Timing disownership experiences in the rubber hand illusion

Rubber hand illusion

In daily life we assume that hands belong to us when they are connected to our bodies in the usual way, and that artificial hands not so connected cannot be ours. But Tastevin (1937) produced evidence to suggest that artificial hands could be experienced as belonging to self when participants simply look at an artificial hand that is appropriately aligned with body position and posture (cf., Schaefer, Heinze, & Rotte, 2009; Ferri, Chiarelli, Merla, Gallese, & Costantini, 2013). This belonging or ownership illusion was replicated by Botvinick and Cohen (1998) under experimental conditions that incorporated tactile sensation. The phenomenon, what has come to be known as the rubber hand illusion (RHI), has since been further replicated, by various techniques, multiple times (for reviews, see Ehrsson, 2009, 2012; Makin, Holmes, & Ehrsson, 2008; Tsakiris et al, 2010; Tsakiris 2011), even in non-humans (Wada, Takano, Ora, Ide, & Kansaku, 2016).

In the standard version of RHI, an artificial hand is experienced as belonging to self when a participant observes strokes applied with a paint brush to a rubber hand while, in synchrony, strokes are also being applied to the occluded, biological hand. When strokes are applied asynchronously, the illusion either fails to occur or is less vivid, although individual sensitivity to perceiving asynchrony varies as a function of a temporal binding window (Costantini et al., 2016). Although the RHI often exhibits an elusive quality (e.g., Hohwy, 2013), the subjective experience of ownership can be investigated by means of questionnaires that generate data amenable to psychometric analyses (e.g., Longo, Schuur, Kammers, Tsakiris, & Haggard, 2008). Indeed, because the illusion can develop so quickly, amazed participants often spontaneously report that the rubber hand “comes alive,” that it belongs to self (Ehrsson, 2012). This startling experience of ownership for a “hand” can even be elicited when participants are looking at nothing but empty space, provided that certain minimal conditions are satisfied, in particular the synchrony of stroking (Guterstam, Gentile, & Ehrsson, 2013).

Is there also a disownership experience?

Although most investigations of the RHI focus on the ownership experience, recently some tantalizing evidence has been adduced to suggest that participants also experience alienation from, or disownership for, the biological hand. For example, Longo et al. (2008; cf., Preston, 2013) discovered that participants deny that they experience having three hands, while they do feel that the biological hand disappears. In a follow-up study that drew upon the same data set (Longo, Schuur, Kammers, Tsakiris, & Haggard, 2009), they adduced further evidence suggesting that disownership can be a significant component of the illusion. Indeed qualitative analyses of the RHI that employed interpretive phenomenological analysis also suggest that participants experience disownership for the biological hand (Lewis & Lloyd, 2010; Moguillansky, O’Regan, & Petitmengen, 2013). A possible mechanism for inducing disownership is the attenuation of neuronal responses in select multisensory regions: that is, these attenuated responses prevent the integration of multisensory signals for the biological hand (Gentile, Guterstam, Brozzoli, & Ehrsson, 2013).

Additional support for the claim that disownership experiences occur during the RHI derives from investigations into homeostatic, sensory, and immunological processes. Moseley et al. (2008), for example, have shown that the illusion engages homeostatic processes in such a way that the skin temperature of the biological hand decreases when participants experience ownership for the rubber hand, apparently because experienced disownership causes a selective reduction in blood flow (see also Hohwy & Paton, 2010; Kammers, Rose, & Haggard, 2011; Moseley, Gallace, & Spence, 2012). These findings, however, should be taken with a grain of salt for, in personal communication, J. Hohwy and others have noted that this experiment is not replicated easily; indeed, our group also failed to detect significant decreases in skin temperature.¹ Most importantly, one study by Rohde, Wold, Karnath, and Ernst (2013) demonstrated that the cooling effect may be present in both the synchronous and the control condition, suggesting that the temperature effect may not be a direct result of the illusion.

In a separate experiment concerning sensory processing, Moseley et al. (2008) have also shown that RHI vividness correlates with a diminished ability to accurately determine the sequence in which tactile stimuli are delivered to the index fingers of the left and right biological hands. They attribute this diminished ability to a decrease in weighting given to tactile information from the biological hand. And Barnsley et al. (2011) have shown that the vividness of the illusion correlates with elevated histamine activity in the biological hand. This elevation of activity they interpret as suggesting that the biological limb is being “rejected.” These findings converge in lending support to the view that experienced disownership might be an important component of the RHI.

