Dear Maria,

what you are observing is very common in the RHI. I have observed this in my own research (Rohde et al., PLoS ONE, 2011 and 2013) on the illusion and have also seen it in a lot of other papers - just check out some studies that look at correlations and therefore give scatter plots - it is very common. Anecdotally, I share your observation that this might be more likely to occur when asynchronous was measured first.

For the reasons for this - I think there are two main factors:

1.) Questionnaires give noisy and subjective measures.

In a questionnaire, you will always have participants that change their criterion between sessions, that are influenced by what they think you might want to hear, or that give random responses - you cannot expect such a rating to be reliable indicators for experience of ownership in each participant - you can only work with the population average, where a strong difference between conditions should occur (otherwise you are really doing something wrong). The presence of noise on this measure does not mean something is wrong with the participants' brains.

2.) Body Ownership works in mysterious ways.

I do not think that reports of ownership in the asynchronous condition are always due to noise - sometimes participants really have a strong illusion in the asynchronous stroking condition. It is a known fact taht there are a lot of inter-subjective differences in the illusion, for instance, some people have it immediately and very strongly, for others it takes a while to build up and is not as freaky, and yet others do not have it at all. There is also the finding by Durgin et al. (Psych Sci, 2007) that participants report feeling touch on a visually stimulated rubber hand without actually receiving any tactile inputs. And, anecdotally, there are some few people who feel ownership for a visually presented or asynchronously stimulated hand. We simply do not understand the mechanisms of body ownership enough to explain this variability, but it exists.

So in my opinion, you are doing nothing wrong (unless the noise is stronger than the signal), you measure a mix of noise and individuals with unusual susceptibility. But you really must not exclude any of these measurements, because this would bias your sample. You would keep the favourable outliers, but not the unfavourable ones and this would distort your measurements.

Hi Maria, I hope some papers will help you (below). All the best. Vladimir

Dear Maria,

what you are observing is very common in the RHI. I have observed this in my own research (Rohde et al., PLoS ONE, 2011 and 2013) on the illusion and have also seen it in a lot of other papers - just check out some studies that look at correlations and therefore give scatter plots - it is very common. Anecdotally, I share your observation that this might be more likely to occur when asynchronous was measured first.

For the reasons for this - I think there are two main factors:

1.) Questionnaires give noisy and subjective measures.

In a questionnaire, you will always have participants that change their criterion between sessions, that are influenced by what they think you might want to hear, or that give random responses - you cannot expect such a rating to be reliable indicators for experience of ownership in each participant - you can only work with the population average, where a strong difference between conditions should occur (otherwise you are really doing something wrong). The presence of noise on this measure does not mean something is wrong with the participants' brains.

2.) Body Ownership works in mysterious ways.

I do not think that reports of ownership in the asynchronous condition are always due to noise - sometimes participants really have a strong illusion in the asynchronous stroking condition. It is a known fact taht there are a lot of inter-subjective differences in the illusion, for instance, some people have it immediately and very strongly, for others it takes a while to build up and is not as freaky, and yet others do not have it at all. There is also the finding by Durgin et al. (Psych Sci, 2007) that participants report feeling touch on a visually stimulated rubber hand without actually receiving any tactile inputs. And, anecdotally, there are some few people who feel ownership for a visually presented or asynchronously stimulated hand. We simply do not understand the mechanisms of body ownership enough to explain this variability, but it exists.

So in my opinion, you are doing nothing wrong (unless the noise is stronger than the signal), you measure a mix of noise and individuals with unusual susceptibility. But you really must not exclude any of these measurements, because this would bias your sample. You would keep the favourable outliers, but not the unfavourable ones and this would distort your measurements.

If real live results from participants give rise to scores that don't make sense then I think you have to assume that the maths of your formula is wrongly designed. (That would be no surprise.) I agree with Marieke that certainly you should not ignore results. It sounds as if you should ask Tony Marcel if you do not have a supervisor who can answer your question. (I would be a bit worried if you do not in fact.) The other person who might help locally is Ophelia Deroy at the Institute of Philosophy at Senate House.

My broader question would be to ask what hypothesis you are wanting to test. You say 'taken the embodiment pattern of the rubber hand illusion as a given'. I am not sure what that means but why should you take anything as a given here? Embodiment seems to me to too vague a concept to be of any great scientific use. Perhaps you should forget about embodiment and have a clear scientific hypothesis - I might understand it! Sense of body ownership sounds a bit more meaningful to me.

But seriously, Tony knows more about this than anyone I can think of and he comes in to Malet Street fairly regularly.