By Guillermo Herrera-Arcos and Tony Shu | K. Lisa Yang Center for Bionics

A control system orchestrating hundreds of actuators to cook, drive a car, and make music with every instrument known to mankind. On-board sensors to detect every crinkle and puff that can possibly be gleaned from the environment. Self-repairing and self-replicating, the human body is the most incredible machine you will ever operate.

The field of bionic prostheses is unique in that its goals are preordained: the body’s form is celestial, its function, our guiding light. As researchers, we seek to enable control and sensation of artificial limbs equal to their biological counterparts. And though success means duplicating the exquisiteness of a musical composer like Liszt, the imagination readily leaps to encompass supraphysical notions such as the virtuosic performance of 12-finger bionic compositions, or swimming at remarkable speeds via neural control of artificial fins.

Even if it is evident that high degrees of bionic integration require technological advancements in both mechatronic and biologic realms, pursuit of the latter has been relatively neglected by the field at large. The development of powerful actuators, batteries, chips, and algorithms has far outpaced our ability to intuitively interface with them through neural connection. Should we tap into the central brain directly? Or do peripheral nerves provide a more accessible port? What about muscles, skin, and all the critical mechanoreceptors within our bodies that provide us rich proprioception and tactile sensation? While it is obvious that bionic limbs should not be controlled via mouse and keyboard, there remains a debate in the field of bionics as to the exact structure of a superior interface.

In a recent article published in Nature Reviews Bioengineering, we delve into mechanoneural interfaces, a new paradigm for bionic integration comprising surgical combinations of soft tissue architectures involving muscles, nerves, and skin with synthetic components, such as implantable sensors and stimulators. We propose that mechanoneural interfaces increase physiological signal selectivity and bandwidth between the brain and peripheral nervous system by leveraging the remarkable capabilities of soft tissues, including the regenerative capacity of peripheral nerves and the ability of muscles to act as biophysical amplifiers of motor signals.

The basic operating principle of a mechanoneural interface is its ability to convert neural signals into mechanical force (e.g., muscle contraction) or convert mechanical force into neural signals (e.g., skin deformation). One example of a mechanoneural interface is targeted sensory reinnervation, wherein a peripheral sensory nerve left over after amputation is introduced to a denervated patch of skin. After reinnervation, simply touching the patch of skin may elicit cutaneous sensations that are perceived as originating from the phantom limb. Another example is the agonist-antagonist myoneural interface, wherein muscles that were naturally paired in the intact limb are surgically relinked in the amputated limb to restore proprioception of the phantom limb to the user. Signals generated by these paired muscles are then used in harmony to control the joint of a bionic prosthesis. Both of these mechanoneural interfaces and many others have already been implemented in persons with amputation to yield more intuitive prosthetic integration with lower and upper extremity prostheses.

Looking ahead, attaining fluency in the body’s nuanced notation requires developing specialized tools and techniques that further our understanding of soft tissues, their regenerative capabilities, the biological signals available, and the I/O required for their interfacing to limb prostheses. The pursuit of more sophisticated mechanoneural interfaces will ultimately enhance the phenomenon of prosthetic embodiment consisting of agency, or the awareness of the control of movement, and ownership, or the feeling that our body parts are inherently part of ourselves. Ultimately, we propose that principles learned from studying mechanoneural interfaces and their functional outcomes will yield a bionic information theory that will allow scientists and physicians to advise patients on which mechanoneural interfacing techniques maximize prosthetic embodiment. Through such advancements, limb prostheses will become an integral part of self, transcending their role as mere assistive instruments.