But the claims about disownership have not gone unchallenged; investigations of limb disownership have not produced conclusive results (Guterstam & Ehrsson, 2012). Schutz-Bosbach, Tausche, and Weiss (2009), for example, designed an experiment aimed at determining whether or how conceptual interpretations of visual and tactile sensations influence the RHI. Their findings led them to infer that the biological hand is retained, not disowned. Folegatti, de Vignemont, Pavani, Rossetti, and Farne (2009) reached similar conclusions. They employed prismatic lenses to create a visual-proprioceptive conflict, albeit one that does not engender the experience of ownership for the rubber hand. They then compared results from this test to results from a standard RHI that includes ownership. Because in both cases they detected somatosensory changes in the biological hand (viz., a slowing of reaction time to tactile stimuli), they suggest that these objective changes result not from disownership of the hand, but from the brain not “knowing” where the biological hand is. Finally, they venture the categorical claim that “there is no experimental setup that can artificially induce the explicit sensation of disownership of one’s own hand.”

de Vignemont (2011) has also challenged the claim that disownership experiences occur; indeed, she too categorically asserts that there is “no evidence” to support such claims. Her concerns are a mix of the conceptual and the experimental. But she has clearly articulated several issues that contribute to confusion over what has been learned from RHI experiments, and what can be learned. She agrees with the commonplace observation that reports of the experience are variable and elusive, and then she underscores the possible role that a widespread failure to distinguish between the “experience of” and the “judgment of” ownership might play in making this illusion so difficult to nail down. She also points out that experimental paradigms tend to rely excessively on questionnaires, tend to pay inadequate attention to updating of “online representations of bodily properties”, and too often neglect observed dissociation between what one experiences and proprioceptive drift, the most frequently employed objective measure of ownership and disownership.

Aims of this study

Within the framework of the RHI paradigm, there is ample evidence that an illusion of ownership occurs for the rubber hand. What remains unclear is whether, within that same framework, an illusion—a conscious experience—of disownership can occur for the biological hand. Moreover, if there is in fact such an experience, it remains unclear how stable or how robust it might be.

Our experiments focus on the contents of conscious experience within the RHI, devoting special attention to disownership. For this reason we designed a series of experiments that began by first distinguishing clearly between what is experienced and what is inferred or judged to be the case, while emphasizing that our concern is with the former. Second, given the importance of subjective report to our experimental aims, we drew upon multiple, previously employed questionnaires, seeking to determine whether responses to similar content that was differently phrased would converge. Third, given the seeming elusiveness of the RHI as well as the lack of adequate information concerning how it is experienced online, we employed novel, online proxies—onset time (OT) and illusion duration (ID). Finally, because some evidence suggests that laterality is a factor relevant to the emergence of disownership experiences, we concentrated on the left rather than on the right hand.

Alterations of the standard RHI paradigm suggest that ownership experiences can be retained for the biological hand (Ehrsson, 2009; Schaefer et al., 2009; Newport, Pearce, & Preston, 2010; Guterstam, Petkova, & Ehrsson, 2011). Our aim is not to challenge these findings. Instead, our aim is to test the categorical denials that disownership experiences can occur within the RHI, by collecting data on conscious experience as it is reported, both while the experience is ongoing and after the fact.

Distinctive methodological approach

A commonly used behavioral proxy for the RHI is proprioceptive drift. But several studies have raised doubts about the reliability of drift as an indicator of the ownership experience for the rubber hand (Rohde, Luca, & Earnst, 2011; Zopf, Savage, & Williams, 2010; Ehrsson, Spence, & Passingham, 2004; Fiorio et al., 2011; Kammers et al., 2009). A more recent study, however, seems to suggest that although drift and ownership experience are positively correlated, crucially, it is the conscious experience that seems to be causing drift, not the reverse (Abdulkarim & Ehrsson, 2016). This is one of the factors motivating our focus on the conscious contents of the RHI.

