Body ownership causes illusory self-attribution of speaking and influences subsequent real speaking

Abstract

When we carry out an act, we typically attribute the action to ourselves, the sense of agency. Explanations for agency include conscious prior intention to act, followed by observation of the sensory consequences; brain activity that involves the feed-forward prediction of the consequences combined with rapid inverse motor prediction to fine-tune the action in real time; priming where there is, e.g., a prior command to perform the act; a cause (the intention to act) preceding the effect (the results of the action); and common-sense rules of attribution of physical causality satisfied. We describe an experiment where participants falsely attributed an act to themselves under conditions that apparently cannot be explained by these theories. A life-sized virtual body (VB) seen from the first-person perspective in 3D stereo, as if substituting the real body, was used to induce the illusion of ownership over the VB. Half of the 44 experimental participants experienced VB movements that were synchronous with their own movements (sync), and the other half asynchronous (async). The VB, seen in a mirror, spoke with corresponding lip movements, and for half of the participants this was accompanied by synchronous vibrotactile stimulation on the thyroid cartilage (Von) but this was not so for the other half. Participants experiencing sync misattributed the speaking to themselves and also shifted the fundamental frequency of their later utterances toward the stimulus voice. Von also contributed to these results. We show that these findings can be explained by current theories of agency, provided that the critical role of ownership over the VB is taken into account.

Keywords: agency; body-ownership illusion; illusory speaking; rubber-hand illusion; vibrotactile stimulation.

Fig. 1.

The experimental setup. The body of the participant was substituted by a sex-matched VB, viewed from 1PP, onto which body and head movements were mapped in real time for the sync condition. The body could also be seen as reflected in a virtual mirror. (A) Participants wore an HMD with earphones, a full-body motion-capture suit, and a vibrotactile device. (B) The female VB. (C) The male VB. B and C illustrate that the VB (here, the arms) could be seen directly when looking toward their own body, and also in the virtual mirror.

Fig. 2.

Boxplots for scores on body ownership. The horizontal black bars are the medians, and the boxes the interquartile ranges (IQRs). The whiskers stretch to the data points that are within the median ± 1.5 IQR, with outliers beyond this shown as single points. MyBody and Mirror are significantly different between async and sync (P < 0.0005, ordered logistic regression) (Table S2).

Fig. 3.

Boxplots for scores on agency and speaking. Agency is significantly different between the visuomotor conditions (P < 0.0005) and the vibrations conditions (P < 0.024). The differences between the visuomotor conditions are significant for VoiceSourceRoom (P = 0.033), VoiceSourceHead (P = 0.004), OwnVoice (P = 0.011), and Speaking (P < 0.0005). There are no other significant differences. All significance levels are with respect to ordered logistic regression (Table S2).

Fig. 4.

Bar chart for dF by visuomotor and vibrations.