An amputee struggles to use his new prosthetic limb. A patient with a frontal-lobe brain lesion insists that her left hand has a mind of its own. The alleged criminal claims in court that he did not fire the gun, even though several eyewitnesses watched him do it. Each of these individuals is grappling with two elements of the mind-body connection: ownership, or an ability to separate ourselves from the physical and social environments, and agency, a conviction that we have control over our limbs.

The human brain typically handles these phenomena by comparing neural signals encoding the intended action with those signals carrying sensory feedback. When we are born, we make erratic reaching and kicking movements to map our body and to calibrate our sensorimotor system. During infancy, these movements solidify our self-awareness, and around the time we first walk, we are quick to investigate a sticker placed on our forehead when looking in a mirror, recognizing the foreign object as abnormal. By the age of four, our brains are proficient at distinguishing self and other. In the amputee, the brain lesion patient, and the defendant on trial, the sense of self is disrupted due to discordance between sensory feedback from the limb and the brain’s expectations of how a movement should feel.

Instead of investigating ownership and agency as two distinct concepts, recent research has sought to understand how body ownership might have developed through the sum of agency experiences that we accrue throughout our life. What we perceive as our body is not only what looks like our body, but what we typically have conscious control over. This control is asserted by learned associations between our muscular movements and the sensory feedback we perceive when performing an action—the so-called “action effects.” What remains unclear, however, is just how multiple selves—including our bodily, social, and autobiographical selves—are integrated, and what kind of agency experiences drive the perception of having a single, stable self.

Bodily illusions

In 1937, French scientist J. Tastevin was testing perception of touch and finger position when he noticed that people often mistook a plastic finger protruding from underneath a cloth near their hand as their real finger. In the 1960s, French philosopher Maurice Merleau- Ponty described the way the body feels as “my hereness” and noted that perceiving oneself in a mirror extends this to a visually perceived self that is part of the external world, which Merleau-Ponty called “my thereness.” In doing so, he anticipated that self-recognition may be more than the immediate experience of the feeling of our bodies, that it may also involve the visual perception of our bodily self, which is quite similar to the way we perceive others. Over the next 40 years, research focused on the senses of touch and limb position, but little, if any, focus was given to mental representations of the body other than case studies of neurological disorders.

Fast-forward to 1998, when Princeton University cognitive scientist Matthew Botvinick headed a study on an illusion similar to the one Tastevin observed to evaluate body ownership.1 Participants sat with one arm under a table. Researchers placed a rubber arm on top of the table in alignment with the real arm below. The experimenters stroked the participant’s arm and the rubber hand either synchronously or asynchronously with a small paintbrush and asked the subjects to respond to a series of questions on body ownership. Subjects reported feeling as if the rubber hand were their own after synchronous but not asynchronous stroking. When asked to point to where they perceived their hand, participants tended to point toward the rubber hand, suggesting they had “embodied” the object as part of their own body. This rubber-hand illusion (RHI) suggests that the sense of self is highly malleable; the perceived location of the arm as estimated from the senses of touch and of limb position (known as proprioception) is superseded by visual input, as this sense is often more reliable.

Over the last 20 years, researchers have used the RHI setup to probe how we perceive our bodies. Some have measured physiological responses following threats to the rubber arm to objectively test if that limb is perceived as a part of the body. In 2003, Vilayanur Ramachandran of the University of California, San Diego, and Carrie Armel of Stanford University observed an increase in sweat production, known as skin conductive response (SCR), when they bent a finger on the rubber hand into a position that would normally be excruciatingly painful.2 A few years later, the Karolinska Institute’s Henrik Ehrsson, then at the Wellcome Trust Centre for Neuroimaging, and colleagues found that threatening the artificial limb with a needle increased activation of brain areas involved in bodily awareness and the anticipation of pain; the stronger the illusion, the stronger the activation of these regions.3

What happens to the “neglected” limb—the one under the table—reinforces the idea that we can learn to embody an artificial limb while disregarding the real one. In 2008, the University of South Australia’s Lorimer Moseley, then at the University of Oxford, and colleagues measured a decrease in the skin temperature of the concealed limb, suggesting reduced blood flow.4 The authors interpreted this physiological disembodiment of the real limb to be a consequence of taking ownership of an artificial body part. A few years later, Moseley and colleagues found that the real limb under the table exhibited increased histamine reactivity, a measure of the innate immune response, suggesting that the body had begun to reject the subject’s actual hand as it accepted the artificial one in its place.5