In view of the seeming unreliability of proprioceptive drift, our focus on conscious experience, and the evidence suggesting that conscious experience causes drift, we opted not to employ this behavioral proxy. Instead, we used two online temporal measures: OT and ID. A few prior studies have recorded OT (Ehrsson et al., 2004; Ehrsson, Wiech, Weiskopf, Dolan, & Passingham, 2007; Lloyd, 2007), but previously the temporal dimension—whether OT or ID—has never been employed as a proxy for the RHI. Peled, Ritsner, Hirschmann, Geva, and Modai (2000), despite not using OT as a proxy, did notice that OT and illusion strength correlate in patients with schizophrenia. We extrapolated from this incidental finding, as well as the observation that the illusion can begin as early as 10–20 s after the start of synchronous stimulation (Lloyd, 2007; Ehrsson et al., 2004), and designed three experiments that explicitly treated the temporal dimension, both OT and ID, as potential online proxies for healthy participants; OT was used for the first two experiments, and ID was used for the third. Although OT might seem problematic, given that in some cases the illusion can begin at the very moment the participant observes the rubber hand, previous data nevertheless suggest that waiting time tends to range between 5 and 116 s, with 27 s being the median (Slater, Perez-Marcos, Ehrsson, & Sanchez-Vives, 2009).

The first experiment aimed at ascertaining whether our claim that OT and illusion strength correlate in healthy subjects could be confirmed, even without first suggesting to the participants what type of conscious content was under investigation. The second experiment applied findings from the first in order to directly test the disownership claim by combining a psychometric approach to post hoc questionnaire data with online OT data. Finally, to further confirm findings from the second experiment, the third experiment employed a different questionnaire, a different scale, and a different online proxy, albeit one that also involves the temporal dimension ID.

Surely even naïve participants could discern from the experimental context that some experience relevant to the rubber hand might be expected. But it is not self-evident that this experience will be of ownership or disownership illusions. As a matter of fact, individual difference in how the RHI is experienced is to be expected (Haans, Kaiser, Bouwhuis, & Ijsselsteijn, 2012). It could just as easily be only touch referral—the feeling of touch on the rubber hand—that is experienced, absent the richer phenomenology involving ownership, because touch referral is actually the illusion’s most distinctive perceptual event (Ehrsson, 2012).

Accordingly, in order to avoid suggesting any specific conscious content, for the first experiment we included a common vertical–horizontal illusion in our instructions, both to illustrate what an illusion is and to distinguish illusion from inference or judgment. Participants were told to indicate OT the moment they experienced an illusion of some kind. We did not divulge to them that the experiment was an investigation of ownership–disownership experiences. Instead, to determine whether participants experienced such illusions, we adopted items from the Longo et al. (2008) questionnaire that had previously been identified as relevant to ownership for the rubber hand and disownership for the biological hand.

Emphasizing the experience–inference distinction is important for this first experiment. A principal reason is that prior results seem to strongly suggest that an ownership experience for the rubber hand can occur. But the contentious point is whether a complementary disownership experience can occur for the biological hand. The worry is that some participants who have an ownership experience might then infer or judge that their biological hand must have been disowned, even if they had no such experience. Avoiding conflation of experience and inference is crucial to clearly determine the conscious contents of the RHI.

Having in the first experiment established the reliability of OT as an online, in-the-moment proxy for the RHI in healthy participants, and having established that participants reported ownership and disownership experiences without either having been suggested to them prior to the experiment, in the second experiment we focused even more directly on the experience of disownership. Again we used the Longo et al. (2008) questionnaire items in conjunction with OT measurements. But here participants were instructed to use OT to indicate the moment of onset of either disownership or ownership. For the third experiment, we sought additional confirmation of findings concerning use of a temporal, online proxy, by using ID, and additional confirmation of subjective report by using an alternative questionnaire and scale (Preston, 2013).

Throughout we focused only on the left hand. The principal reason is that right hemisphere tactile activation tends to evince a more vivid illusion of ownership for the rubber hand (Ocklenburg, Ruther, Peterburs, Pinnow, & Gunturkun, 2011). On the assumption that experiences of ownership and disownership are intrinsically related, we conjectured that right hemisphere tactile activation also tends to evince a more vivid illusion of disownership. In fact, certain pathologies suggest that disownership experiences are lateralized in just this way (Vallar & Ronchi, 2009; Karnath & Baier, 2010; Giummarra, Bradshaw, Hilti, Nicholls, & Brugger, 2012).

Significance of this study

Schutz-Bosbach et al. (2009), Folegatti et al. (2009), de Vignemont (2011), and others adduce evidence to infer that standard RHI induction procedures do not or cannot induce disownership experiences for one’s biological hand. Our investigations are designed to test their views (1) by focusing on the conscious content of the RHI; (2) by emphasizing the difference between experience and inference or judgment; (3) by collecting data online, while participants are having the experiences, and after the fact, using dual measures for both; and (4) by focusing on just one hand, the left. Proceeding in this way we are able to marshal support for the view that not only can disownership occur during the RHI, but also that disownership is consciously experienced.

