Receptor Potentials and Neural Coding

A receptor potential (also called a generator potential when it directly triggers spiking in the same cell) is a local, graded change in membrane potential produced by a sensory stimulus acting on a receptor cell or specialized sensory ending. It arises from the opening or closing of ion channels — mechanically gated channels in touch receptors, cyclic-nucleotide-gated channels in photoreceptors, ligand-gated channels in taste and olfactory receptors — and it behaves like an transduced analog signal: its amplitude and time course scale continuously with stimulus strength, it decays with distance (it is not regenerated), and it can summate temporally and spatially. This is fundamentally different from an action potential, which is regenerative, all-or-none, and propagates without decrement. The nervous system's problem is that most receptor potentials occur too far from the central nervous system, or are too small and too slow, to be read out directly by the brain. The solution is rate coding: the graded receptor (or generator) potential depolarizes the membrane at a specialized spike-initiating zone, and if it exceeds threshold, it triggers a train of action potentials whose instantaneous frequency scales with the receptor potential's amplitude. Stronger stimuli produce larger receptor potentials, which produce higher firing rates, which the central nervous system reads as greater intensity. This is why sensory coding is often described as having two stages: an analog stage (transduction) and a digital, frequency-modulated stage (spike encoding) — conceptually similar to how an analog voltage can be converted to a pulse-frequency-modulated digital signal in engineering, though the biological implementation is far messier and adapts over time. Two further principles complete the picture. Labeled lines refer to the fact that each sensory axon's central target is fixed by anatomy and development, so activity in that line is always interpreted the same way regardless of how it was generated — a mechanically-, electrically-, or artifactually-triggered spike in an optic nerve fiber is always perceived as light. Population coding refers to the fact that many stimulus dimensions (color, sound frequency, spatial location, movement direction) are not signaled by single "grandmother" receptors tuned narrowly to one value, but by the relative activity across a population of broadly-tuned neurons, whose combined pattern the brain decodes.

Step 1 — Stimulus energy opens or closes ion channels. In a Pacinian corpuscle, mechanical indentation deforms the lamellated capsule and stretches the underlying nerve terminal membrane, opening mechanically-gated cation channels (Piezo2 channels are now known to be central to this process in many mechanoreceptors). In a rod photoreceptor, light closes cyclic-GMP-gated channels via the phototransduction cascade, producing a hyperpolarizing receptor potential — an instructive exception to the general rule that stronger stimuli depolarize.

Step 2 — Ion flux produces a graded receptor potential. The current produced by these channels flows through the receptor membrane's resistance, producing a local voltage change (Ohm's law: V = IR) whose amplitude depends on stimulus intensity, duration, and the receptor's transduction efficiency. Over a useful working range, this relationship is monotonic and often compressive (a large range of stimulus intensities maps onto a smaller range of receptor-potential amplitudes) — this compression is what allows receptors to encode many decades of stimulus intensity within a membrane voltage range of only tens of millivolts.

Step 3 — The receptor potential spreads passively (electrotonically) to a spike-initiating zone. Because a receptor potential is not regenerative, it decays with distance according to the cell's length constant. In many mechanoreceptors and in secondary sensory cells (like hair cells or taste cells), the graded potential must first trigger neurotransmitter release onto a separate afferent neuron, adding a synaptic step before spike generation; in primary sensory neurons (like the Pacinian corpuscle's own afferent, or olfactory receptor neurons), the receptor potential directly depolarizes the first node of Ranvier or spike-initiation zone.

Step 4 — Threshold crossing triggers a train of action potentials. If the depolarization at the spike-initiating zone exceeds threshold, voltage-gated sodium and potassium channels generate action potentials exactly as in any excitable membrane. Critically, the amplitude of the receptor potential is recoded as the frequency of the spike train, not as spike amplitude (spikes are all-or-none) and not usually as spike shape. Larger sustained depolarizations bring the membrane back to threshold faster after each spike's refractory period, so instantaneous firing rate rises monotonically with receptor-potential amplitude, at least until channel inactivation and saturation intervene at high intensities.

