The Neuroprosthetic Revolution: an Agonist-Antagonist Myoneural Interface
Hiteyjit Singh Gujral
Introduction:
It is quite difficult to imagine what someone with a prosthetic feels. A part – made not of your own skin, but some metallic substance – attached to your body to mimic the function of what was once part of your own self.
For millions of individuals living with limb amputations worldwide, the dream of fully restored movement and sensation has long remained out of reach. Traditional amputation techniques, while life-saving, often leave patients with significant functional limitations. However, there is hope. Recent innovations seeking to bridge the gap between neuroscience and technology are bringing an exciting change to the landscape of amputation surgery and prosthetic technology by preserving the body's natural sensory feedback mechanisms.
Proprioception:
Before the AMI procedure can be understood, it is necessary to describe the basis on which it acts. How do our limbs sense how they move?
Proprioception is a sense to help the mind understand positions of the body’s components in open space and relative to oneself (Blumer et al., 2024). If one closes their eyes, an accurate determination of the location of one’s legs, arms, and hands is still possible, even without visual input. This unique sensation, often termed as the ‘sixth sense’, utilizes receptors placed within the body itself, relying on cues from muscles and joints – rather than external stimuli (Proske & Gandevia, 2012). Leading theories suggest three major components at play: muscle spindles, Golgi tendon organs, and joint receptors. Muscle spindles, present in nearly every skeletal muscle, are responsible for informing the central nervous system of changes in muscle length (Kröger & Watkins, 2021). Golgi tendon organs, placed in series (contrasting with the parallel placement of muscle spindles) with muscle fibers, respond more as ‘tension sensors’, accurately providing information about rate of contraction and loads placed on a muscle (Morimoto & Takada, 1993). Joint receptors, as the name suggests, are present within joints, and are responsible for perceiving a change in angles of a joint (Macefield, 2024).
Almost all muscles are present in pairs. Within these pairs, muscles act antagonist to each other, hence their name: the agonist-antagonist pair. For example, when one attempts to contract their bicep, their tricep will stretch (BBC Bitesize, 2019). Similarly, when one attempts to stretch the tricep, the bicep relaxes and stretches. In essence, this linkage is critical for proprioception to occur, where muscle spindles and Golgi tendon organs are only able to function due to the true changes in length and tension of each muscle. The information of stretch and tension, created by these pairs working together, picked up by their respective proprioceptors, can then be communicated actively to the nervous system for further use (MIT Media Lab & Herr, 2017).
Traditional amputations:
Surgical amputations, or the act of removing an individual’s limb through surgery (Shores, 2019), has been around for many centuries, with the earliest trace being more than 30,000 years old (The University of Sydney, 2022). Since then, many advancements in surgical techniques have been devised to help provide a much better transition back to everyday life. Techniques have been standardized, mortality rates have improved (O’Banion et al., 2022), quality of life has been made better (Ovadia & Askari, 2015), post-op therapeutics have increased in efficacy (Schug & Gillespie, 2011), and modern prosthetic development has catalysed a major restoration in previously-lost functions.
However, even since modern surgical techniques have been established, prioritization has remained largely on wound healing and creation of a viable prosthetic socket attachment region. As a result, muscle pairs and neuromuscular connections are often severed, with muscle tissue instead forming a padded region around the bone (Herr & Carty, 2021). Neurological connections and systems are often severed and given less attention, causing nerve endings to simply be left in free space. Although traditional amputations do attempt to reduce such consequences, the disruption of neural connections can often cause painful repercussions. Nearly 25% of amputees experience Neuromas (O’Reilly et al., 2016), causing extreme pain especially if occurring in areas where there is high external pressure (eg. a prosthetic socket). Almost 70% of amputees experience forms of phantom/residual limb pain (Hobusch et al., 2020), and many simply reject their prosthetic after trying it. Moreover, severance of musculature and neuronal connections at the site of amputations would make it extremely difficult for the brain to experience sensation (including proprioception) in that area. Researchers have now attempted to devise methods to re-establish these lost sensations, with a particular technique showing promise: the Agonist-Antagonist Myoneural Interface (AMI).