Overview and hypothesis

Two measures were used to determine whether standard RHI induction procedures bring about disownership experiences. First, the OT was recorded while biological and rubber hands were being stroked. Second, after completion of the experiment, a questionnaire designed to probe ownership of the rubber hand and disownership of the real hand was administered. If standard RHI induction procedures engender disownership, the questionnaire should evince both ownership of the rubber hand and disownership of the biological hand, while the strength of both should positively correlate with OTs. If standard RHI induction procedures do not engender disownership, however, it should neither be evinced by the questionnaire nor correlate with OTs.

Methods

Experimental setup

Participants were individually tested in a small, quiet room. The experimenter stood in front of the participant, who was seated with both hands placed on a table top. Prior to beginning, participants were shown a vertical–horizontal illusion and were advised that the test was designed to see whether or not they would experience an illusion, of some unspecified kind. The purpose of showing them this illusion was to illustrate the difference between what is experienced and what is judged to be the case. As has been emphasized in the previous section, given our special focus on the conscious contents of the RHI, along with skepticism expressed by some investigators that the experience of disownership even occurs, we found clarification of this distinction to be crucial to our design.

First, participants were asked to insert their hands into a black cardboard tube, so that they would be hidden from view. Second, a towel was placed over the tube and in such a way that it concealed participants’ elbows and forearms, as well as the space separating the rubber hand from the body. Third, the experimenter proceeded to use two paintbrushes to stroke corresponding fingers of the rubber and biological hands, at an approximate rate of one stroke per 2 s, and continuing for a period of 5 min (Fig. 1). Because evidence suggests right hemispheric dominance for the experience of body ownership (Ocklenburg et al., 2011), strokes were applied to the real left hand and a corresponding left, rubber hand. The left hand was used for all three experiments. Two foot pedals connected to a computer were placed under the participants’ feet to measure OTs (Fig. 2a).

Onset time for the illusion

We used mean ± 2.5 standard deviations (SD) as cut-off criteria and no participants had to be excluded in follow-up analyses. By measuring the elapsed time from the moment that the left foot pedal was pressed (when stroking of hands begins) to the moment when the right foot pedal was pressed (when participants begin experiencing the RHI), we obtained the OT of the RHI for each participant. For trials when participants did not step on the right pedal before cessation of the stroking, OT was recorded as 300 s (duration of stroking); this ensured that no cells were left empty. (In Experiment 3 we show that this methodological compromise did not contaminate our data.) A one-way repeated measure analysis of variance (ANOVA) on OT showed a significant difference between synchronous and asynchronous stroking (F(1, 31) = 34.193, p < 0.001, η_p ² = 0.5245). In other words, participants experienced RHI more quickly when the stroking was synchronous than when it was asynchronous (Fig. 3a).

Questionnaire

We categorized items 1–5 as ownership items, and 6–10 as disownership items. We then conducted a two-way repeated measure ANOVA concerning the synchronous–asynchronous difference and ownership type—that is, ownership versus disownership items (Fig. 3b). Both the main effects of the synchrony–asynchrony difference (F(1, 31) = 59.907, p < 0.001, η_p ² = 0.6590) and ownership type (F(1, 31) = 48.643, p < 0.001, η_p ² = 0.6108) were significant. Moreover, interaction between the synchrony–asynchrony difference and ownership type was also significant (F(1, 31) = 6.876, p < 0.05, η_p ² = 0.1815). Simple main effects revealed that both ownership and disownership items were rated more positively under the synchronous condition than under the asynchronous condition, and ownership items were rated more positively than disownership items under both synchronous and asynchronous conditions. None of the post hoc Tukey tests were significant.

To test whether there was any relationship between the experiences of rubber hand ownership and biological hand disownership, we calculated the correlation between scores of ownership items and disownership items, using the difference scores between the synchronous and the asynchronous conditions. We discovered a significant correlation between ownership and disownership (r = 0.38, p < 0.05). That is to say, the more strongly a participant experienced a feeling of ownership for the rubber hand, the more strongly they experienced a feeling of disownership for the real hand.

Correlation between onset time and questionnaire

To further probe the relationship between strength of the illusion as reflected by questionnaire ratings and the emergence of the illusory experience, we calculated two correlations. Both the correlations between OT and ownership items (r = −0.39, p < 0.05) and between OT and disownership items (r = −0.35, p < 0.05), when the asynchronous condition was subtracted from the synchronous condition, were significant. These two correlations revealed that the more quickly participants experienced the illusion, the stronger were the ratings of ownership for the rubber hand and disownership for the biological hand items, and vice versa. In short, the experience of RHI is reflected not only by questionnaire item ratings, but also by OT.

