With eyes closed, a person is asked to place one hand where an examiner has positioned the other. The hands do not quite align. That error may reflect an imprecise sense of limb position, but it might also reflect difficulty moving the matching arm, remembering the target, comparing two sides of the body or understanding the instruction. Calling the result “poor proprioception” assumes the task has isolated a sensory contribution it may not have isolated. Measurement begins by naming which bodily information is being tested and which other operations the response requires.
Proprioception includes senses of limb and trunk position, movement, effort and force, as Proske and Gandevia’s review explains. These components are related but need not share a single score. This Methods & Measures entry concerns how researchers make their differences visible. It is distinct from sensorimotor prediction, which asks what sensory consequence an action is expected to produce, and motor imagery, which asks about simulated movement without executing it. A person may predict poorly despite receiving accurate position signals, or report vivid movement imagery despite imprecise passive-motion detection.
Choose the component before the instrument
A joint-position task asks where a limb or joint is. A passive-motion task asks whether it moved, in which direction, or by how much. A force task asks what effort or load is sensed. “Body awareness” is too broad for any one of these. Researchers should state the target component before choosing a convenient clinic test, because different tasks recruit different sensory inputs and different memory and motor demands.
Position can be measured through a matching task: one limb is placed at a reference angle and the participant reproduces that angle with the same or opposite limb. Same-limb matching may require a delay and therefore memory. Opposite-limb matching avoids some delay but requires transforming information between sides and may be affected by asymmetry. Active matching adds motor planning and execution; passive matching reduces those demands but adds reliance on the examiner or apparatus to move the limb. Neither version is a pure reading of receptors.
Movement detection has a different logic. An apparatus moves a relaxed joint slowly while vision is blocked, and the participant signals when movement starts or states its direction. The threshold depends on speed, starting position, direction, attention and response rule. A low threshold suggests sensitivity under that exact protocol. It is not automatically equivalent to accurate position reproduction or skilled movement in everyday life.
Psychophysical discrimination asks whether two positions or movements differ. That is not the same as merely detecting that something moved. A participant may detect very small motion yet be less precise when comparing two amplitudes after a delay. A paper comparing matching methods with psychophysical thresholds highlighted this distinction and the influence of active versus passive movement. A study should avoid putting detection threshold, discrimination threshold and matching error into one undifferentiated “acuity” category.
Understand the sensory pathway without claiming to read it directly
Muscle spindles, tendon and joint afferents contribute to proprioceptive experience, while motor commands and skin information also shape the estimate. Goodwin, McCloskey and Matthews applied tendon vibration and observed systematic illusions of elbow angle. This is powerful evidence that muscle-related signals can influence perceived position. It does not mean every position error in a clinical test is caused by a spindle deficit. The task perturbs several sensory and motor relations and demonstrates a mechanism in a particular setting.
Vibration can be used experimentally to test weighting of input. If the apparent arm position changes when a tendon is vibrated, a researcher can compare the illusion across posture, vision and movement conditions. The vibration’s frequency, site and intensity must be reported, because physical differences can change the effect. The participant may also notice the stimulation and alter their response consciously. A mismatch between perceived and actual angle is evidence about the integrated estimate, not a count of firing from a single receptor class.
Vision often improves performance and can compensate for uncertain limb signals. Removing visual feedback is therefore common, but it changes the task. A person may be less confident, move more slowly, or rely on memory and attention differently. If the aim is a functional everyday outcome, vision may be part of the relevant system rather than a nuisance to eliminate. If the aim is a proprioceptive estimate, vision should be controlled and the narrowed inference stated.
Effort and force deserve separate treatment. Sensing how much force one produces is not identical to knowing joint angle. Muscle fatigue can alter effort for the same external force, and an external load can change force sensation without the same movement. A force-matching task should distinguish the force applied from effort to generate it. Proske and Gandevia’s broader account is a reminder that the proprioceptive senses form a family, not a single pathway with one universal threshold.
Measure matching error carefully
Absolute error records how far a reproduced position is from the target, without regard to direction. Signed error records bias, such as consistently overshooting clockwise. Variable error records consistency across trials. These summaries answer different questions. A participant might have a large stable bias but little trial-to-trial noise, while another has little average bias because errors cancel but high variability. Reporting only the mean signed error can make poor precision look like excellent accuracy.