What is the AMI:
Invented in 2014 at the Massachusetts Institute of Technology, AMI takes advantage of the naturally occurring agonist-antagonist muscle pairs within the body at the site of amputation (Herr & Carty, 2021). Following the surgical removal of a limb, instead of leaving muscles to simply heal on their own, the AMI involves mechanically reconnecting otherwise severed muscle pairs. Since nerves and neuromuscular architecture can be conserved through creating an AMI unit, the residual limb’s proprioceptive abilities remain established (Herr & Carty, 2021). Especially since a muscle system’s fundamental functionality of contracting an agonist and stretching an antagonist continues to function, the brain is able to obtain kinesthetic-like feedback for signals that would otherwise be sent to a ‘phantom’ limb (Herr & Carty, 2021).
Through this method, even patients with a passive prosthetic limb have reported greater control and feel as if the limb is not purely an ‘external attachment’ (MIT Media Lab & Herr, 2017). From a neurophysiological standpoint, functional activation of brain regions associated with proprioception did not significantly differ between those with AMI amputations and non-amputee controls (Srinivasan et al., 2020). Compared to a marked decrease in activation of this region for those with non-AMI amputations, this difference exhibits a high potential for AMI to conserve traditionally lost sensorimotor abilities within amputees. Further, observations of an AMI patient have shown greater intentional control of their prosthetic’s movement compared to those with traditional amputation. In addition, during tasks such as navigating stairs, all natural movement patterns – that would otherwise not be possible with passive prosthetics – can also be observed as an involuntary, reflexive behavior in an AMI patient (Clites et al., 2018).
Moreover, in certain cases, when impulses are sent from the brain, sensors placed around the AMI units can detect changes in length/contraction of these reconstructed muscles and transmit such information into robotic prosthetics for translation into desired movements. According to the creators of AMI, each myoneural interface can allow for one axis of movement within a prosthetic, as each AMI is responsible for a singular muscle system (Herr & Carty, 2021). Moreover, through this method, the AMI is also able to facilitate reverse communication, i.e the communication of signals from the neuroprosthetic back to the user (Herr & Carty, 2021). With a prosthetic providing feedback on its functioning through electrical measurements (eg. impedance, etc), muscles within the AMI can be electrically stimulated to artificially replicate proprioceptive signals that can then be sent to the brain (Herr & Carty, 2021).
When created, the AMI framework could only make use of muscle systems that were pre-planned to be leveraged by AMIs post-amputation. However, more recent developments of AMI through integrating techniques such as regenerative grafting has enabled even those with prior amputations to have functionality restored with the AMI framework (Srinivasan et al., 2019).
Looking Forward:
Although originally developed primarily for lower-limb amputations, AMI’s principles can be adapted even further for upper extremity applications, where dexterous control and sensory feedback may be more critical. Where AMI has helped resolve amputative proprioceptive loss, combining it with sensory reinnervation techniques could allow for potential restoration of tactile feedback alongside what AMI can currently achieve. Furthermore, for higher-level amputations, researchers can even look towards exploring hierarchical arrangements of multiple AMI pairs to control different prosthetic joints simultaneously.
Conclusion:
The Agonist-Antagonist Myoneural Interface represents a significant advancement in amputation surgery by preserving the body's natural proprioceptive mechanisms. By maintaining the dynamic relationship between opposing muscle groups, AMI restores critical neural feedback pathways that inform the brain about limb position and movement.
Patients have shown improvements across the board following the conservation of the proprioception, especially with greater phantom limb awareness, more intuitive control of prosthetic devices, and greater natural movement patterns
Very few modern techniques are capable of such sensory integration with one’s neuroprosthetic. It is very exciting to witness such innovations allow those with artificial limbs to not feel isolated from their prosthesis. The AMI approach represents more than just an incremental improvement in amputation surgery – it signals a paradigm shift in how we can approach interfaces between human neuromuscular systems and prosthetic technology.
About the Author
Hiteyjit Singh Gujral (‘27) is a sophomore at Harvard College concentrating in neuroscience.
References
It is quite difficult to imagine what someone with a prosthetic feels. A part – made not of your own skin, but some metallic substance – attached to your body to mimic the function of what was once part of your own self.
For millions of individuals living with limb amputations worldwide, the dream of fully restored movement and sensation has long remained out of reach. Traditional amputation techniques, while life-saving, often leave patients with significant functional limitations. However, there is hope. Recent innovations seeking to bridge the gap between neuroscience and technology are bringing an exciting change to the landscape of amputation surgery and prosthetic technology by preserving the body's natural sensory feedback mechanisms.