Overview and hypothesis

In Experiment 1, we asked participants to indicate the moment at which they began to experience an illusion. Questionnaire responses reflected that participants were able to distinguish between the experience of ownership and that of disownership, and that these responses correlated with their online recording of OT. Accordingly, we surmised that the OT measure can also be employed to distinguish between these two experiences when participants are aware of what type of illusion to expect. To verify this, for Experiment 2 we asked participants to indicate the instant when they began to experience ownership of the rubber hand, as well as the instant they began to experience disownership of the biological hand. We predicted that, if the disownership view holds true: (1) OTs for both ownership and disownership would be earlier under the synchronous than the asynchronous condition; (2) ownership and disownership OTs would correlate; and (3) ownership–disownership ratings would correlate with ownership–disownership OTs.

Methods

Onset time for the illusion

We used mean ± 2.5 SD as cut-off criteria and one of the participants was excluded in follow-up analyses. We conducted a two-way repeated measure ANOVA concerning the synchronous–asynchronous difference and belonging type (Fig. 4a). Both the main effects of the synchrony–asynchrony difference (F(1, 18) = 339.638, p < 0.001, η_p ² = 0.9497) and belonging type (F(1, 18) = 32.197, p < 0.001, η_p ² = 0.6414) were significant. This suggests that participants experienced the feeling of ownership and disownership more quickly under the synchronous stroking condition than the asynchronous stroking condition, and experienced the feeling of ownership faster than disownership. The interaction between the synchrony–asynchrony difference and belonging type was also significant (F(1, 18) = 16.027, p < 0.001, η_p ² = 0.4710). The simple main effect reveals that both ownership and disownership were experienced more quickly under the synchronous than the asynchronous condition, and that ownership was experienced more quickly than disownership under the synchronous condition, but that there was no difference between ownership and disownership obtained under the asynchronous condition. These data imply that participants experienced ownership for the rubber hand prior to disownership for the biological hand, but that if they did not experience ownership, neither did they experience disownership.

Furthermore, we assessed the relationship between the OTs of ownership and disownership, when the asynchronous condition was subtracted from the synchronous condition. Here, too, we discovered a significant correlation (r = 0.56, p < 0.001). This discovery implies that the more quickly participants experience ownership for the rubber hand, the more quickly they experience disownership for the biological hand.

Questionnaire

We conducted a two-way repeated measure ANOVA concerning the synchronous–asynchronous difference and belonging type (Fig. 4b). Both the main effects of the synchrony–asynchrony difference (F(1, 18) = 24.641, p < 0.001, η_p ² = 0.5779) and belonging type (F(1, 18) = 16.879, p < 0.001, η_p ² = 0.4839) were significant. This is to say, participants were more likely to affirm questionnaire items under the synchronous condition than under the asynchronous condition, and ownership items were more strongly affirmed than disownership items. There was also a significant interaction between the synchrony–asynchrony difference and belonging type (F(1, 18) = 14.189, p < 0.001, η_p ² = 0.4408). In addition, simple main effects revealed that both ownership and disownership items were rated more positively under the synchronous condition than under the asynchronous condition, and ownership items were rated more positively than disownership items under the synchronous condition, but not under the asynchronous condition. None of the post hoc Tukey tests were significant.

In addition, we calculated the correlation between scores of ownership items and disownership items when the asynchronous condition was subtracted from the synchronous condition. We again discovered a significant correlation between ownership and disownership (r = 0.69, p < 0.001). In short, the stronger the experience of ownership for the rubber hand, the stronger the experience of disownership for the biological hand, and vice versa.

Correlation between onset time and questionnaire

To further probe the relationship between questionnaire ratings and OTs, we calculated the correlation between ownership questionnaire scores and ownership OTs, as well as the correlation between disownership questionnaire scores and disownership OTs. Both correlations were significant (ownership: r = −0.44, p < 0.05; disownership: r = −0.44, p < 0.05). That is, the earlier the onset of ownership or disownership, the more robustly were those illusions experienced.