Illusions of embodiment can be induced in ways other than discordant visual-tactile stimuli. Receptors in skeletal muscle known as muscle spindles make a large contribution to our sense of proprioception. (See “Proprioception: The Sense Within,” The Scientist, September 2016.) Moseley and colleagues wondered if muscle spindles might also contribute to the sense of ownership. The researchers applied a nerve block to the index finger on one hand and blocked the participant’s view of it, effectively removing both visual and tactile inputs, while the subject’s other hand visibly grasped a rubber index finger. The experimenters found that when they synchronously moved the real index finger and the rubber finger, the participants reported feeling that the fake finger was embodied, suggesting that input from muscle spindles in conjunction with vision is sufficient to generate ownership.6

While the above experiments probing the sense of body ownership are admittedly contrived scenarios, the fact that people can learn to embody a limb that they weren’t born with has major implications for amputees. Knowing what we now do about body ownership, can we help amputees fully embrace their prosthetic limbs?

There’s some evidence that we can. In the mid-1990s, researchers at the Toho University School of Medicine in Tokyo, Japan, trained macaques to use a rake to retrieve objects and found that neurons responsive to touch stimuli from the hand and to visual input now also respond to the rake when it was in use.7 The RHI is one way to induce a person to accept an artificial limb as her own, but amputees obviously lack skin, muscles, and neurons in their missing limb to stimulate. As a workaround, researchers have begun to use electrical stimulation of the brain regions thought to be involved in representing the body to mimic the effects of stroking a real limb. Last year, working with two patients undergoing brain surgery for epilepsy, Kelly Collins of the University of Washington and colleagues stimulated the region of the somatosensory cortex corresponding to one hand while touching a rubber hand visible to the participants.8 Both patients had a strong sense of ownership over the artificial limb, which attenuated when electrical stimulation was moved to other regions of the cortex. This study provides hope that a similar procedure could help train amputees to embody their prosthetic limbs.

THE RUBBER-HAND ILLUSION
A classic experiment to test the idea of body ownership is to have volunteers place a hand out of view under a table and set a rubber hand on the top the table. When researchers touch the real hand and the rubber hand synchronously, participants will feel as if the rubber hand were their own. Researchers have observed that brain areas including the premotor, somatosensory, and parietal cortices, candidate regions for identifying and representing self, are activated in response to the now-embodied fake hand 􀁑1 . If the rubber hand is physically threatened, volunteers will often begin to sweat, indicating they feel as if they are at risk of injury. At the same time, activity increases in the insula and anterior cingulate cortex, deep brain regions responsible for bodily awareness and pain anticipation (PNAS, 104:9828-33, 2007) 􀁑2 . Meanwhile, the real hand that is under the table—known as the “neglected” limb—experiences a decrease in skin temperature, suggesting reduced blood fl ow, and an increased innate immune response, suggesting that the body has begun to reject the real hand as it accepts the artificial one as its own 􀁑3 .

Increased embodiment should not only improve amputees’ control over their prosthetics, but may help reduce the phantom pain that many amputees suffer. To date, effects of the RHI on pain reduction are equivocal, but a couple of studies suggest that variations of the illusion could have potential, as they generate stronger,more holistic feelings of ownership. In 2011, Ehrsson and colleagues completed an experiment in which participants, who were upper limb amputees, viewed a mannequin through goggles that relayed a video feed with a first-person view of the artificial body. The researchers stroked either the intact hand of a complete mannequin or the stump and the area below the stump of an amputated mannequin, while at the same time stroking the stump of the participants. Although the illusion only worked under certain condi tions, two of four participants did report remarkably reduced pain after synchronous stroking of their stump and the mannequin's stump or the area below it.9 And earlier this year, James Pamment and Jane Aspell at Anglia Ruskin University in the U.K. induced a similar full-body illusion in 18 people suffering from various pain conditions. They found that the illusion reduced pain ratings by 37 percent in this cohort.10 Together, these studies point to the possibility that wholebody illusions could be used to overcome phantom limb pain in amputees as well as other pathologies marked by pain.

MEETING EXPECTATIONS
The reafference principle posits that we are constantly comparing feedback from sensors in our limbs (termed reafference) and other sensory systems (e.g., visual, auditory, etc.) with our expectations based on our intended movements (termed efference copy). Disagreement between these signals (termed exafference) can lead to reduced feelings of embodiment and agency.