Trial count matters. One matching attempt can be disturbed by distraction or awkward posture. Repeated attempts allow estimation of variability, but repetition also lets a participant learn the task or infer the examiner’s preferred answer. Randomize target positions, include more than one direction and report whether feedback was given. If participants can see the true angle after each attempt, later trials measure a mixture of proprioception and calibration learning.
Range of motion matters too. Near a joint limit, tissue tension and skin stretch may provide extra cues. A test at one comfortable angle may not generalize to another part of the range. The shoulder, wrist and knee have different mechanics and available cues. An elbow result should not be treated as a global body-position score. Researchers should choose tested positions based on the intended function and specify them in the report.
Dukelow and colleagues used a robotic arm-position matching task after stroke and found quantitative differences that a conventional clinical thumb-localizing test did not always agree with. The lack of complete agreement is informative. The methods may sample different components and have different resolution, not simply one right and one wrong score. Their study also reported reliability evidence for the robotic method, which is necessary if a test is to track change rather than only describe group means.
Equipment introduces its own error. A goniometer, photograph, motion tracker or robot measures physical angle with finite precision and may constrain movement. Examiners can differ in how they place a limb or judge a response. Calibration and blinding reduce avoidable variation. A systematic review of knee joint-position tests found that reliability differed by method and was sometimes variable. This makes protocol details part of the construct validity argument rather than administrative trivia.
Separate sensation from movement ability
An active matching test can fail because the participant cannot move to the target even if they sense where it is. Weakness, pain, spasticity or limited range of motion can constrain the response. A passive test can reduce motor demands but may still require communication and sustained attention. Before interpreting an error as sensory, the examiner should document movement capacity and whether the participant can execute the response format.
After stroke, impairment can be heterogeneous. One arm may provide an uncertain reference while the other also moves imprecisely. Opposite-arm matching assumes the “unaffected” side is a trustworthy instrument, which may not always be true. A task that measures each arm’s position against an external reference can help, but it introduces visual or apparatus cues. No single protocol removes all assumptions; the design should identify which side supplies the target and which side produces the answer.
Memory and language are further confounds. If a position is shown, moved away and reproduced several seconds later, the task includes retention of the target. Verbal judgments such as “higher” or “lower” require understanding the terms and mapping a bodily sensation onto words. A participant with language difficulty may answer incorrectly despite intact sensation. Nonverbal response modes may help but create other motor requirements. Clinical interpretations should account for the person’s available communication.
Attention can change thresholds. A participant listening for tiny passive movement may perform differently when tired or distracted. Repeated testing can be especially demanding when the movement is deliberately near threshold. A threshold estimated from a few yes/no responses is sensitive to willingness to say “yes.” Forced-choice designs and catch trials can help separate sensitivity from response criterion, provided that the participant understands the choice and does not merely guess in a patterned way.
Link laboratory measures to meaningful function
It is tempting to infer that a small joint-position error predicts better balance, sport skill or recovery. That link needs direct testing. Daily movement combines proprioception with vision, vestibular information, motor planning, strength and environmental feedback. A laboratory difference may be real yet explain little of functional performance. Researchers should choose functional outcomes that match the joint and task, then measure whether the proprioceptive test predicts them beyond other relevant capacities.
A training study introduces another inference problem. Repeated practice on the same matching task can improve scores through familiarity with the apparatus or strategy. To claim improved proprioceptive function, include an appropriate comparison group and a transfer measure that differs from the practiced test. If only the trained angle improves, the conclusion should be narrow. If several movement and position measures improve, the claim may be stronger, but each has its own error sources.
Reliability should be considered before change. If a person’s score varies widely from day to day, a small post-training gain may be indistinguishable from noise. Test-retest intervals, rater consistency and measurement error should be reported for the intended population. Reliability in healthy adults does not automatically transfer to children, older adults or people with pain or neurological injury. A score’s clinical meaning depends on how it behaves where it will be used.