Step 5 — The stimulus–response function typically obeys a compressive law. Across many receptor types, plots of firing rate versus stimulus intensity are well fit either by a power function (Stevens' power law, R ∝ Sⁿ, with exponent n often between about 0.3 and 1.7 depending on modality) or, over a more restricted range, by a logarithmic function (consistent with the older Weber–Fechner law, which states that the just-noticeable difference in stimulus intensity is a constant fraction of the background intensity, ΔI/I ≈ constant, called the Weber fraction). Both laws describe the same underlying biological fact: receptors have limited dynamic range and must compress a stimulus range spanning many orders of magnitude (for light, roughly 10 log units of intensity; for sound pressure, about 12 orders of magnitude) into a firing-rate range of only about 0–500 spikes/second.

Step 6 — Population and labeled-line coding recover stimulus identity, location, and quality. Because a single afferent's spike train only signals "how much" activity is occurring in that line, the brain must combine information across many afferents to know "what" and "where." Labeled lines fix modality and rough location by anatomical wiring: a spike arriving via the optic nerve is always seen as light because of where that line terminates in the thalamus and cortex, not because of anything intrinsic to the spike itself. Population codes then refine this: for example, the perceived color of light depends on the relative activation of three cone types (each individually only signaling luminance along its own broad, overlapping spectral tuning curve), and perceived sound frequency depends partly on which set of auditory-nerve fibers (tuned to overlapping frequency bands by their position along the cochlea) is most active, in addition to phase-locking codes at low frequencies.

The single most important founding experiment in this field is Edgar Adrian's work in the late 1920s recording from single sensory nerve fibers — the first time anyone had isolated the electrical activity of one axon at a time. Using a capillary electrometer and later valve amplifiers connected to a frog muscle-stretch receptor and to cutaneous touch receptors, Adrian showed two things that became foundational laws of sensory physiology. First, individual action potentials in a given fiber were essentially identical in size and shape regardless of stimulus strength — confirming the all-or-none law at the single-fiber level. Second, and more importantly, as he increased the stretch on a muscle or the pressure on skin, the frequency of firing in the same fiber increased smoothly and monotonically, while spike amplitude stayed constant. Adrian explicitly stated that intensity is signaled by "the frequency of the impulses and not by their size," a conclusion he reached by directly counting spikes per second from oscillographic records under graded, quantified mechanical loads. This work — for which Adrian shared the 1932 Nobel Prize in Physiology or Medicine with Charles Sherrington — established rate coding as an empirical fact rather than a hypothesis.

Stephen Kuffler, working in the 1950s on the frog and cat muscle spindle stretch receptor (a favorite because its receptor potential and spike-initiating zone can be studied with relative anatomical simplicity), extended this by directly recording the generator potential itself, separate from the spikes it produced. Using local anesthetics or focal cooling to block spike generation at the first node of Ranvier while leaving the sensory ending intact, Kuffler and colleagues showed that the underlying generator potential was indeed a graded, non-propagating depolarization whose amplitude scaled with the degree of muscle stretch, and that this graded potential decremented with distance exactly as predicted for a passive electrotonic signal. When spiking was not blocked, the firing rate downstream tracked the size of this generator potential in an orderly, monotonic fashion. This was the direct demonstration that a graded, analog signal precedes and causes the rate-coded spike train — closing the causal loop that Adrian's data had only implied. A third pillar of evidence concerns labeled lines, and here the classic demonstration is almost embarrassingly simple: mechanical, electrical, or even chemical stimulation of a sensory pathway produces a sensation appropriate to that pathway's usual modality, never the modality of the actual stimulus. Pressing on the eyeball mechanically stimulates retinal photoreceptors and produces a visual sensation (a phosphene) rather than a feeling of pressure, because the optic nerve is "labeled" for vision by its central connections. This principle was formalized physiologically by Vernon Mountcastle's studies of the somatosensory system in the 1950s–60s, in which he showed that neurons at every level of the dorsal column–medial lemniscal pathway, from peripheral afferent to cortical neuron, retained modality-specific and place-specific response properties, and that electrically stimulating a single, identified afferent fiber (or its central target) reliably evoked a sensation matching that fiber's natural modality and receptive field location — evidence later reinforced by microstimulation studies in awake primates (Vallbo, Johansson, and colleagues) showing that stimulating a single identified tactile afferent could evoke a distinct, localized tactile percept in a human subject. Together, Adrian's rate-coding experiments, Kuffler's generator-potential recordings, and the labeled-line microstimulation studies form the empirical backbone of the modern theory of sensory coding.

