Phantom limbs can contort into impossible configurations

FOLLOWING the surgical removal of a body part, amputees often report sensations which seem to originate from the missing limb. This is thought to occur because the brain's model of the body (referred to as the body image) still contains a representation of the limb, and this leads to the experience that the missing limb is still attached to their body. Occasionally, amputees say that they cannot move their phantom limbs - they are perceived to be frozen in space, apparently because they cannot be seen.

Yet, research shows that the body image is malleable and easily manipulated. And according to a new paper published in the Proceedings of the National Academy of Sciences, phantom limbs can be altered by internal brain mechanisms alone. The study shows that some amputees can make their phantom limbs defy the anatomical constraints of the physical body, using visual imagery to make them perform movements which could not possibly be performed by a real limb.

Lorimer Moseley and Peter Brugger recruited seven arm amputees, all of whom had experienced vivid and enduring phantom limb sensations, and encouraged them to use visual imagery to contort their phantoms into configurations which would be normally be impossible because of the biomechanical constraints of the body. Initially, all the participants reported that it felt as if they were observing somebody else's arm performing the movements, or that the phantom they visualized did not feel like their own.

But with training, four of them reported that they had successfully learned how to perform the impossible movement, by visualizing their phantom wrist spinning the wrong way. After learning, these participants also reported feeling a change in the structure of their phantom limbs, involving the development of a new phantom wrist joint which could rotate freely on the forearm, which permitted the limb to perform the impossible movement. One also reported that his phantom now included a new axis which protruded into the hand, and drew a picture to illustrate its structure (below left).

These subjective reports were corroborated by the participants' performance on two motor imagery tasks which provided an objective and quantifiable measure of implicit motor behaviour. In the first of these tasks, the participants were shown a series of images of human hands and asked to judge, as quickly as possible, whether each was of a left or a right hand. Included in the task were pictures of left and right hands in two different postures at either end of the impossible movement which had been practiced earlier, and some of the trials required the participants to hold their phantom hand in one of these postures.

Correct judgement on this task is known to require mental rotation, whereby one imagines rotating one's hand until it reaches the position depicted in the image. The response time is related to the extent of the mental manoeuvres required to imagine one's own hand rotating from its resting position to one which matches the hand in the picture. For healthy participants, reaction times on the task are shorter when the posture of the real hand matches that in the image. Earlier studies have also shown that amputees take longer than non-amputees to correctly identify an image corresponding to their missing limb. 

phantom limb impossible movement.JPG

At the outset of the experiment, the performance of all the participants was consistent with these earlier findings. Their reaction times were shorter when their hand posture matched that depicted in the presented image, and reaction times for pictures corresponding to the missing limb were longer than for those of the intact limb. After training, however, some differences were observed in the four participants who had reported successful learning of the impossible movement. In these participants, but not in the other three, reaction times for pictures of the missing limb decreased, and were the same as those for pictures of the unaffected limb. Their performance also improved on trials in which the position of their phantom limb was opposite to that of the hand shown in the image.

In the second task, the participants were shown two alternating images, depicting a hand at either end of the impossible movement, and asked to report the apparent path of motion required to move their phantom hand between one posture and the other. This involves either a long rotation of the wrist or a shorter, but anatomically impossible, rotation in the opposite direction (above, middle and right). Normally, non-amputees perceive the hand moving through the longer motion path when the interval between the two flashed images is equal to, or slower than, the time taken to actually move their hand in that way (0.7 - 1 seconds). The shorter path is only reported when the interval is outside this range.

Again, the participants initally reported perceiving the long rotational movement, but only when the alternating images were separated by intervals of greater than 0.7 seconds. After training, however, the four participants who claimed to have learnt the new movement successfully perceived the shorter motion path regardless of the time interval between the alternating hand images. In all four participants, this was found to be true only for images which corresponded to the amputated limb. In trials involving images of the unaffacted hand, manipulations of the time interval caused no change in performance. And there was no change the responses of the three participants who were unable to learn the new movement.  

These results provide good evidence that the four successful participants had modified the neural representation of their phantom limbs, to incorporate a new joint which enabled the phantoms to perform impossible movements of which they were previously incapable. Furthermore, they had also generated novel motor commands with which the movement can be executed. Remarkably, the newly configured phantom limbs were apparently still constrained by the laws of physics. Two of the successful participants reported that the new wrist joint had made it more difficult for them to move their phantom hands from side to side. It seems that the brain's representations of body parts are constrained by the relationship between structure and function, even when the representations are not based on a physical substrate.

Although the body image has a genetic basis, it is also thought to be continuously updated by sensory feedback from the body. Manipulating the information which flows into the brain through the senses can profoundly alter the way we perceive our bodies. For example, conflict between visual and tactile information can give rise to out-of-body experiences, or to the illusion that the sensations one feels are originating from someone else's body and not one's own. But as this study shows, the body image can also be altered by internally generated mechanisms, in the complete absence of feedback from the body. This suggests that visual imagery could be used to re-adjust the body image distortions that occur in a variety of neurological and psychiatric conditions.


Thanks @missippy!

Moseley, G. L. & Brugger, P. (2009). Interdependence of movement and anatomy persists when amputees learn a physiologically impossible movement of their phantom limb. Proc. Nat. Acad. Sci. DOI: 10.1073/pnas.0907151106.


More like this

Shades of Gil "The Arm" Hamilton!

Consider that in the future, it will likely become straightforward for people to control mechanical limbs with the same sorts of mental impulses we all use to control our normal limbs. This research suggests that at least some such people need not be constrained to human-like mechanical limbs. At minimum, they could very likely adapt their body image to accomodate limbs with non-physiological joints.

