From the authors (original comment enclosed by ****):

*******************
1) The authors employ a binary factor approach for their analyses. This approach allows the assessment of the main factors, but cannot be used to examine the interaction-terms between those factors.
*******************

See p8, Section ‘Statistical Methods’. The analysis was presented using both the proportional odds model and classical analysis of variance with all two-way interactions (and even 3-way interaction). The results are the same as for the proportional odds model, and the interaction terms are nowhere near significant.

Of course it is true that maybe there are interaction effects but the sample size would need to be much larger in order to stand a chance to find them. There is always the need for experiments with greater sample sizes, so our findings remain to be replicated (or not) with other experiments.

Now since the M (movement) factor was not directly significant, we could consider just perspective position (P) and T (touch) as factors, using a two-way ANOVA with 6 observations per cell. This would increase the power, and make it more likely to be able to find an interaction effect between perspective position and touch. Again we find no interaction effect.

To emphasise: this is not to say that there is no interaction effect ‘in reality’, it is just that if there is one, it is too weak to be discovered with the particular sample size that we had.


*******************
Thus, strictly speaking, the data in this study does not enable the authors to draw any conclusions about the relative importance of the different factors. For that purpose, a standard 2x2x2 factorial design should have been used.
*******************

This was a standard 2x2x2 factorial design. It is not clear why this is doubted. There were three factors, each had two levels, 3 subjects were arbitrarily assigned to each combination of levels. So it is a 2x2x2 design with 3 replications per cell.


*******************
Furthermore, the fact that there were only 24 participants across all possible combinations of the three binary conditions (i.e. Persepctive - 1PP vs. 3PP; Motion - synchronized vs. asynchronized; Touch - synchronized vs. asynchronized), led to the situation that only 3 participants were experiencing the exact same combination of conditions. In that sense Figure 3D and Table 4 should be interpreted with great caution.

*******************

The sample size is relatively small. But if in spite of the small sample size we still found significant main effects, even though the power is low, this means that in fact the result is quite strong. It is important to note that the assumption of normality of the residual errors of the models were tested, as reported in the paper.

*******************
2) The questionnaire results do not seem to support the authors’ claims that the perspective "clearly dominates as an explanatory factor for subjective and physiological measures of ownership" (p.4, 4th sentence in the Discussion), and that "when perspective position is included as factor in the experimental design the importance of visual-tactile synchronization diminishes in comparison to what would be expected from the literature" (p.2, last sentence of the Introduction). The results of the questionnaire (Figure 3A-C) show that the scores for both binary conditions of the factors Perspective and Touch to 5 out of 8 questions (body, woman, mirror, cnct and attack) look identical.
*******************

The ANOVA (and proportional odds models) includes both P (perspective condition) and T (touch). P is significant and T is not. The comment above is therefore not supportable.


*******************
Surprisingly, the scores to the touch question in the asynchronous touch condition are somewhat high, which might however be due to the fact that the participants were not explicitly told to observe the touches.
*******************

It is important to distinguish between the ‘touch’ question(s) and the Touch condition. The latter refers to whether there was synchronous visual-tactile stimulation (i.e., the woman’s visual strokes were synchronous with the felt strokes).

Regarding the touch questions, these were

5. How strong was the feeling that the woman you saw was directly touching you on the shoulder?
10. How strong was the feeling that the touch you felt was caused by the woman that you saw?

It is quite possible that answers to 5 and 10 could be high even in the asynchronous touch condition, because the dominant factor here is 1PP. I.e., in the 1PP condition people would experience the touch as being caused by the woman simply because this is what they saw, and what they saw dominated what they felt. This is very clear from Figure 3A.


*******************
The only real differences between the Perspective and Touch factors are the scores to the cloth and the hurt questions. The cloth question is, however, less important to the issue of body ownership and is probably biased by the visual input. In that sense, there is no clear evidence in the data to support the authors’ claim that the visual first person perspective is sufficient or more important than the visuotactile integration as a factor producing body ownership illusions.
*******************

We disagree strongly with this statement. The evidence points to the conclusion that in this experiment 1PP dominated everything else. The T factor simply was not as important in any analysis that we considered.

It is not appropriate to rely simply on the charts shown in the Figures. Inevitably these consider each variable individually, and they do not take into account their joint influence on the responses. This can only be done through the statistical models.


*******************
3) If one looks at the data it seems that the elevated perspective abolished the illusory perception of owning the virtual body. From the paper it was not clear to me why the authors chose to include this condition. In this condition participants were experiencing an elevated perspective (from the ceiling) while no tactile stimulation was delivered to their body; only the virtual body, which was observed from the third person perspective, was visually stimulated. When I read the paper it seemed to me as if the authors were using this condition to support the claim that the visual perspective is sufficient to establish the illusory perception of body ownership. However, the scores to the questions cnct ("When you were looking down from above, how much did you feel a strong connection with the seated girl as if you were looking down at yourself?") and attack ("When the standing woman hit the seated woman, how much did you feel this as if this was an attack on your body?") in the questionnaire (Figure 3A-C) suggest that this situation led to a weak or actually non existing illusion of owning the virtual body. The ratings of those two questions corresponded to a "Low" score and did not differ at all between the different binary conditions of the three main factors. Thus, in my opinion the data cannot be used to support the claim that the visual perspective alone is much more important than the visuotactile synchrony.

*******************

There is no evidence to suggest that the elevated perspective in itself had any particular effect one way or the other, and it was not our purpose to maintain or eliminate the body ownership illusion during this period, and this was never claimed in the paper.

