Fake Hands and Tables-The Malleability of the Body Image

In " The Classical Rubber Hand Illusion", we discussed the original experiments of Botvinick and Cohen (1998). Their hypothesis for the rubber hand illusion was that vision has higher reliability and spatial acquity than proprioception, so the brain gives more weight to visual information. People would thus localize a body part to it's apparent visual location, particularly when the visible location falls within the possible range dictated by proprioception. Some support for this theory lies in the fact that placing the fake hand perpendicular to the real occluded hand destroys the illusion that the fake hand is one's own.

Armel and Ramachandran  (2003) reported a closely related but bizarre illusion. The subject is made to place his or her real hand on a table and the hand is hidden from view. However, instead of stroking a fake rubber hand, the researchers simply stroked and tapped the table in precise synchrony for a minute. Astonishingly, subjects reported sensations arising from the table surface, despite the fact that it bears no physical resemblance to a hand. Whereas Botvinick and Cohen interpret their result in terms of resolving incongruities between visual versus proprioceptive location of the hand, Armel and Ramachandran's experiment would lead one to argue that the illusion arises mainly  from Bayesian logic of all perception; the brain's remarkable ability to detect statistical correlations in sensory inputs in constructing useful perceptual representations of the world, including one's own body. It is especially intriguing that the bizarre perceptual representation (assimilating the table into the body image) is resistant to the "top-down" knowledge of the absurdity of the situation!

To measure the extent to which subjects incorporated the external objects into their body image, they were asked to rate the vividness of the illusion. The experimenters also recorded the skin conductance response (SCR), to provide an objective test of whether the table had indeed become informationally coupled with the subject's body image. If the external objects became integrated into the body's image, would they be aroused when the table (or a fake hand for that matter) was 'injured'?

If a finger of the fake hand is bent backwards to seem painful, does the subject register an SCR? To what extent is the fake hand assimilated into the subject's body image? To address this, after ca. 2.5 minutes of the touching procedure, both the real and the fake fingers were lifted, but only the fake finger was bent backwards into a 'painful position'. SCR was recorded at this point and a free response description and intensity rating of the illusion were obtained. The control for this experiment was a 'delayed synchrony' condition wherein touch to the real and fake hands were identical, the only difference being, that the touch to the real hand was delivered 1 sec after the touch on the fake hand. Mean intensity and mean SCRs showed that subjects identified with the fake hand more in the condition where the touch was synchronized rather than the one where the touch was synchronized but delayed (see Fig. 1)

Fig. 1: Mean Intensity ratings (gray bars) and SCR (black circles) in the first experiment where synchronous touch to the real hand was delayed by a second in the control condition. The error bars indicate one SEM (Adapted from Armel and Ramachandran, 2003)

Would subjects still experience the illusion if the form of the external object was manipulated? To explore this, a barren table was stroked  and tapped in the same manner and in the same relative location (see Fig.2 ). 

Fig. 2: Form manipulation where subjects received the table condition (Adapted from Armel and Ramachandran, 2003)

Band-aids were placed on both the real hand and the table and subjects were told that the band-aid would be pulled off the table but not off their real hands. At the end of the 2.5 min touching period, in lieu of pulling back a fake finger, the band-aid on the table was partially pulled off. In the control condition for this experiment, the real hand was made visible by removing the occluder. Subjects were instructed to look back and forth between their real hands and the table. Furthermore, the real hand and the location where the table was touched were close together so that they could be seen simultaneously even while looking at one or the other. The band-aid was pulled off the table while the subject viewed it. To ensure that the subject was only looking at the table, the experimenter occluded the subject's real finger at this time. In a comparison of the conditions in which the table and real hand were touched with the partition in place or removed, intensity ratings and SCR were significantly different (see Fig. 3). However, the same experiment (using a band-aid) carried out with a fake rubber hand in lieu of the table is more effective at inducing the illusion in terms of intensity ratings but only marginally so for SCR (see Fig. 3)

Fig. 3: Mean Intensity ratings (gray bars) and SCR (black circles) in the second experiment where form of the external object was manipulated. The error bars indicate one SEM (Adapted from Armel and Ramachandran, 2003)

Would subjects still experience the illusion if the form of the external object was manipulated? Each subject viewed touch to a fake hand in a 'realistic' location in one condition and then to a distant fake hand in another (see Fig. 4).

Fig. 4: Location manipulation where subjects received the distant-hand condition (Adapted from Armel and Ramachandran, 2003)

The fake arm was extended so that it lay 3 feet beyond the real hand. In the distant fake hand manipulation, a fake finger was bent back for a painful stimulus. In the control condition, the touch applied to the fake and real hands was asynchronous i.e.  touch was random and there was no correlation (see Table 1 at the end for summary of conditions for all experiments). Mean intensity and mean SCR showed that subjects identified with the fake hand and the distant fake hand, more in the conditions where touch was synchronized than when not synchronized. However, the 'anatomically correct' fake hand condition was more effective than the distant fake hand condition (both with synchronous touch, see Fig. 5)  

Fig. 5: Mean Intensity ratings (gray bars) and SCR (black circles) in the second experiment where the location of the fake hand was manipulated. The error bars indicate one SEM (Adapted from Armel and Ramachandran, 2003)

The so-called body image appears to be highly malleable. Despite  it's appearance of durability, it can be profoundly altered by stimulus contingencies and correlations that one encounters. Taken together, these experiments illustrate an important principle underlying perception: that the mechanisms of perception may be involved in extracting statistical correlations. Further investigations will further our understanding of phenomena such as body-dysmorphic disorder and anorexia nervosa

Table 1: Experimental design for all experiments. For a given experiment, each subject received all conditions in one of different possible orders (Adapted from Armel and Ramachandran, 2003)

Reference:

Armel, K., & Ramachandran, V. (2003). Projecting sensations to external objects: evidence from skin conductance response Proceedings of the Royal Society B: Biological Sciences, 270 (1523), 1499-1506 DOI: 10.1098/rspb.2003.2364