One wonders how far that could be taken. Could they adapt their body image to accomodate more than 2 arms? What about flexible appendages, a la tentacles? Is this a skill that only some will be able to learn?

Fascinating stuff!

I love it when you go out on a limb like this!

This is a lovely, fascinating article. However, though it is admittedly hard to find amputees that fit the criteria of the experiment, seven IS rather a small sample.

So, this experiment was done with video, am I correct, modified to include a physically impossible movement for a human wrist. That's, as qetzal says, quite fascinating. Perhaps non-human-like "waldoes" could be designed as prosthetics or even non-amputees might learn to use them with these techniques.

Did they test the subjects for differing mapping of the phantom on sections of the skin (as so often happens with phantom limbs in others)? Perhaps the mapping would be onto a joint that COULD make this movement, ie, the elbow (as I can see it). Do you think it would have made a difference with this training, and would there be MULTIPLE phantoms, one "normal" and one able to perform this "abnormal" movement?

My last question is, were the subjects sorted for age? Did age play a factor in the malleability of the body image that was seen in this experiment? Were the folks who learned the "impossible" movement older or younger? Again, it would be difficult to see a trend in such a sample size, but the information would be interesting.

@qetzal: It makes me wonder what other types of body image modifications might be possible through visual imagery. I'm also curious about why only 4 out of the 7 amputees managed to learn the impossible movement successfully. Here's an earlier post about a woman who experienced a phantom third arm following a stroke.

@Lorimer: Thanks for dropping by and commenting. It's good to know that you approve of my write-up! Nice blog, by the way; I'll be keeping a close eye on it, and will get in touch in the near future...

@E Wolke: Yes, a sample size of 7 is on the smallish side, but that doesn't make the results any less fascinating. Undoubtedly the same learning task will be tried on more amputees in the future. The study used images, not video; there's no mention of mapping the phantom onto the skin surface and if I remember rightly, all the participants had their arms amputated above the elbow. Regards the participants' age, I initially thought that the time elapsed between amputation and the study may be a more important factor, but looking back at the paper, this seems irrelevant.

I am not sure that study *shows* that body image can be altered. It implies it, we all want to believe it. But *show*? I believe, that that research makes subjective sensations a bit more plausible, and helps the test subjects to express themselves. When I participated in a neurophysiological study and described my phantom thing, it became apparent that they *understood* what I was saying but they were not able to emotionally weigh the sensations the way I would do it, or other amputees that I know would. So there is a really wide gap in communication there.

Of course the brain will learn to adapt a terminal device even if it is a tool and not a hand. It does not have to be hand shaped and the learning process does not feel painful if the device works. I started to do automatical and subconscious quick grabs with a body powered hook just after two weeks of wearing it - and so when I mindlessly picked something up that had fallen down (that was in a supermarket) a person nearby said they were surprised at the speed and fluent motion. And I had not even noticed.

I started mirror therapy - and my very strong but painless phantom sensations changed to mostly weak phantom sensations and occasional phantom pains. So, things are changing, obviously. The art in getting things going there definitely consists in starting from the current situation, and by doing lots and lots of frequent little steps. I actually managed to wiggle my phantom hand for a while but now it's back to its normal "freeze" state.

To maximize fast learning I believe the art of prosthetics consists in building a robust control and a robust mechanism - something that always works more or less the same way. Because there are things that drive me nuts, too. If the prosthesis hurts, or if parts wiggle to a point that I feel to lose control, I really get antsy. And more intuitive just is better. Now I wear a Becker hand that I painted red, my self constructed wrist unit (the guys that helped me work in high tech, so don't expect Lego parts) now looks rather harmless but it functions like a drillhammer's fast release/lock (since it's built like one), and for what it is, I love it.

Thanks for this post on a very curious aspect of proprioception. For the benefit of your readers, I'll repeat the tip I sent your way on Twitter a few days ago: Ian Robertson at Trinity College Dublin has done some interesting work on brain rehabilitation. I encountered him through his popular neuroscience book Mind Sculpture, which includes a chapter on phantom limbs and related phenomena.

Hi, I am an above knee amputee, 68 operations, 23 broken bones. Very long road. AKA for 13 years. Having one leg is the easy part. The crushing, burning, gnawing feels like a vice with razors is enclosing on my phantom foot which feels half way between where it should be and my short stump. It's difficult to use a prosthesis because of the excruciating pain. I have done over thousand hours research. Very difficult to treat. Would love to go see Vilayanur S. Ramachandran MD, PhD.
He is the expert. Anyone here have anything they can suggest like a great doctor or one who is studying phantom pain please email me at

This may sounds strange or off-topic, but I've always been able to -feel- limbs that have never even existed. I can "imagine" the sensation of having a tail. I can send the impulses that I would use to move it, only I don't and never have had an actual tail. I can repeat this with other typical vertebrate body parts such as having large floppy ears, or a penis (I am a woman). I am also a chronic pain patient and have done a lot of research into somatosensory dysfunction and phenomena. I believe the fairly unique modulation of my nervous system allows my perception to reach beyond requiring actual nerves and tissue to perceive their existence in space, and to know what it would feel like for someone to step on my tail. Perhaps this has something to do with my persistent chronic pain as well. I don't seem to need an actual stimulus to have a noxious sensation such as pain.

The whole tail thing is neat, and I've never really shared it, but the chronic seemingly uncontrollable pain aspect is really frustrating. Two sides of the novelty coin, I suppose. Hopefully someday, I will learn to control some of my pain -mentally,- as I do my totally wicked tail. ;)

Just thought you all might find that interesting. Maybe someone should study me! ;)