Skip to content

F-I curve

The firing frequency a neuron produces across applied current levels under a stated protocol and rate convention.

Version
v1 · 2026-10-17 · History
Domain-specific #
14149
Domain group
Natural Sciences
Origin domain
Neuroscience
Subdomains
Electrophysiology, Computational Neuroscience → Neuroscience
Aliases
Frequency-current curve, Firing frequency-current relationship

Core Idea

An F-I curve relates applied current, I, to the firing frequency, f, of a specified neuron or neuron model.[1][2] Its points are meaningful only with the current protocol and a rate convention: a finite measurement window for an experiment, or a stated steady-state rule for a model. The curve is a repeatable type of relation, not one universal shape shared by all neurons.

In cortical slices, Tateno, Harsch and Robinson varied injected current and measured firing from spike trains.[1] Drion, O’Leary and Marder instead applied current steps to a conductance-based model and calculated its steady firing rate.[2] Both trace current to neuronal firing frequency, while one is experimental measurement and the other a model result. Their agreement on the variables does not make their neuron types, timescales or curve shapes identical.

Structural Signature

  • Neuronal bearer. Identify the biological neuron and preparation, or the specified neuron model. A population activity response is a different carrier unless an explicit neuron-level bridge is supplied.
  • Applied-current protocol. Vary injected or simulated current over a range; state pulse duration and step history where relevant. Tateno and colleagues explicitly distinguish current-input f-I from conductance-input f-g plots.[1] Replacing current with conductance changes the relation being reported.
  • Firing-frequency convention. Convert spikes to a rate using a declared estimator. Tateno and colleagues used 600-ms pulses and, for one steady-state estimate, the last five interspike intervals; Drion and colleagues report a model steady-state rate.[1][2]
  • Current-to-rate pairing. Associate each tested input level with the rate produced under those conditions. This relation, rather than a single threshold or a plotted line's visual style, is the F-I curve.

An onset threshold, slope, saturation or firing class may describe a particular curve. Tateno and Drion exhibit different onset behaviors in experimental and modeled curves, respectively.[1][2] None of those behaviors is an extra constitutive slot.

What It Is Not

An F-I curve is not a list of spike times, a voltage trace, or one rheobase estimate. Those observations can help construct or characterize it, but they do not map firing rates across current levels. It is also not automatically an f-g curve: a conductance input changes the tested quantity. Tateno and colleagues explicitly measured both types of relationship separately.[1]

Nor is any population-level rate transfer function automatically a single-neuron F-I curve. Aggregation can change the bearer and response definition; an equivalence must be demonstrated rather than inferred from two axes with similar names. Tateno's measured cells and Drion's model show differing onset behaviors.[1][2] Thus a sigmoidal shape is not part of the name's definition.

Scope of Application

The literal scope is neuronal electrophysiology and neuron modeling when applied current and firing frequency are paired under stated conditions. Experimenters can compare cells or states; modelers can ask how conductances alter excitability. In Tateno's young rat somatosensory cortical slices, regular-spiking pyramidal cells and fast-spiking inhibitory cells yielded different threshold behavior.[1] In Drion's Connor–Stevens model, changing A-type potassium conductance changed the modeled onset class.[2]

The scope does not expand simply because another field plots output against input. A generic response curve has a thinner relation; the F-I name retains the neuronal carrier, current input and firing-rate output.

Clarity

For any reported F-I curve, ask: whose firing, which current, and which frequency convention? A current-step experiment and a model both answer these questions, but they answer them differently. Tateno's methods distinguish initial instantaneous from later firing rates.[1] An unspecified “firing rate” can therefore conceal a measurement choice.

Onset type is a finding about a curve, not the definition of the curve. In the studied cortical preparation, the regular-spiking cells approached threshold continuously and the fast-spiking cells began repetitive firing at a nonzero frequency.[1] The same kind of graph records both outcomes.

Manages Complexity

An F-I curve compresses many voltage trajectories or spike trains into one conditional current-to-rate relation. That compression makes thresholds and gains easier to compare, while discarding exact spike timing and subthreshold dynamics. It therefore needs its operating conditions attached. Tateno's distinct rate estimators show how a measurement convention changes the reported rate.[1] In Drion's model, increasing versus decreasing current steps can reveal hysteresis that a one-direction protocol obscures.[2]

Abstract Reasoning

First fix the bearer and protocol, then vary current while applying one rate rule consistently. Examine the resulting mapping before naming its shape or attributing a mechanism. To compare two curves, check that their current units, duration, rate estimator and step histories are compatible. Drion's specified Connor–Stevens model changes its curve as A-type potassium conductance varies.[2] That model result does not establish a universal channel rule for neurons.

The collapse test is concrete: if the input is no longer current, the output is no longer neuronal firing frequency, or only one input point is known, the asserted F-I relation has not been established. If an experiment keeps those roles but finds a new shape, the F-I identity survives.

Knowledge Transfer

The measurement logic transfers between different neuron types and between an experiment and a biophysical model: hold the typed axes fixed, state the operating convention, and compare the mapping. Tateno's cortical cells and Drion's Connor–Stevens model are unlike carriers with the same current and rate axes.[1][2] They support this specific transfer.

Transfer does not carry every curve feature with it. Tateno's type 1/type 2 contrast and Drion's conductance-dependent Type II → Type I → Type II* sequence show differing onset classes.[1][2] The current-to-rate question remains the same. A population-rate model or synaptic-input relation needs its own bridge before borrowing the single-neuron F-I name.