Group comparisons also need age and context checks. Older and younger participants may differ in joint mobility, strength, vision and familiarity with testing. An athletic group may move more confidently without having a lower passive detection threshold. A clinical group may take medication or experience pain. Matching or measuring these factors helps constrain interpretation. Proprioception can be a plausible mechanism without being the only explanation for a group difference.
Make the construct visible in the report
A useful report says whether the measure was active or passive; which joint, angle, speed and side were used; what visual and tactile cues were available; how many trials occurred; and how error was scored. It distinguishes detection from discrimination and position from movement. These details allow a reader to understand what sensation could have supported the task and what non-sensory demands were present.
When several tests disagree, the investigator should examine what each asks. A person may detect passive wrist motion accurately yet overshoot an actively matched arm position because motor execution is difficult. Another may match positions well by using skin stretch or learned anchors while showing a higher passive-motion threshold. Divergence is not an excuse to choose the most favorable score. It can identify a narrower mechanism worth testing.
The Institute’s measurement questions synthesize the cited psychophysical, robotic and clinical methods; they are not a validated assessment scale. The practical discipline is to resist saying “proprioception improved” until the particular component, protocol and uncertainty are named. A score is a result of a person interacting with a task, not a direct quantity taken from a muscle spindle.
What the evidence does not show
The cited studies do not establish one test that covers position, movement, effort and force equally well. A low error in matching one joint does not prove precise proprioception throughout the body. A high error does not establish a peripheral sensory lesion; motor, cognitive, pain and task factors may contribute. Tendon-vibration illusions show that muscle-related input can shape position perception, not that vibration responses diagnose everyday bodily awareness.
Reliability evidence for one robot or knee protocol does not validate every clinic or home version. A statistically different average between groups does not tell whether an individual has a meaningful impairment or whether training would help them. Functional predictions require their own tests. Even when proprioceptive information is altered, vision and other senses can support action, so a deficit on a blindfolded task need not describe every real-world ability.
These measures cannot by themselves establish how the nervous system predicts the sensory effects of a movement. Prediction may influence performance, but it is a separate construct requiring its own perturbations and controls. Nor does vivid motor imagery guarantee accurate proprioceptive input. Keeping these distinctions allows each task to contribute evidence without claiming to measure the whole body in one number.
Return to the two hands
The original mismatch becomes more informative if the researcher records whether the target hand was moved passively, whether the participant matched with the same or opposite arm, how long the delay was, and whether vision or feedback was available. Repeated trials can show whether error is biased or variable. A passive-motion threshold might then test a related but distinct aspect of sensing movement. The answer is not simply whether the hands met; it is what the test required to bring them together.
Proprioception lets a body act without watching every joint. Measuring it well requires the same care as any other sensory science: define the signal, control the alternative cues, choose a response the person can make and report uncertainty. A precise result may be narrower than “body awareness,” but it is more useful. It tells us which information was available for a particular movement and where the evidence for that conclusion ends.
Research references
- Proske, U., & Gandevia, S. C. (2012). The proprioceptive senses: Their roles in signaling body shape, body position and movement, and muscle force. Physiological Reviews. View source
- Goodwin, G. M., McCloskey, D. I., & Matthews, P. B. C. (1972). Proprioceptive illusions induced by muscle vibration: Contribution by muscle spindles to perception? Science, 175, 1382–1384. View source
- Elangovan, N., Herrmann, A., & Konczak, J. (2014). Assessing proprioceptive function: Evaluating joint position matching methods against psychophysical thresholds. Physical Therapy, 94, 553–561. View source
- Dukelow, S. P., et al. (2010). Quantitative assessment of limb position sense following stroke. Neurorehabilitation and Neural Repair. View source
- Smith, T. O., Davies, L., & Hing, C. B. (2013). A systematic review to determine the reliability of knee joint position sense assessment measures. The Knee, 20, 162–169. View source
- Ager, A. L., Roy, J.-S., Roos, M., Fournier Belley, A., Cools, A., & Hébert, L. J. (2017). Shoulder proprioception: How is it measured and is it reliable? A systematic review. Journal of Hand Therapy, 30, 221–231. View source