Example 1 — Applying Stevens' power law to a Pacinian corpuscle. Suppose a Pacinian corpuscle's firing rate R (spikes/s) obeys Stevens' power law R = k·Sⁿ, where S is the indentation depth in micrometers, k = 8, and the measured exponent for this receptor is n = 0.6 (mechanoreceptor exponents are typically sub-linear, reflecting compression). At an indentation of S = 25 μm, predict the firing rate, and then find how much you must increase S to double the firing rate.

• R₁ = 8 × 25^0.6. First, 25^0.6 = e^(0.6·ln25) = e^(0.6×3.219) = e^1.931 ≈ 6.90.
• R₁ ≈ 8 × 6.90 ≈ 55.2 spikes/s.
• To double firing rate: R₂ = 110.4 = 8·S₂^0.6 → S₂^0.6 = 13.8 → S₂ = 13.8^(1/0.6) = 13.8^1.667.
• ln(13.8) ≈ 2.625; ×1.667 ≈ 4.376; e^4.376 ≈ 79.4.
• So S₂ ≈ 79.4 μm — roughly 3.18× the original indentation is needed to double the firing rate, illustrating the compressive, sub-linear nature of the code: because n < 1, proportionally large increases in stimulus are needed to produce equal proportional increases in response, a direct electrophysiological analogue of the psychophysical Weber–Fechner compression.

This matters clinically: a receptor with a compressive power-law exponent allows a huge dynamic range (light touch through firm pressure) to be represented within a firing-rate range the nervous system can actually transmit (typically under a few hundred spikes/second), at the cost of reduced absolute sensitivity to changes at high intensities — exactly the trade-off predicted by information theory for a channel with a limited output range.

Example 2 — Diagnostic reasoning from a labeled-line deficit. A patient reports that gentle stroking of a small patch of skin on the forearm produces a burning, painful sensation rather than a light-touch sensation, even though the stimulus is objectively innocuous (a phenomenon called allodynia). Using the labeled-line/rate-coding framework, reason through two possible mechanisms and how you would distinguish them electrophysiologically.

• Hypothesis A — peripheral sensitization: nociceptor free nerve endings in that patch have become abnormally excitable (e.g., after tissue injury, inflammatory mediators such as prostaglandins and bradykinin lower their transduction threshold), so an innocuous mechanical stimulus now generates a suprathreshold generator potential in the nociceptive labeled line itself. Prediction: microneurography (recording directly from single cutaneous nerve fascicles) should show nociceptor afferents firing action potentials in response to light touch that normally would not activate them at all.
• Hypothesis B — central rewiring/crosstalk: the low-threshold mechanoreceptor afferents (normally signaling touch) fire completely normally, but their central targets in the dorsal horn have become abnormally connected to (or disinhibited onto) pain-signaling projection neurons, so a normal touch-coded spike train is now centrally misread as pain — a violation of the labeled-line assumption at the level of central wiring rather than the periphery. Prediction: microneurography of the touch afferent shows a normal low firing rate appropriate to light touch, but the central pain pathway is nonetheless activated (detectable via, e.g., functional imaging or reflex withdrawal).

