Proprioception: The Sense Within

Knowing where our bodies are in space is critical for the control of our movements and for our sense of self.

In 1971, at the age of 19, Ian Waterman suffered a bout of severe viral gastroenteritis. The illness triggered an autoimmune response that stole his ability to gauge where his limbs were in relation to their environment. As described by Columbia University neurologist Jonathan Cole, Waterman was not paralyzed; his limbs moved, but he had no control over them. He felt disembodied as if he was floating in the air.

The five basic senses—sight, hearing, smell, taste, and touch—enable us to perceive the world around us. But what about sensations generated by the actions of our own bodies? As Waterman’s case demonstrates, the ability to sense our bodies is critical for telling us where we are in our surroundings as well as for the execution of normal movements. Sometimes referred to as the “sixth sense,” proprioception includes the sense of position and movement of our limbs, the senses of muscle force and effort, and the sense of balance. These senses, triggered by our everyday activities, allow us to carry out our tasks successfully, without thinking; absent feedback from proprioceptors, we, like Waterman, would be lost.

We remain largely unaware of the actions of the sense organs responsible for generating our proprioceptive senses. Have you ever stood in a darkened room and tried to touch the tip of your nose with your index finger? Most of us can do that with uncanny accuracy without even trying. But if we cannot see it, how do we know where our arm is as it travels through the air aiming for the nose? And how do we know where our nose is? To make things even more bizarre, if the biceps muscle of the arm touching the nose is vibrated, it generates the sensation that the arm is lengthening and the nose is beginning to grow. All of this is the subject of proprioception.

Research on proprioception has lagged behind work on the five basic senses, perhaps because it is a sense we are largely unaware of. However, during the last 50 years, neuroscientists have taken advantage of new stimulation and imaging techniques to achieve further insight into this elusive yet essential sense, acquiring new knowledge both at the receptor level and on the central processing of proprioceptive information.

Sensing limb position and movement

Reflections on how we sense our body’s movements date back to Galen in the 2nd century CE. But it wasn’t until the early 1800s that specific ideas were formulated about the mechanisms underlying proprioception. German physiologists proposed that there was no need for any peripheral sense organs in proprioception; rather, they believed that neurons in the brain driving muscle contractions sent copies of their signals to adjacent sensory areas to generate the required sensation. It was called a “sensation of innervation.”  At the turn of the 20th century, English neurophysiologist Charles Sherrington challenged this idea, on the grounds that we were aware of the positions of our limbs even when they lay relaxed, unmoving. Sherrington believed that there were sensory receptors in peripheral tissues that signalled position and movement. Today, elements of both sets of ideas contribute to the accepted view.

The most obvious place to locate a sense organ that signals position and movement of a limb is in the joints, and for many years it was believed that joint receptors were the principal proprioceptors. Recordings of responses of central neurons during joint movements supported this idea. But there were other possibilities. When the forearm is rotating about the elbow joint, there is a movement not only at the joint; muscles inserting into the joint—the elbow flexors and extensors—are stretched and shortened as well. In a landmark series of experiments in 1972, Guy Goodwin and colleagues at the University of Oxford provided evidence that receptors in muscles, not in joints, were the most likely candidates for generating our sense of limb position and movement.

Goodwin’s team showed that if the biceps muscle of one arm of a blindfolded subject was vibrated, the subject perceived the arm as extending, even though it had not moved at all. The subject indicated the illusion by tracking the movement with their other arm. The authors argued that vibration had stimulated muscle spindles, stretch-sensitive capsules found in most of our skeletal muscles. The response to vibration mimicked spindle activity generated by muscle stretch, leading to the illusion of a stretching biceps, that is, an extension of the arm. The vibration of the triceps muscle led to sensations of the arm moving into flexion—that is, the illusion that the triceps was being stretched. Importantly, vibrating the elbow joint did not produce any sensations of movement or displaced position, so the illusion could not be attributed to responses of the skin or joint receptors. 

