Biological neuroscience

Read the cell as interacting compartments

Dendrites, the cell body and the axon describe structure, but a diagram alone does not specify how every neuron behaves. The familiar branched cartoon is a teaching abstraction; many fly neurons have different geometries. Membrane potential is measured across a membrane. When channels open, ion movement depends on both concentration and electrical differences. Depolarization means the potential becomes less negative relative to the chosen reference.

A spike is an event with a mechanism

Action potentials involve changing conductances and regenerative feedback. A computational spike timestamp keeps the event while discarding its waveform. That reduction can be useful when timing is the question, and inappropriate when channel dynamics are the question. Refractoriness limits responsiveness after activity; it should be specified explicitly rather than inferred from a plotting convention.

From synapse to network weight

A chemical synapse is not simply a signed number. Transmitter release, receptors, reversal potentials, timing and postsynaptic state shape its effect. A scalar artificial weight compresses these mechanisms into a task-dependent approximation. Excitation and inhibition should therefore be explained through effect and context. A predicted transmitter label in a connectome does not, by itself, measure every downstream response.

Reading a population

Rate and timing are different descriptions of the same spike train. A count over 100 milliseconds loses the order of the events. A population decoder may recover information that any one cell does not provide. To choose a representation, define the behavior and the observation window first, then test whether information survives your simplification.

SEE THE RELATIONSHIPS

From structure to behavior.

MAKE THE IDEA PRECISE

The mathematics.

Build the equation, step by step

1. Choose the observation window before counting. A count without a duration cannot define a rate.
2. Divide the spike count N by window length T in seconds. The resulting unit is spikes per second, conventionally reported in hertz.
3. Changing the window changes the estimate. A longer window reduces time resolution; a short window can be dominated by whether a single spike fell just inside or outside its boundary.

WORK THROUGH THE NUMBERS

Worked example

Five spikes in 0.1 seconds give 50 Hz. This does not mean spikes are exactly 20 ms apart. Two trains can have identical counts and very different event timing.

A small Python experiment.

Python 3 · Standard library only · Synthetic educational example. Run this in your own notebook or a local file.

spikes_ms = [4, 6, 20, 70, 98]
window_ms = 100
rate_hz = len(spikes_ms) / (window_ms / 1000)
intervals = [b-a for a,b in zip(spikes_ms, spikes_ms[1:])]
print(f"rate = {rate_hz:.0f} Hz; intervals = {intervals} ms")

Expected output

rate = 50 Hz; intervals = [2, 14, 50, 28] ms

Change one quantity, predict the result, and then run the code again. Explain any difference before adding complexity.

THE NEXT LAYER

Chapter depth map.

This introduction provides a first model. The following chapters are outlines for deeper future editions; their full treatments have not yet been published.

Cells and signals

Communication and adaptation

Read with the original sources.

These readings support the concepts and deeper investigation. The worked numbers and illustrations above are teaching examples.