Distinguishing these requires exactly the tool Adrian and Kuffler pioneered — direct electrophysiological recording from identified peripheral fibers — to determine whether the "mislabeling" occurs in the receptor's own transduction/generator-potential stage or downstream in the labeled line's central connections; this distinction is directly relevant to choosing peripheral versus centrally-acting analgesic strategies in chronic pain syndromes.

:::warning[The single most dangerous confusion is treating action-potential amplitude as a variable that carries information.] Spikes are essentially digital — fixed in amplitude and shape within a given axon — precisely so that they can propagate long distances without distortion or decay. All stimulus-intensity information that survives the receptor-to-brain journey must be repackaged into a variable that survives axonal propagation: firing rate, spike timing/pattern, and the number and identity of active fibers (population code). Forgetting this leads students to imagine graded "big spikes" for strong stimuli, which is not how the nervous system works.

Edgar Douglas Adrian trained originally in physiology and clinical medicine at Cambridge and at St Bartholomew's Hospital in London, but his defining scientific contribution came from an engineering problem: how to record electrical activity from a single, isolated sensory nerve fiber, a feat that required exquisitely fine dissection of nerve-muscle preparations and amplification technology sensitive enough to detect microvolt-scale signals — technology that barely existed when he began. Working initially with Keith Lucas and later independently at Cambridge through the 1920s, Adrian dissected single fibers from frog and cat sensory nerves (notably muscle spindle afferents and cutaneous touch receptors) down to the point where only one active axon remained in the recording, then applied precisely graded mechanical stimuli while photographing the resulting oscillograph traces. His 1926–1928 papers, culminating in the influential monograph The Basis of Sensation (1928), laid out the frequency code for intensity with a rigor that turned a plausible hypothesis into an established physiological law. He shared the 1932 Nobel Prize in Physiology or Medicine with Charles Sherrington for discoveries concerning the function of neurons, and went on to lead Cambridge's physiology department and serve as President of the Royal Society, mentoring a generation of electrophysiologists who extended single-unit recording to nearly every sensory system.

Cochlear implants are perhaps the most direct clinical application of receptor-potential and rate-coding theory: because hair-cell transduction is destroyed in sensorineural deafness, the implant bypasses the receptor potential stage entirely and directly injects patterned electrical current into auditory-nerve labeled lines, exploiting the fact that the brain, not the ear, ultimately interprets which line is active as a given pitch (tonotopic place coding) and uses stimulation rate to convey additional temporal/intensity cues. Retinal prostheses for inherited photoreceptor degenerations (such as retinitis pigmentosa) work on the identical logic, electrically driving surviving retinal ganglion cells so that their labeled-line projection to the visual thalamus is interpreted as light regardless of the true (electrical) origin of the signal. Understanding compressive stimulus–response functions underlies clinical audiometry and perimetry (visual field testing), where thresholds and just-noticeable differences are quantified against known Weber fractions to detect early sensory loss before a patient consciously notices a deficit. In pain medicine, distinguishing peripheral sensitization (altered receptor-potential thresholds in nociceptors) from central sensitization (altered decoding of normal labeled lines) directly guides whether a topical/peripheral agent or a centrally-acting drug is more likely to help a given chronic pain patient. In agricultural and veterinary contexts, understanding rate-coded thermoreceptor and nociceptor responses informs humane handling and analgesia protocols in livestock. And in basic neuroscience and brain–machine interface research, population-coding models derived from exactly this framework (broadly tuned units whose joint activity is decoded, rather than single dedicated "detector" neurons) are what allow engineers to decode limb movement intent from cortical spike trains for neuroprosthetic limb control — a population-vector decoding strategy directly descended from the tuning-curve logic first quantified in sensory systems.

Ask the tutor: How does adaptation (a receptor's firing rate declining during a sustained, unchanging stimulus) interact with rate coding — does it mean intensity information is lost over time?