Proprioception:
Before the AMI procedure can be understood, it is necessary to describe the basis on which it acts. How do our limbs sense how they move?
Proprioception is a sense to help the mind understand positions of the body’s components in open space and relative to oneself (Blumer et al., 2024). If one closes their eyes, an accurate determination of the location of one’s legs, arms, and hands is still possible, even without visual input. This unique sensation, often termed as the ‘sixth sense’, utilizes receptors placed within the body itself, relying on cues from muscles and joints – rather than external stimuli (Proske & Gandevia, 2012). Leading theories suggest three major components at play: muscle spindles, Golgi tendon organs, and joint receptors. Muscle spindles, present in nearly every skeletal muscle, are responsible for informing the central nervous system of changes in muscle length (Kröger & Watkins, 2021). Golgi tendon organs, placed in series (contrasting with the parallel placement of muscle spindles) with muscle fibers, respond more as ‘tension sensors’, accurately providing information about rate of contraction and loads placed on a muscle (Morimoto & Takada, 1993). Joint receptors, as the name suggests, are present within joints, and are responsible for perceiving a change in angles of a joint (Macefield, 2024).
Almost all muscles are present in pairs. Within these pairs, muscles act antagonist to each other, hence their name: the agonist-antagonist pair. For example, when one attempts to contract their bicep, their tricep will stretch (BBC Bitesize, 2019). Similarly, when one attempts to stretch the tricep, the bicep relaxes and stretches. In essence, this linkage is critical for proprioception to occur, where muscle spindles and Golgi tendon organs are only able to function due to the true changes in length and tension of each muscle. The information of stretch and tension, created by these pairs working together, picked up by their respective proprioceptors, can then be communicated actively to the nervous system for further use (MIT Media Lab & Herr, 2017).
Traditional amputations:
Surgical amputations, or the act of removing an individual’s limb through surgery (Shores, 2019), has been around for many centuries, with the earliest trace being more than 30,000 years old (The University of Sydney, 2022). Since then, many advancements in surgical techniques have been devised to help provide a much better transition back to everyday life. Techniques have been standardized, mortality rates have improved (O’Banion et al., 2022), quality of life has been made better (Ovadia & Askari, 2015), post-op therapeutics have increased in efficacy (Schug & Gillespie, 2011), and modern prosthetic development has catalysed a major restoration in previously-lost functions.
However, even since modern surgical techniques have been established, prioritization has remained largely on wound healing and creation of a viable prosthetic socket attachment region. As a result, muscle pairs and neuromuscular connections are often severed, with muscle tissue instead forming a padded region around the bone (Herr & Carty, 2021). Neurological connections and systems are often severed and given less attention, causing nerve endings to simply be left in free space. Although traditional amputations do attempt to reduce such consequences, the disruption of neural connections can often cause painful repercussions. Nearly 25% of amputees experience Neuromas (O’Reilly et al., 2016), causing extreme pain especially if occurring in areas where there is high external pressure (eg. a prosthetic socket). Almost 70% of amputees experience forms of phantom/residual limb pain (Hobusch et al., 2020), and many simply reject their prosthetic after trying it. Moreover, severance of musculature and neuronal connections at the site of amputations would make it extremely difficult for the brain to experience sensation (including proprioception) in that area. Researchers have now attempted to devise methods to re-establish these lost sensations, with a particular technique showing promise: the Agonist-Antagonist Myoneural Interface (AMI).
What is the AMI:
Invented in 2014 at the Massachusetts Institute of Technology, AMI takes advantage of the naturally occurring agonist-antagonist muscle pairs within the body at the site of amputation (Herr & Carty, 2021). Following the surgical removal of a limb, instead of leaving muscles to simply heal on their own, the AMI involves mechanically reconnecting otherwise severed muscle pairs. Since nerves and neuromuscular architecture can be conserved through creating an AMI unit, the residual limb’s proprioceptive abilities remain established (Herr & Carty, 2021). Especially since a muscle system’s fundamental functionality of contracting an agonist and stretching an antagonist continues to function, the brain is able to obtain kinesthetic-like feedback for signals that would otherwise be sent to a ‘phantom’ limb (Herr & Carty, 2021).
