Fig 1.
Rubber Hand Illusion as Causal Inference.
Spatial signals (χ) and temporal signals (τ) coming from the visual (χv, τv) and somatosensory modalities (proprioception: χp, tactile: τt) are either integrated or segregated depending on whether the brain infers a common cause or independent causes for the sensations.
Fig 2.
a) Synchronous Stroking: Distributions are the likelihoods representing the objective stimulus locations/timings. Marked points are the model estimates (MAP) of stimulus location/timing. b) Synchronous Stroking: The frequency of simulation runs in which a common cause is inferred is shown in the shaded bar. c) Asynchronous Stroking: Distributions are the likelihoods representing the objective stimulus locations/timings. Marked points are the model estimates (MAP) of stimulus location/timing. d) Asynchronous Stroking: The frequency of simulation runs in which a common cause is inferred is shown in the shaded bar.
Fig 3.
Simulation Results: Spatial Extent.
The probability of experiencing the illusion is plotted as a function of the distance (in centimeters) between the rubber hand and the real hand. As the distance between the two increases, the illusion becomes weaker and eventually fails to occur. These results are qualitatively and quantitatively consistent with empirical findings from human participants [25].
Fig 4.
Simulation Results: No Stroking.
a) Removing the temporal dimension from the model retains the illusory effect of overlapping spatial estimates. Marked points represent model estimate (MAP) of hand location. b) The frequency of simulation runs in which a common cause is inferred is shown in the shaded bar.
Fig 5.
a) RHI Apparatus b) Experiment 1 procedural design.
Fig 6.
Ownership Ratings Prior to Tactile Stimulation.
a) Median pre-test ratings from groups ‘sync’, ‘async’, and ‘no-stroke’ indicated by black square. Bars indicate interquartile range. b) Histogram of ownership ratings. The ratings are on a scale of -3 to 3, whereby -3 and +3 correspond to strong disagreement and strong agreement, respectively, with the statement I feel as though the rubber hand is my hand. **** p < 0.0001
Fig 7.
a) Proprioceptive Drift: The change in proprioceptive localization from pre-test to post-test. n = 21. * p < 0.05, ** p < 0.01, *** p < 0.001. b) Ownership: The median change in subjective ownership report from pre-test to post-test indicated by black squares. Bars display interquartile range. ** p < 0.01, *** p < 0.001
Fig 8.
Ownership and Proprioceptive Drift.
Scatterplot of ownership ratings plotted against proprioceptive drift in the (a) pre-test and (b) post-test from the three groups which were presented with a rubber hand. Large outlined circles represent means for those who gave the same ownership response.
Fig 9.
Experiment 2 procedural design.
Fig 10.
a) Median ownership ratings after the end of the experiment indicated by black squares. Bars display interquartile range. b) Elicited SCR at two time points, “eye-opening” and “threat”.
Fig 11.
Scatterplot of ownership ratings plotted against the logarithm of the SCR to Eye-Opening (a) and to Threat (b) from the two groups which were presented with a rubber hand. Large outlined shapes represent means for those who gave the same ownership response.