Welcome to chapter seven, nerves, sensation and accessible design. This chapter gives you enough neurophysiology to understand why moving, feeling, seeing, hearing and interpreting are separate capabilities supported by separate machinery, and enough design method to act on it. It is also the chapter closest to classical biomedical engineering, because nerves and muscles are electrical, and electrical signals are measurable. If you ever touch an EMG, ECG or EEG instrument, this is its foundation.
Can the user operate this control sounds like one question. It is at least four, and they fail independently. One person may have plenty of strength and not enough pinch precision. Another may operate the mechanism perfectly and have no way to see whether it latched. A third may see it and understand it and be unable to reach it. A fourth may do all of that and be unable to feel that the strap underneath is cutting in. Each of those is a different failure with a different fix, and this chapter teaches you to separate them.
The central nervous system is the brain and spinal cord; the peripheral system is the nerves connecting them to the body. Sensory, or afferent, pathways carry information toward the centre. Motor, or efferent, pathways carry commands away. Learn those two words, because clinical language uses them constantly and because the distinction is exactly the one your design review needs: input to the person versus output from the person. The autonomic division is largely outside conscious control. It is why heart rate rises with anxiety, and why "the patient was agitated" is a physiological statement as well as a behavioural one.
A neuron's dendrites receive inputs, the cell body decides whether their sum reaches threshold, and the axon conducts the resulting signal, sometimes over a metre. Myelin insulates the axon in segments, and the signal regenerates only at the gaps between them, so it effectively jumps. That is called saltatory conduction, and it is much faster and cheaper than continuous conduction. Speeds range from about half a metre per second in small bare fibres to around a hundred and twenty metres per second in large myelinated ones. Conditions that strip myelin slow or block conduction without destroying the axon, which is why their effects can be profound and variable.
At rest a neuron holds about minus seventy millivolts inside, a battery kept charged by the sodium potassium pump using ATP. Cut the energy supply and the gradients decay, and disturb potassium levels in the blood and nerves and the heart are affected immediately. When a stimulus reaches threshold, sodium channels open, sodium rushes in, and that opens more channels: positive feedback, and a rapid upstroke. Two properties matter for engineering. The action potential is all or nothing; its amplitude is effectively digital. And a stronger stimulus does not make a bigger spike. It makes more spikes per second, and recruits more fibres. That is why nerve stimulators control pulse rate and pulse width rather than simply voltage.
At a chemical synapse the arriving signal lets calcium in, which releases neurotransmitter into a narrow gap. The transmitter binds receptors on the next cell, which may be excited or inhibited, and that cell fires only if the total reaches threshold. The neuromuscular junction is the special synapse between a motor nerve and a muscle fibre, and its transmitter is acetylcholine. This is where chapter six's muscle physiology connects: nerve signal, then acetylcholine, then calcium released inside the fibre, then contraction. Many drugs and toxins act at this junction, which is why it appears constantly in anaesthesia and pharmacology.
Touch is not one sense. Merkel complexes report sustained touch and edges. Meissner corpuscles report light touch and slip. Pacinian corpuscles report vibration and deep pressure. Ruffini endings report stretch. Free nerve endings report warmth, cold and potentially damaging stimuli, and muscle spindles and tendon organs report muscle length and tension. The general lesson is adaptation. Rapidly adapting receptors stop reporting a constant stimulus, which is why you stop noticing your clothes within seconds. That is efficient, and it is a design hazard: a pressure you can feel at minute one you may not notice at minute thirty.
Receptor density varies enormously across the body, and you can measure it as two-point discrimination: the smallest separation at which two touches feel like two. On a fingertip it is about two to four millimetres. On the palm about ten. On the forearm thirty to forty, and on the back forty to fifty or more. These are orientation values that vary with method and person. The design consequence is direct: tactile markers must be separated by considerably more than the threshold at the site where they will be felt. Two dots three millimetres apart on a handle gripped with the palm are one dot.