A better understanding of how body ownership is encoded in the brain could also one day help treat patients with more extreme body illusions, such as the brain lesion patient who has lost control of her left hand or the defendant who insists he did not fire the gun. Damage in multisensory areas of the brain—particularly the transition between the parietal and the temporal cortex, the so-called temporoparietal junction, and parts of the medial frontal cortex11—may result in an object being incorrectly embodied or in the disembodiment of a limb or even the whole body, as in the case of patients having outof- body experiences.12 The neural correlates and brain mechanisms leading to self/other discrimination and out-of-body illusions could one day be targeted to help patients suffering from disorders that result in abnormal bodily consciousness.

Being in control

Beyond ownership, the sense of agency is a conviction that we have control over the events we initiate. We have control when we reach for a glass of water, when we kick a football, and when we put pen to paper.

Based on theoretical ideas of 19th century physician and physicist Hermann von Helmholtz, German scientists Erich von Holst and Horst Mittelstaedt demonstrated the reafference principle in 1950 to distinguish between self-generated movements and external perturbations. Any time we move, we generate a motor command (efference) to control the muscles. At the same time, we also generate a prediction—based on prior experience of the sensation resulting from the movement—termed the efference copy. The actual movement-related sensory input, which comes from receptors in the muscle and skin, is referred to as reafference. Any difference between the two signals (reafference and efference copy) is the result of environmental input, which is termed exafference. Understanding errors that may occur within this system is probably central to understanding problems in agency and ownership perception.

In a letter to Oliver Sacks, which the late neurologist and author published in his 1984 book A Leg to Stand On, Russian neuropsychologist Alexander Luria stated, “If a part of the body is split off from action, it becomes ‘alien’ and not felt as part of the body.” This occurs in patients with alienhand syndrome, for example, a debilitating condition that leaves sufferers with no control over an arm. Some patients have to strap their arm to their chest before sleep so that they don’t punch themselves in the middle of the night. Luria’s view suggests that, in this situation, the sensation returning from the limb is considered pure exafferent input, as if there were no conscious prediction that the limb should move, and as such it is disembodied. But are agency and ownership really this dependent on each other?

In 2005, Manos Tsakiris of Royal Holloway, University of London, and colleagues found that indeed they are. In the group’s study, a lever that lifted a participant’s passive right index finger was operated either by the participant’s own left hand or by an experimenter, so that movement of the right hand was effected either voluntarily by the subject or externally by the researcher. Participants could not see their hands, but on a screen they saw a video stream of a gloved right hand—either their own or someone else’s— with its index finger being lifted by a lever and they were asked to say whether or not it was their own. The subjects were substantially more accurate at identifying their own hand when the movement was voluntary— and thus the motor command and sensory feedback they received were in agreement— suggesting that agency is critical to self-recognition, a key component of ownership.13

Another line of evidence in support of the interdependent relationship between agency and ownership comes from the work of Hiroshi Ishiguro of Osaka University in Japan. As humans, we can harness the power of imagination to test things out before we enact them; imagining moving a limb produces substantial activation of the limb-specific movement-planning areas of the brain, and people controlling prosthetic limbs to perform basic tasks have activity in these same areas. Ishiguro and colleagues fitted participants with a head-mounted display through which they viewed robotic hands. Participants imagined moving the hands, and the resulting neural activity was recorded via EEG and used to command the robotic hands. When participants could not move the hands, their feelings of ownership, as measured by response to limb threat, were halved, suggesting that embodying a new limb is optimized when one is able to move it and receive visual feedback.14

Past control over an object—the experience of agency over that object—might also contribute to embodiment. In a study performed by one of us (R.L.) and colleagues, researchers found that a rubber hand, a smartphone, and a wooden block were in principle all perceived as embodied in an adapted version of the RHI. But when participants were asked to verbally estimate the position of a limb, covered by a box placed next to the rubber hand and object, they perceived their hand as being closer to the smartphone and the rubber hand, but not toward the wooden block, which people had no previous agency experience with.15 This finding suggests that there is a direct impact of past agency experience on ownership.

Other research has suggested that agency is partly separable from ownership, however. In 2012, Ehrsson, along with his then graduate student Andreas Kalckert, designed a rubber-gloved wooden model hand to make finger movements that were either linked by a wooden rod to (and thus synchronous with) movements of the participant’s own hidden hand, or detached and controlled independently by the experimenter. 16 Initiation of synchronous movements by the participant elicited a strong sense of ownership and agency over the model hand; linked, synchronous movements initiated by the experimenter (passive movements) abolished the sense of agency, while the sense of ownership remained intact. Conversely, when the experimenters rotated the robotic hand by 180 degrees—putting it in an anatomically implausible position, with the fingers facing toward the body—participants maintained a sense of agency, but not of ownership.