Muscle spindles are unique in that they have two kinds of sensory nerve endings: the primary ending responds to both stretch of a muscle and the rate of a stretch; the secondary ending responds only to the stretch. Earlier animal experiments had shown that the primary endings are especially sensitive to muscle vibration, while the secondary endings are vibration-insensitive. In 1973, Ian McCloskey at the University of New South Wales in Australia showed that the illusion of arm extension was largest with vibration frequencies of 80–100 Hz; when the frequency was lowered, the illusion, predominantly of movement, morphed into one of displaced position. On the basis of these findings, he proposed that there were two senses: the sense of limb movement, generated largely by the primary endings of muscle spindles, and the sense of limb position, generated by both the primary and secondary endings.

Our “sixth sense” not only enables us to control the movements we make, but provides us with the ability to perceive ourselves moving in space and acting in relation to our surroundings.

Since then, the vibration illusion has been demonstrated many times at a variety of different joints, essentially confirming Goodwin’s initial findings. And in 1986, J.C. Gilhodes and colleagues of Centre National de la Recherche Scientifique (CNRS) in France further showed that if the two antagonist muscle groups acting at the elbow joint—the flexors and extensors—were both vibrated at the same time, there was no vibration illusion. If the frequency of vibration of one antagonist was lowered, an illusion gradually began to emerge, its size is directly proportional to the difference in vibration frequencies applied to the two muscles. These observations suggest that the brain is not processing signals from each muscle in isolation, but compares signals coming from the antagonist muscle groups and computes arm position from their difference.

In 2014, one of us (U.P.) and colleagues proposed that for this kind of arm-position matching task, it is not only the difference in signals from the antagonists that matter, but, in addition, the brain computes the difference in signals coming from both arms; when the difference is small, the arms are closely aligned.Supporting evidence for that view came the same year from Naoyuki Hakuta and colleagues of Showa University School of Medicine in Japan who showed that the size of the vibration illusion in one arm could be halved by vibrating the equivalent muscle of the other arm. It seems that the brain is constantly monitoring the movements of our arms relative to one another, most likely to be able to accurately align them for tasks such as the manipulation of objects and tools.

Matching and pointing

MATCHING GAME: Volunteers are blindfolded and their arms are strapped to two hinged paddles. Experimenters move one arm to a test position and ask the subjects to use their other arm to match the position of the first. In a different experimental setup (not pictured), only one arm is strapped in, and study participants indicate the position of the arm by moving the empty paddle to align with the perceived position of the hidden arm. Position errors from these two scenarios suggest that different receptors operating at the elbow joint and its attached muscles are responsible for the subjects’ assessments of arm position. COURTESY OF UWE PROSKE In addition to knowing where our limbs are in space, at least two other sensations contribute to our physical self-awareness: a sense of force and a sense of effort or heaviness. When we are asked to compare the heaviness of two objects of nearly the same weight, we typically juggle them up and down in our hands before making our judgment. It suggests that our sense of heaviness is closely allied to the sense of movement.

The vibration illusion used a limb position–matching task: subjects are asked to indicate the sensation generated in one arm by tracking it with their other arm. It is, in fact, a sensation-matching task. But in everyday life, we don’t go around matching the perceived positions of our limbs. I asked where we think our limbs are, we point to them.

Earlier this year, Anthony Tsay and colleagues at Monash University in Australia once again tested the vibration illusion, but this time, rather than tracking the illusion with the other arm, they asked subjects to point to the perceived position of the vibrated arm that remained hidden from view. Surprisingly, when the experiment was carried out in this way, subjects did not indicate any illusory displacement of their arm during vibration. Yet when Tsay and his team used the more traditional matching task, the same subjects demonstrated normal vibration illusions. Furthermore, characteristic position errors seen in a matching task following a muscle contraction that has been attributed to muscle spindles were no longer present in a pointing task. Taken together, these findings, suggest that in a pointing task the muscle spindles no longer play the dominant role of position sensors that they do in matching tasks; the source of the position signal changes depending on the nature of the task.

