Neurons & Spikes · Lab · ~8 min

The mystery neuron

Every lesson so far handed you the model with all its dials visible. This instrument holds a neuron whose firing threshold is hidden, and the only way in is experiment: inject a current, watch $300$ ms of membrane voltage, count the spikes. Your job is to measure the cell's rheobase — the smallest sustained current that ever makes it fire. Fewer trials means better science.

Prerequisite This lab drives the integrator from Lesson 2 — The neuron integrates. If the climb-and-fire loop is unfamiliar, run that instrument first.
Experiment

Pick a current with the input current slider and press Run 300 ms trial. The trace shows what an electrode would: the voltage climbing from $-70$ mV, and a spike each time the cell fires. No threshold line is drawn — that is the unknown. Every trial lands in the log below, and the trials used counter keeps score.

Try itStart coarse — one trial near $0.5$ nA, one near $3.0$ nA — then split the difference. Bracketing halves your uncertainty with every trial; idly sweeping the slider does not.
Trial log
no trials yet
About this model

What this lab models: a leaky integrate-and-fire cell, $\tau \frac{dV}{dt} = -(V - V_{rest}) + RI$, with $\tau = 20$ ms, $R = 10$ MΩ, $V_{rest} = -70$ mV, and an instant reset to $-70$ mV at threshold. Each trial is $300$ ms of constant current integrated numerically in $0.1$ ms steps — the same protocol as a current-clamp rheobase measurement.

What it leaves out: real spikes have a shape and a refractory period; real membranes are noisy, so near rheobase the same current can fire on one trial and stay silent on the next; real cells adapt, firing slower as a trial goes on; and a sharp voltage threshold is itself an approximation of continuous channel dynamics. In a real experiment your bracketing strategy would be fighting all four.

Source: the leaky integrate-and-fire model in Neuronal Dynamics, chapter 1.3 (Gerstner, Kistler, Naud, Paulsen).