Its main purpose was a practical one. We wanted to have the ‘slap’ in order to measure the physiological response to this. We could not have the slap from a 1PP perspective because we were not going to deliver an actual slap to the subject. Therefore we elevated the viewpoint. Then our question was how they would respond physiologically to the slap given their particular experimental condition (P,M,T).

As stated in the paper the spatial coordinates of this viewpoint transformation were chosen based on such transformations in neurological patients with out-of-body experiences.

It is true that there were no significant effects for the factors ‘connect’ and ‘attack’. But there was an effect for the factor ‘hurt’, meaning that after the participants returned back down inside the girl’s body, in the 1PP condition they were far more likely to experience a fear that they might be hurt by the standing woman. Thus the statement that the elevated condition eliminated the effect is not relevant (and was not the question we asked) but what is relevant is that while in the elevated condition they had a different physiological response had they been in the 1PP compared to the 3PP condition.

The conclusion of the commentary is not valid. Once again the ONLY significant factor was P (1PP resulting in a greater effect than 3PP).



*******************
4) The physiological measurement (HRD) needs further validation. It is very commendable that the authors introduce the Heart Rate Deceleration (HRD) as a novel objective measure for the illusory ownership of a body. However, I need to point out that the HRD results are not supported by the questionnaire results. From Figure 4 it is clear that the HRD during the slap procedure was higher for the 1PP than for the 3PP. However, the questionnaire scores to the attack and cntc question (as mentioned earlier) do not differentiate between the two conditions of the Perspective factor. Thus, it is unclear why there is a difference in the heart rate during the slap between the two perspectives, when there is no concomitant difference in the subjective experience of the attack as being directed towards the own body between the two perspectives.

*******************

This comment is unsupportable. The last paragraph before the discussion is as follows:

“There is, furthermore, consistency between physiological responses (HRD) and the subjective questionnaire responses. During the slap (duringS) and after the period of being in the elevated position (down) the HRD was significantly positively correlated with a feeling of the participant’s own body being attacked (attack), the feeling that they might be hurt by the woman (hurt) and body ownership (body). However, there were no significant correlations between any of the questionnaire responses and HRD for the control periods (beforeS and across). The full set of correlations and significance levels is in Table 4.”




*******************
Further, in the discussion of the paper the authors mention that they performed Skin Conductance Response (SCR) measurements but that these did not yield any significant difference "between seeing the slap in the 1PP and 3PP conditions or TS and TS' conditions" (p.6, 4th paragraph). They also state that the SCR "measures arousal, the valence of which is not known" and that in this experiment it is "simply an arousing effect to see someone slapped" (same paragraph). It is highly regrettable that neither in the methods section nor the results section there is any explanation of the procedure and presentation of the actual SCR data. Thus, it is not meaningful to compare the SCR results in this study with those of Petkova & Ehrsson (2008). As described above, the subjective scores to the cnct and attack questions make it clear that the participants did not experience the illusion of owning the body of the girl while they were observing it from the elevated perspective. Since the slaps were observed from this perspective, it is not surprising that the SCR did not show any differentiation between the binary conditions of the Perspective and Touch factors. If anything, the negative result in the questionnaire data is consistent with the negative SCR result, thus validating the SCR measure further. I would also like to add that the SCR is a well-established measure of autonomic nervous system reactions to emotional stimuli. In Petkova & Ehrsson’s (2008) study there were several control experiments that ensured that the reported SCR was specific to a knife threat in the illusion condition only. Unfortunately, such extensive validation of the HRD is still missing and would merit further investigation.

*******************

We found no significant results using skin conductance at all. We used a standard method to obtain skin conductance, using a Nexus 4 sampling at 128Hz, with electrodes on the fingers of the non-dominant hand. We looked at both the number of skin conductance responses and also the maximum rise in skin conductance level 5s after events. (We also tried various other durations such as 3s, 6s, etc.).

Our results are not comparable with those of Petkova and Ehrsson (2008), since in that paper there was no variation in perspective position - i.e., it was always 1PP and seen from a fixed viewpoint direction. However, as discussed in our paper (p6) a special case of our experiment does support the results found in PE2008.


*******************
I hope my comments on your paper will foster fruitful scientific discussions!
*******************

To summarise, in reality our experiment was quite simple. We had 3 factors and two main types of response variable (questionnaire and physiological). Whatever type of analysis we do, whether full ANOVA (not really appropriate for ordinal data but nevertheless used very often) or the proportional odds model (more appropriate for ordinal data), we always find the same thing. P is the significant factor and 1PP shows greater evidence of the illusion than 3PP, and the synchronous touch condition shows some effect but not as strong. This is the case for the most important questions relating to body ownership, and this is just what the results show.

A scientific discussion is to be encouraged, but the commentary has an erroneous understanding of the experimental design and method of analysis, and its conclusions are not supported by the data itself.

The result of our study suggests that 1PP in a wide field of view head-mounted display with low latency head-tracking, when you can look down and see another body replacing your own, induces an illusion of body ownership, and is stronger than synchronised touch. (This result cannot be compared - or only with caution - with the commentary writer’s own work, as mentioned above, since that experiment did not have manipulation of perspective position as a factor).

We do not think that our results are surprising at all. Suppose you were able to look directly at your own real body, but some experimental setup was established so that you saw something repeatedly touch your body at a different time to when you felt the touch. Would you cease to have the feeling that it was your own body? (Perhaps you might, this is an empirical question).

In reality the best way to further this discussion is not more argument about the data, but through the carrying out of more experiments.