What, then, might be the position sensor in this pointing task? One possibility is skin receptors. In a matching task from one of our groups (S.G.’s), rhythmic stretching of the skin overlying a muscle can generate illusions of limb movement. When Tsay and colleagues tested this hypothesis, they did not find any evidence that skin stretch receptors contributed to position sense in a pointing task. Another candidate is the sensory nerve endings in the joints—as had been originally hypothesized during the first half of the 20th century. There is evidence from our experiments measuring movement-detection thresholds that joints can contribute a signal, at least for the fingers. The role of joints has not been studied in position matching and pointing tasks, however. This is a challenge for the future.

The body model

During an arm-matching task, the brain uses the difference in signal strength from the two arms to determine their relative positions. But in a pointing task, because we are determining the position of only one arm, the two-arm signal-difference mechanism cannot be used to indicate the position of the hidden arm.

In addition to knowing where our limbs are in space, at least two other sensations contribute to our physical self-awareness: a sense of force and a sense of effort or heaviness.

So how is position sense generated in a pointing task? Tsay and colleagues postulate that in pointing tasks, the position signal coming from the hidden arm accesses a map of the body located in the brain to determine arm position. In a 2010 study, Matthew Longo and Patrick Haggard of University College London asked subjects to place one hand under a table, out of sight, while with their other hand they pointed to the perceived positions of different landmarks on the hidden hand, such as the fingertips and knuckles. When responses were plotted on a map, they revealed a distorted shape, squatter and wider than the actual hand. The authors proposed that the brain uses information coming from the hand, including proprioceptive inputs, and combines them with a central map, or body model, to determine the location of the landmarks. The distortions in shape resembled those seen in sensory maps drawn on the human cortex many years earlier by neurosurgeons Wilder Penfield and Edwin Boldrey —the famed homunculus, which represents differences in the cortical innervation density of the body’s various parts. The distortions in the perceived shape of the hand described by Longo and Haggard may be related to the density of receptors on the surface of the hand, but it remains unclear how the information provided by the distorted proprioceptive maps is used to locate the position of the limbs in space.

When subjects were shown drawings of differently shaped hands, they were able to correctly select the one that was closest to the true shape of their hand. So while a body model generated by proprioceptive inputs showed characteristic distortions, another map called the body image—probably based on remembered visual information—provided a more accurate representation.

There are a number of serious, debilitating conditions associated with disturbances to the body image. These include eating disorders such as anorexia nervosa; conditions in which the patient denies that a part of their body actually belongs to them; and out-of-body experiences, where the patient believes that their body is no longer under their own control. Indeed, our very sense of self is believed to be generated in association with the central processing of proprioceptive information. Another related and well-known phenomenon is a phantom limb, where an amputated limb is perceived to continue to exist.

Our body image is labile and can be modified. Simultaneous touching of a hidden hand and a visible rubber hand lying near it leads the rubber hand to be adopted as part of the body. As we move about during everyday activities, there must be a continuous updating of the map, based on incoming movement information. This was first recognized many years ago and emphasizes the key link between the proprioceptive sensory system and the motor system.

While we have learned a lot in recent years about the peripheral signals responsible for the senses of limb position and movement, the picture continues to evolve. We are beginning to recognize that the source of the signals can change, depending on the task undertaken. Yet we still know relatively little about the central processing of the incoming information. How do we derive the metrics of body parts, for example, or process constantly changing spatial signals during ongoing body movements? This is an area where we should focus future research efforts.

The senses of effort, force, and heaviness

If our muscles are infused with a muscle relaxant, whatever we are holding, and even our limbs themselves, feel much heavier. According to work from one of our labs (S.G.’s), this is a disturbance in the sense of effort triggered by muscle weakness.  In response to this weakness, our motor neurons fire at an increased rate to generate the required level of muscle force, and this higher firing rate results in a greater perceived effort, giving the impression of increased weight. Similarly, whenever our muscles become weaker as a result of fatigue from exercise, the motor neurons increase their firing rate to compensate for the loss in force. That is why our limbs feel like lead at the end of vigorous exercise.