Proprioception is the sense of body position and movement, built from muscle spindles, tendon organs and joint receptors. It is the sense people forget they have, and the one whose loss is most disorienting. Without it you cannot find your own hand in the dark, cannot grade your grip without watching, and cannot hold a posture without visual attention. Tasks that look trivial consume all of someone's attention. So an interface that can be operated without looking, with distinguishable shapes and clear mechanical detents, places far less load on that person, and on everyone whose eyes are needed elsewhere.
Nociception is the detection of potentially damaging stimuli; pain is the experience, and the two dissociate in both directions. Absence of a pain report is not evidence of absence of harm, and the reasons stack up. Sensation may be reduced in that area, through neuropathy, spinal injury, surgery or local anaesthetic. Rapidly adapting receptors stop reporting sustained pressure. Damage from sustained loading starts deep, where there are fewer nerve endings. The person may be unable to communicate, or reluctant to complain to someone who has gone to trouble for them. And attention elsewhere suppresses awareness. So a design must never rely on the user's discomfort as its safety mechanism. If protection depends on someone noticing, and they may not notice, the protection does not exist.
A reflex is a rapid, involuntary, stereotyped response. The withdrawal reflex runs from a pain receptor in the fingertip into the spinal cord and straight back out to the flexor muscles, without waiting for the brain. The hand has already withdrawn by the time it hurts. That cuts both ways for a designer. A startle or withdrawal can make someone drop, jerk or pull away, so a mechanism should tolerate a sudden release. And reflexes protect only against hazards that trigger them. Sustained pressure triggers no withdrawal reflex at all.
Now the engineering bridge. If you move toward instrumentation, this next table is where every biomedical instrumentation course begins.
Here are the signals. An ECG from the heart is about half a millivolt to four millivolts at the skin. Surface EMG from muscle is similar. EEG from the brain is only ten to a hundred microvolts. A surface nerve action potential is one to twenty microvolts. Look at those amplitudes against mains interference, which is volts. The signal is five to six orders of magnitude below the ambient noise, and the wire is a gel electrode on skin whose impedance drifts with moisture and movement. Everything in the field follows from that: high input impedance instrumentation amplifiers, high common mode rejection, careful shielding and grounding, motion artefact rejection, and above all patient electrical safety.
This is an absolute boundary, not a cautious preference. Never connect anything you have built to a person. Patient-connected electrical equipment is governed by the standard IEC 60601-1 and its particular parts, which set leakage current limits, isolation, and how applied parts are classified. Mains-powered bench equipment, development boards and anything with a USB connection to a laptop are not isolated to those limits. If a project needs biosignals, use clinically approved instruments operated by someone qualified, or work with recorded or synthetic data.
Two ideas from psychophysics. The just-noticeable difference, the smallest change someone reliably detects, is roughly proportional to the baseline. That is Weber's law: the louder the room, the bigger the change in sound you need to notice it. So a feedback cue sized for a quiet workshop may be undetectable in a busy clinical area. And reaction time: a simple visual reaction is around two hundred milliseconds, anything involving a choice is much longer, and fatigue, age, medication and divided attention all lengthen it. If your interface has a timeout, that timeout has real human limits behind it, and your framework's default was not chosen with any of them in mind.
This is the method the chapter exists to deliver. Any interaction with any object decomposes into at least four independent demands. Perception: can they detect it, see the label, hear the click, feel the detent? Cognition: do they know what it means, what the symbol is, what the sequence is? Action: can they reach it, grip it, exert the force, achieve the precision, sustain it? And feedback: do they know what happened, whether it registered, what state it is in now? Each stage can fail on its own, and each failure is a separate, often cheap, intervention.