Examples

Canonical: measured cortical cells

Tateno and colleagues injected 600-ms current pulses into layer ⅔ cells of young rat somatosensory cortex. They measured current against firing frequency in regular-spiking pyramidal neurons and fast-spiking inhibitory interneurons. Under that preparation and estimator, the former showed near-continuous type 1 onset, whereas the latter showed a nonzero-frequency type 2 onset.[1] These are distinct curves of the same relation type, not proof that all pyramidal or inhibitory neurons behave that way.

Mapped back: measured cortical neuron → injected current series → spike-train rate under the reported interval convention → experimental F-I relation. The type contrast is an observed curve property, not a defining role.

Applied: a conductance-based model

Drion and colleagues used the Connor–Stevens neuron model to plot steady-state firing frequency against applied current density. At selected A-type potassium conductance values, its curve passes through Type II, Type I and Type II* onset regimes.[2] The model demonstrates how the same typed relation can reveal parameter-dependent excitability; it is a simulation, not a second cortical-slice measurement.

Mapped back: specified Connor–Stevens model → simulated applied-current steps → modeled steady firing rate → model F-I relation. Conductance changes shape, while current and rate remain the axes.

Structural Tensions

T1: comparable summary versus protocol fidelity. Current-to-rate curves let investigators compare excitability across cells and models. A compact curve can, however, hide adaptation, rate-window choices and step-history effects. Drion's hysteretic Type II* regime can look different when current steps run downward rather than upward; Tateno distinguishes initial from steady estimates.[1][2] Fixing one protocol supports comparison but may omit a relevant dynamic feature; changing protocols exposes it but prevents naive point-for-point comparison. Diagnostic: Are two curves commensurate under the same current and frequency convention, and what behavior was lost by that convention?

Structural–Framed Character

The entry is predominantly structural within a biological frame: it pairs quantitative input and output across cases. Its evaluative weight is descriptive; a steep or shallow curve does not make a neuron better. Human-practice dependence enters through current injection, pulse schedule and rate estimation, which shape the measured representation rather than creating neuronal firing itself. Tateno's finite pulse and interval convention and Drion's model protocol illustrate those choices.[1][2] Institutional origin supplies naming and reporting conventions, not the electrophysiological effect. Vocabulary portability stops where the neuron, current and spike-rate roles stop. Import versus recognition asks whether a new case literally has those roles or only resembles an input-output plot.

Its character: structurally defined within a neuroscience frame. A curve's quantitative form decides how current relates to firing, while the required neuronal carrier, electrical input and spike-rate convention keep the named identity domain-specific. The descriptive test neither rewards a curve's shape nor depends on the institution that reports it. Removing the neuronal and spike-train roles leaves a more general input-response mapping, not an F-I curve.

Structural Core vs. Domain Accent

The thin skeleton is a conditional quantitative mapping from an applied input level to a measured or modeled output rate. The domain accent is indispensable here: a neuronal carrier, applied electrical current, spikes, and a firing-frequency convention. Tateno supplies the measured cortical case and Drion the modeled Connor–Stevens case.[1][2] Those commitments separate the F-I curve from a generic dose-response or function graph. The two cases establish measured and modeled realizations within neuroscience, not a cross-domain F-I identity.

The portable skeleton raises a future-Prime question: can a conditional input-magnitude-to-output-rate relation be shown to retain its decision rules across at least two materially unlike domains, with explicit carrier, rate convention and protocol-dependent boundary? These two neuroscience cases alone cannot answer it. The F-I name remains attached to neuronal firing.

Input/output and Dose–Response Relationship are related ways to reason about inputs and responses. Input/output tracks directed exchange across a system boundary; Dose–Response Relationship emphasizes a measured response across input magnitudes and can carry different curve-shape assumptions. The F-I curve's literal claim is narrower: applied current and neuronal firing frequency under specified conditions. A shared word such as “response” does not by itself settle a genus or prerequisite relation.

Neighborhood in Abstraction Space

F-I curve sits in a sparse region of the domain-specific corpus (95th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2721 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-17

Not to Be Confused With

  • Frequency–conductance relation (f-g): the varied quantity is conductance rather than applied current.[1]
  • Spike-train or timing code: precise spike times can differ even when a rate summary agrees; the F-I curve reports frequency under its convention.[1]
  • One threshold or gain estimate: a feature of a curve, not the current-to-rate relation across a range.
  • Population rate transfer function: a different carrier unless its relation to a neuron-level applied-current curve is independently shown.
  • One required onset shape: the studied cells and model display different onset classes.[1][2]

References

[1] T. Tateno, A. Harsch and H. P. C. Robinson, “Threshold Firing Frequency–Current Relationships of Neurons in Rat Somatosensory Cortex: Type 1 and Type 2 Dynamics”, Journal of Neurophysiology 92(4) (2004), 2283–2294. Full original article inspected at an external PDF mirror, matched to publisher DOI; especially printed pp. 2284 (Methods) and 2286–2287 (Fig. 4 and discussion). registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t

[2] Guillaume Drion, Timothy O’Leary and Eve Marder, “Ion channel degeneracy enables robust and tunable neuronal firing rates”, Proceedings of the National Academy of Sciences 112(38) (2015), E5361–E5370. Full original PDF inspected in an institutional author repository; especially printed E5361–E5362 (Results and Fig. 1), E5364 (Fig. 4), and E5368 (Methods). registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o