Through this method, even patients with a passive prosthetic limb have reported greater control and feel as if the limb is not purely an ‘external attachment’ (MIT Media Lab & Herr, 2017). From a neurophysiological standpoint, functional activation of brain regions associated with proprioception did not significantly differ between those with AMI amputations and non-amputee controls (Srinivasan et al., 2020). Compared to a marked decrease in activation of this region for those with non-AMI amputations, this difference exhibits a high potential for AMI to conserve traditionally lost sensorimotor abilities within amputees. Further, observations of an AMI patient have shown greater intentional control of their prosthetic’s movement compared to those with traditional amputation. In addition, during tasks such as navigating stairs, all natural movement patterns – that would otherwise not be possible with passive prosthetics – can also be observed as an involuntary, reflexive behavior in an AMI patient (Clites et al., 2018).
Moreover, in certain cases, when impulses are sent from the brain, sensors placed around the AMI units can detect changes in length/contraction of these reconstructed muscles and transmit such information into robotic prosthetics for translation into desired movements. According to the creators of AMI, each myoneural interface can allow for one axis of movement within a prosthetic, as each AMI is responsible for a singular muscle system (Herr & Carty, 2021). Moreover, through this method, the AMI is also able to facilitate reverse communication, i.e the communication of signals from the neuroprosthetic back to the user (Herr & Carty, 2021). With a prosthetic providing feedback on its functioning through electrical measurements (eg. impedance, etc), muscles within the AMI can be electrically stimulated to artificially replicate proprioceptive signals that can then be sent to the brain (Herr & Carty, 2021).
When created, the AMI framework could only make use of muscle systems that were pre-planned to be leveraged by AMIs post-amputation. However, more recent developments of AMI through integrating techniques such as regenerative grafting has enabled even those with prior amputations to have functionality restored with the AMI framework (Srinivasan et al., 2019).
Looking Forward:
Although originally developed primarily for lower-limb amputations, AMI’s principles can be adapted even further for upper extremity applications, where dexterous control and sensory feedback may be more critical. Where AMI has helped resolve amputative proprioceptive loss, combining it with sensory reinnervation techniques could allow for potential restoration of tactile feedback alongside what AMI can currently achieve. Furthermore, for higher-level amputations, researchers can even look towards exploring hierarchical arrangements of multiple AMI pairs to control different prosthetic joints simultaneously.
Conclusion:
The Agonist-Antagonist Myoneural Interface represents a significant advancement in amputation surgery by preserving the body's natural proprioceptive mechanisms. By maintaining the dynamic relationship between opposing muscle groups, AMI restores critical neural feedback pathways that inform the brain about limb position and movement.
Patients have shown improvements across the board following the conservation of the proprioception, especially with greater phantom limb awareness, more intuitive control of prosthetic devices, and greater natural movement patterns
Very few modern techniques are capable of such sensory integration with one’s neuroprosthetic. It is very exciting to witness such innovations allow those with artificial limbs to not feel isolated from their prosthesis. The AMI approach represents more than just an incremental improvement in amputation surgery – it signals a paradigm shift in how we can approach interfaces between human neuromuscular systems and prosthetic technology.
About the Author
Hiteyjit Singh Gujral (‘27) is a sophomore at Harvard College concentrating in neuroscience.