In a simplified view, the sense of effort is produced by impulses in the motor cortex that both travel down the spinal cord to the lower motor neurons to trigger muscle contraction and relay back to sensory areas of the brain, where the sensation of effort is generated. However, there is evidence from fatigue experiments, in which magnetic brain stimulation was used to mimic motor commands, that the sense of effort is generated somewhere upstream of the motor cortex and that the effort-force relationship undergoes constant adjustment. Furthermore, if the sensory and motor neurons supplying a limb are blocked and an attempt is made to try to move the paralyzed, anaesthetized limb, this can generate sensations of changed limb position and movement in the absence of any actual movement, S.G.’s group showed.  So the sense of effort is linked in some way not only to the sense of force but also to the senses of position and movement.

As well as having a centrally generated sense of effort, we are able to sense muscle force from the action of receptors specifically designed as tension sensors, the tendon organs. At each end of a muscle is a tendon, which anchors the muscle to bone; at the junction between tendon and muscle fibres lies a population of sensors called the Golgi tendon organs. Each consists of a large sensory axon that terminates on the tendon strands that connect at one end to the tendon proper and at the other to each of 10–20 individual muscle fibres. Each muscle fibre belongs to a different motor unit. The tendon organ will respond to the contraction of a single motor unit. Contraction of the whole muscle will engage a population of tendon organs, which send their impulses to the cerebral cortex with information about the amount of force exerted. So whenever we contract our muscles we have a centrally generated sense of effort accompanied by a sense of muscle force arising in our tendon organs.

In an experiment aimed to show this, subjects were asked to compare the stiffness of a series of compression springs. The subjects pressed a spring with one hand and used their other hand to select from a range of springs one with matching stiffness. Under control conditions, subjects were quite accurate in their choice. When the muscles of one hand were weakened by infusing a muscle relaxant, even though subjects now complained that their weakened hand required much more effort to compress the springs, they were still surprisingly accurate in choosing a spring of matching stiffness. But when subjects were instructed to match efforts, not forces, they made large errors in their comparison of spring stiffness. It seems we have the ability to selectively choose between our senses of effort and of muscle force, depending on the nature of the task.

Researchers have recently proposed that force signals of peripheral origin arise in both tendon organs and muscle spindles. If a muscle is progressively paralyzed, at the onset of the paralysis lifted objects feel heavier. As the paralysis deepens, paradoxically, lifted objects become lighter again. This result has been attributed to the action of muscle spindles. When the muscle starts to weaken from paralysis, the spindles remain unparalyzed, and their signals remain strong, contributing to the sense of increased heaviness. When the paralysis gets deep enough, the intrafusal fibers of spindles become paralyzed as well, leading to a reduction of the spindle signal, and as a result, the object feels less heavy than before. Thus, in addition to a centrally generated sense of effort, we have a peripherally generated sense of force or of heaviness, which arises from signals in both muscle spindles and tendon organs.

Getting a grip

Over the many months, after he suffered his proprioceptive loss, Waterman gradually learned to move again. At first, just standing stably was difficult. Using vision and a conscious will to move, Waterman, now in his mid-60s, is able to slowly combine muscle actions to achieve desired movements, such as lifting a cup of coffee. The more complicated the movement, the harder he has to think. Seeing his body and trunk are of critical importance. In the dark, he remains completely helpless.

Remarkably, Waterman is also able to compare the heaviness of objects of identical size as well as the rest of us who have an intact proprioceptive system—provided his eyes are open. It seems that the judges the heaviness of objects by observing the speed and extent of their movement when he lifts them.

Waterman’s unique case emphasizes the importance of proprioception in our daily lives. Our “sixth sense” not only enables us to control the movements we make but provides us with our sense of self, the awareness of our body and its movements as we navigate through our surroundings. As we unravel the neural mechanisms that underlie proprioception, we are learning more about how the brain processes sensory information. And that will ultimately lead us to a better understanding of ourselves.