possibility that superior proprioceptive ability in blind individuals could explain the lack of somatic rubber hand illusion in this group.
In conclusion, in this study, we found that blind individuals do not experience the somatic rubber hand illusion.Our results thus underscore fundamental differences in how blind and sighted individuals integrate or segregate sensory signals from the body for the maintenance of a coherent body representation and sense of body ownership.Further research on bodily awareness following blindness could expand our understanding of the neural processes involved in perceiving the body as one's own without visual experience.Such studies could not only explore the differences in bodily perception in blind and sighted individuals but also help to better understand the contribution of vision to the development and maintenance of the bodily self.

Funding
The Polish National Science Centre (NCN; grant no: 2018/30/A/ HS6/00595 awarded to MS), Swedish Research Council (VR; grant no: 2017-03135 awarded to HHE) and Göran Gustafssons Stiftelse (prize awarded to HHE) supported this work.The funding sources were not involved in the study design, collection, analyses, and interpretation of the data or writing of this manuscript.

Declaration of Competing Interest
The authors declare no competing interests.D. Radziun et al.

Fig. 1 .
Fig.1.Illusion induction.The setup used to induce the somatic rubber hand illusion.From left to right: (i) the experimenter's left hand guiding the left index finger of the participant, (ii) the participant's left hand, (iii) the rubber hand, (iv) the experimenter's right hand touching the participant's right hand with the index finger, (v) the participant's right hand.For illustrative purposes, the participant is wearing orange gloves, the experimenter green gloves, and the rubber hand a yellow glove.

D
.Radziun et al.

D
.Radziun et al.

Fig. 5 .
Fig. 5. Correlation between questionnaire ratings on S1 in the synchronous condition and interoceptive accuracy in the sighted (A) and blind (B) groups.

Table 1
Blind participant characteristics.