Overview and hypothesis

In Experiments 1 and 2, we asked participants to rate ownership items and disownership items adopted from Longo et al. (2008) and to indicate OT. For Experiment 3, three principal changes were made. First, because Longo et al. did not include control items, and in order to determine whether the manner in which ownership–disownership queries were expressed could alter results, we added a questionnaire used by Preston (2013; cf., Guterstam et al., 2011) that includes controls and alternative formulations of questionnaire items. We conjectured that if the ownership–disownership experiences are sufficiently distinct, semantic differences between the two questionnaires would not significantly affect the results. Second, we amended the numerical scale that is typically matched to questionnaires of this sort so that it more clearly indicated whether participants intended to affirm experiencing disownership, as opposed to merely indicating uncertainty. Third, to further explore the utility of the temporal dimension as an online proxy for ownership–disownership experiences, as well as to avoid the need to attribute a 300 s OT to participants who did not experience the RHI, we investigated ID to determine whether it too might evince the same pattern of correlations as those evinced by OT.

Methods

Illusion duration

We used mean ±2.5 SD as the cut-off criteria and no participants had to be excluded from follow-up analyses. We conducted a one-way repeated measure ANOVA concerning the synchronous–asynchronous difference and ID (Fig. 5a). The main effects of the synchrony–asynchrony difference were significant (F(1, 11) = 20.409, p < 0.001, η_p ² = 0.6498). This shows that participants experienced the feeling of ownership longer under the synchronous stroking condition than under the asynchronous stroking condition.

Questionnaire

First, we conducted a two-way repeated measure ANOVA concerning the synchronous–asynchronous difference and belonging type based on data collected from Longo et al.’s (2008) questionnaire. The main effects of the synchrony–asynchrony difference (F(1, 11) = 9.855, p < 0.01, η_p ² = 0.4725) were significant. This is to say, items were rated more positively under the synchronous condition than under the asynchronous condition. There was also a significant interaction between the synchrony–asynchrony difference and belonging type (F(1, 11) = 5.210, p < 0.05, η_p ² = 0.3214). In addition, simple main effects revealed that both ownership and disownership items were rated more positively under the synchronous than under the asynchronous condition, and ownership items were rated more positively than disownership items under the synchronous condition, but not the asynchronous condition. None of the post hoc Tukey tests were significant.

Second, we conducted a two-way repeated measure ANOVA concerning the synchronous–asynchronous difference and belonging type in Preston’s (2013) questionnaire (Fig. 5b). The main effects of the synchrony–asynchrony difference (F(1, 11) = 8.985, p < 0.05, η_p ² = 0.4596) were significant. This is to say, items were rated more positively under the synchronous condition than under the asynchronous condition. There was also a significant interaction between synchronous–asynchronous difference and belonging type (F(1, 11) = 8.101, p < 0.05, η_p ² = 0.4241). In addition, simple main effects revealed that both ownership and disownership items were rated more positively under the synchronous condition than under the asynchronous condition, and ownership items were rated more positively than disownership items under the synchronous condition, but not under the asynchronous condition. As for control items, there was no significant difference between synchronous and asynchronous conditions, indicating that the difference between synchronous and asynchronous in ownership and disownership items was not due to response bias. Moreover, the post hoc Tukey indicated that the ownership items were rated more positively than control items in the synchronous condition (p < 0.05) and the disownership items were rated more positively than control items in the synchronous condition (p < 0.05). There were also quadratic trends for both the synchronous condition (F(1, 44) = 5.940, p < 0.05) and asynchronous condition (F(1, 44) = 8.093, p < 0.01).

Third, we calculated the correlation between scores of ownership items and disownership items when the asynchronous condition was subtracted from the synchronous condition for data collected from both questionnaires. We discovered significant correlations between ownership and disownership in Longo et al.’s (2008) questionnaire (r = 0.59, p < 0.05) and in Preston’s (2013) questionnaire (r = 0.63, p < 0.05). In short, the stronger the experience of ownership for the rubber hand, the stronger the experience of disownership for the biological hand, and vice versa. Moreover, we calculated correlations between the two questionnaires: correlations between ownership items (r = 0.57, p < 0.05) and disownership items (r = 0.76, p < 0.01) for the two questionnaires were significant.

Correlation between illusion duration and questionnaire

We further probed the relationship between questionnaire ratings and ID. There were significant correlations between the ownership–disownership questionnaire items of Longo et al. (2008) and ID (ownership: r = 0.65, p < 0.05; disownership: r = 0.66, p < 0.01), as well as between the ownership–disownership items of Preston (2013) and ID (ownership: r = 0.55, p < 0.05; disownership: r = 0.65, p < 0.05). In other words, the longer the ID of ownership, the more robustly felt was the illusion.