References
- BBC Bitesize. (2019). Muscular System - Edexcel - Revision 4 - GCSE Physical Education - BBC Bitesize. BBC Bitesize. https://www.bbc.co.uk/bitesize/guides/zpkr82p/revision/4
- Blumer, R., Carrero-Rojas, G., de la Cruz, R. R., Streicher, J., & Pastor, A. M. (2024, January 1). Extraocular Muscles: Proprioception and Proprioceptors. ScienceDirect; Elsevier. https://www.sciencedirect.com/science/article/abs/pii/B9780443138201000281
- Clites, T. R., Carty, M. J., Ullauri, J. B., Carney, M. E., Mooney, L. M., Duval, J.-F., Srinivasan, S. S., & Herr, Hugh. M. (2018). Proprioception from a neurally controlled lower-extremity prosthesis. Science Translational Medicine, 10(443), eaap8373. https://doi.org/10.1126/scitranslmed.aap8373
- Herr, H., & Carty, M. J. (2021). The Agonist-antagonist Myoneural Interface. Techniques in Orthopaedics, Publish Ahead of Print. https://doi.org/10.1097/bto.0000000000000552
- Hobusch, G. M., Döring, K., Brånemark, R., & Windhager, R. (2020). Advanced techniques in amputation surgery and prosthetic technology in the lower extremity. EFORT Open Reviews, 5(10), 724–741. https://doi.org/10.1302/2058-5241.5.190070
- Kröger, S., & Watkins, B. (2021). Muscle spindle function in healthy and diseased muscle. Skeletal Muscle, 11(1). https://doi.org/10.1186/s13395-020-00258-x
- Macefield, V. G. (2024). Proprioception: Role of Joint Receptors. Encyclopedia of Neuroscience, 3315–3316. https://doi.org/10.1007/978-3-540-29678-2_4826
- MIT Media Lab, & Herr, H. (n.d.). AMI: Frequently Asked Questions. MIT Media Lab. https://www.media.mit.edu/projects/agonist-antagonist-myoneural-interface-ami/frequently-asked-questions/
- MIT Media Lab, & Herr, H. (2017). Project Overview ‹ Agonist-antagonist Myoneural Interface (AMI) – MIT Media Lab. MIT Media Lab. https://www.media.mit.edu/projects/agonist-antagonist-myoneural-interface-ami/overview/
- Morimoto, T., & Takada, K. (1993). The sense of touch in the control of ingestion. Elsevier EBooks, 79–97. https://doi.org/10.1016/b978-0-08-041988-6.50010-9
- O’Banion, L. A., Qumsiyeh, Y., Matheny, H., Siada, S. S., Yan, Y., Hiramoto, J. S., Rome, C., Dirks, R. C., & Prentice, A. (2022). Lower extremity amputation protocol: a pilot enhanced recovery pathway for vascular amputees. Journal of Vascular Surgery Cases, Innovations and Techniques, 8(4), 740–747. https://doi.org/10.1016/j.jvscit.2022.08.003
- O’Reilly, M. A. R., O’Reilly, P. M. R., Sheahan, J. N., Sullivan, J., O’Reilly, H. M., & O’Reilly, M. J. (2016). Neuromas as the cause of pain in the residual limbs of amputees. An ultrasound study. Clinical Radiology, 71(10), 1068.e1–1068.e6. https://doi.org/10.1016/j.crad.2016.05.022
- Ovadia, S., & Askari, M. (2015). Upper Extremity Amputations and Prosthetics. Seminars in Plastic Surgery, 29(01), 055–061. https://doi.org/10.1055/s-0035-1544171
- Proske, U., & Gandevia, S. C. (2012). The Proprioceptive Senses: Their Roles in Signaling Body Shape, Body Position and Movement, and Muscle Force. Physiological Reviews, 92(4), 1651–1697. https://doi.org/10.1152/physrev.00048.2011
- Schug, S. A., & Gillespie, G. (2011). Post-amputation Pain (R. Fitridge & M. Thompson, Eds.). PubMed; University of Adelaide Press. https://www.ncbi.nlm.nih.gov/books/NBK534280/
- Shores, J. (2019). Amputation. Johns Hopkins Medicine Health Library. https://www.hopkinsmedicine.org/health/treatment-tests-and-therapies/amputation
- Srinivasan, S. S., Diaz, M., Carty, M., & Herr, H. M. (2019). Towards functional restoration for persons with limb amputation: A dual-stage implementation of regenerative agonist-antagonist myoneural interfaces. Scientific Reports, 9(1). https://doi.org/10.1038/s41598-018-38096-z
- Srinivasan, S. S., Tuckute, G., Zou, J., Gutierrez-Arango, S., Song, H., Barry, R. L., & Herr, H. M. (2020). Agonist-antagonist myoneural interface amputation preserves proprioceptive sensorimotor neurophysiology in lower limbs. Science Translational Medicine, 12(573), eabc5926. https://doi.org/10.1126/scitranslmed.abc5926
- The University of Sydney. (2022, September 8). Stone age surgery: earliest evidence of amputation found. The University of Sydney. https://www.sydney.edu.au/news-opinion/news/2022/09/08/stone-age-surgery-earliest-evidence-of-amputation-found-archaeology.html