This double dissociation suggests these two components of self are partly processed separately when deprived of the usual multisensory inputs. But in the real world, the evidence all seems to point toward the interdependence of agency and ownership. Perhaps the best example of this is the bizarre case of Ian Waterman, one of very few people without sensation of touch or limb position below the neck, lost in an autoimmune episode when he was 19. Neurophysiologist Jonathan Cole of Bournemouth University, who has studied Waterman for many years, explains that “Ian felt ‘disembodied’ only at the beginning, when he had no agency” and when he was not looking directly at his body; Waterman only required vision with crude movement control to regain ownership of his body. Because Waterman receives no peripheral feedback, he has to consciously think about his movements, Cole adds, and as a result, “he feels more cognitively embodied than we might.” During a recent visit to NASA, Ian was able to control a full-body robot.17 When a trolley careened toward the robot, he immediately tried to protect his “new self ” by avoiding it.

Given our seemingly boundless potential to attribute agency and ownership to inanimate extensions of ourselves, it is hard to predict how we might interact with our surroundings in the future. It is possible that we might one day control robots with our bodies and our minds. Nearer-term, the rise of brain-machine interfaces and neuroprosthetics will further blur the line between “me” and “mine,” and will inform the design of prosthetics that move more naturally so that they can be more easily “embodied.” A better understanding of the link between the sense of agency and actions themselves will also have implications for treating rare disorders of self, and raise ethical questions about the legal treatment of those who claim at some point to have lost control of their bodies. 􀁊

Roman Liepelt is a senior lecturer in psychology at the German Sport University in Cologne. Jack Brooks is a PhD student in physiology at the Neuroscience Research Australia at the University of New South Wales.

References

1. M. Botvinick, J. Cohen, “Rubber hands ‘feel’ touch that eyes see,” Nature, 391:756, 1998.

2. K.C. Armel, V.S. Ramachandran, “Projecting sensations to external objects: Evidence from skin conductance response,” Proc R Soc Lond, 270:1499-506, 2003.

3. H.H. Ehrsson et al., “Threatening a rubber hand that you feel is yours elicits a cortical anxiety response.” PNAS, 104:9828-33, 2007.

4. G.L. Moseley et al., “Psychologically induced cooling of a specific body part caused by the illusory ownership of an artificial counterpart,” PNAS, 105:13169-73, 2008.

5. N. Barnsley et al., “The rubber hand illusion increases histamine reactivity in the real arm,” Curr Biol, 21:R945-R946, 2011.

6. L.D. Walsh et al., “Proprioceptive signals contribute to the sense of body ownership,” J Physiol, 589:3009-21, 2011.

7. A. Iriki et al., “Coding of modified body schema during tool use by macaque postcentral neurones,” Neuroreport, 7:2325-30, 1996.

8. K.L. Collins et al., “Ownership of an artificial limb induced by electrical brain stimulation,” Nature, 114:166-71, 2017.

9. L. Schmalzl et al., “Pulling telescoped phantoms out of the stump: Manipulating the perceived position of phantom limbs using a full-body illusion,” Front Hum Neurosci, 5:121, 2011.

10. J. Pamment, J.E. Aspell, “Putting pain out of mind with an ‘out of body’ illusion,” Eur J Pain, 21:334-42, 2017.

11. R. Liepelt et al., “The medial frontal cortex mediates self-other discrimination in the Joint Simon task: A tDCS study,” J Psychophysiol, 30:87-101, 2016.

12. B. Lenggenhager et al.,. “Video Ergo Sum: Manipulating bodily self-consciousness,” Science, 317:1096-99, 2007.

13. M. Tsakiris et al., “A specific role for efferent information in self-recognition,” Cognition, 96:215-31, 2005.

14. M. Alimardani et al., “Humanlike robot hands controlled by brain activity arouse illusion of ownership in operators,” Sci Rep, 3:2396, 2013.

15. R. Liepelt et al., “Self-perception beyond the body: The role of past agency,” Psychol Res, doi:10.1007/s00426-016-0766-1, 2016.

16. A. Kalckert, H.H. Ehrsson, “Moving a rubber hand that feels like your own: A dissociation of ownership and agency,” Front Hum Neurosci, doi:10.3389/fnhum.2012.00040, 2012.

17. J. Cole, Losing Touch: A man without his body, Oxford, U.K.: Oxford University Press, 2016.