Work an example. A control requires someone to see a small symbol, pinch a latch, hold it open and slide a part into place. Can use the control conceals at least eight demands. Seeing the symbol depends on acuity, contrast, lighting and glare. Knowing what it means depends on convention and memory. Reaching depends on range of motion. Pinch force and pinch precision are different capabilities. Holding it open is an endurance demand. Sliding the part may need two hands. And knowing it is fully engaged depends on whether there is any indication at all. The same decomposition works for software, and it is not a coincidence that the web accessibility guidelines are organised as perceivable, operable, understandable and robust.
A diagnostic label is not a specification. Two people with the same diagnosis may differ in strength, range, sensation, endurance, tremor, vision, cognition, priorities, living situation and what they actually want to do. Designing from a diagnosis produces a device for a category, and categories do not use devices. So ask instead: what are you trying to do, how do you do it now, what specifically becomes difficult and when, and what have you already tried? A therapist helps interpret movement, sensation and function. The person is the expert on their own experience. Both inputs are needed, and they are not interchangeable.
Every single-channel cue fails for someone. Colour alone fails for colour vision deficiency, which affects around eight percent of men of northern European descent, and for poor lighting and monochrome printing. Sound alone fails with hearing loss, in a noisy ward, or when the operator is facing away. Small text fails for low vision and with wet or gloved hands. The principle is redundant coding: carry the same information on more than one channel, so no single impairment removes it. And the strongest control of all is one where the wrong thing simply cannot be done, such as an asymmetric connector that only fits one way. That sits at the top of the risk control hierarchy for good reason.
Neurology and rehabilitation have the densest concentration of assistive technology in medicine. Clinical devices include EEG and EMG systems, nerve stimulators, cochlear implants and hearing aids, and augmentative and alternative communication devices, where the access method is the core design problem. Switch and alternative access hardware is highly individual, and it is where maker and occupational therapist collaboration is most productive. Software includes screen readers, the free NVDA among them, and accessibility test tools you can run today. Training includes lumbar puncture trainers, nerve block phantoms, and simple sensory testing kits that are cheap and makeable.
Accessibility is the highest-leverage contribution in this book, in software and in hardware. On the software side it needs no workshop, no materials, no infection control review and no patient contact. The tools are free, the standards are public, and most clinical and educational software is measurably bad at it. Run an automated checker over a site, then navigate it with a screen reader. On the hardware side the work is alternative access: a switch interface that turns a large, low-force button into a keypress; a mounting arm that holds it exactly where one person can reach it; a robust, wipeable enclosure; a strap that keeps it in place. Each draws on electronics, machining and textiles, and each is made with a therapist and the user.
Your elbow model has two cord ends, and a learner cannot tell which one bends it. The first attempt, red and green, is better than nothing and fails for colour vision deficiency, poor light and a photocopy. The second attempt uses redundant coding: text labels saying pull to bend and support, do not pull; different tab shapes you can tell apart by touch; a curved arrow showing the motion; the active end always on the same side; and a line drawn from the pull point to the moving segment so the cause is visible. Keep the colour, as a redundant cue. Then test it on someone who has never seen it, without hints. If they cannot explain it, your labelling failed, not their reading.
For the workshop: decompose pressing a touchscreen button into at least eight demands across perception, cognition, action and feedback, then do the same for a physical control. Review a control panel for five barriers, each tied to a stage, and redesign it, noting what each change costs. Apply redundant coding to your model and test it on someone unfamiliar with it. With a willing adult and a bent paperclip, find the two-point threshold on a fingertip and a forearm; it is the ratio that matters. And remember: a one-handed or gloved trial is a way to find problems, not a way to understand disability.
The limits. You know sensation, strength and cognition are distinct capabilities; measuring them is clinical work. The biosignal table is orientation for a future instrumentation course, not a licence; the standard exists because the failure mode is electrocution. A one-handed trial reveals problems but does not reproduce lived experience. Accessibility conformance is testable, but whether something is usable by a particular person is established with that person. And tremor, spasticity and neuropathy appear here only as reasons for variation. Next, the head and neck: eye, ear, nose, throat and mouth.