# Integration, Inhibition and Neuromodulation

> In a passive compact cell, each synaptic input is a conductance that pulls the voltage toward its own reversal potential and adds to the total, so the balance point is a conductance-weighted average. This lesson treats each input as one more rope on the balance point from the action potential lesson, with no current injected through an electrode. With that rule you can predict how inputs sum in time and along a dendrite, why two inputs give less than double, why an inhibitory input can veto a spike without moving a resting cell's voltage, and why the same GABA input can help or hinder. It ends with neuromodulators, which change the ropes themselves: afterwards you can read a current-clamp experiment in which dopamine changes firing to matched current steps.

Course: Neuroscience (Synapses and plasticity). Level: university elective. About 38 minutes.
Lesson page (link this one when citing): https://stemhop.com/lessons/synaptic-integration/

Builds on: The Resting Membrane and Passive Signals · The Action Potential and Conduction · Synaptic Transmission and Quantal Release

## What you'll learn

- Predict how EPSPs and IPSPs sum in time (overlap set by the time constant) and in space (decay set by the length constant), treating synaptic inputs as conductances that add rather than as voltages, so that, in a passive isopotential cell with fixed, voltage-independent conductances and no injected current, the steady voltage is the conductance-weighted average of their reversal potentials.
- Explain shunting inhibition, define an IPSP by its effect on firing rather than its sign, and relate E_GABA to the chloride gradient and the timing and site of the input.
- Explain how dopamine acting on D1 receptors (through cAMP and PKA) and acetylcholine acting on muscarinic receptors (closing KCNQ channels through loss of PIP2) change excitability over hundreds of milliseconds to minutes.
- Interpret a current-clamp recording in which a modulator changes the firing evoked by matched current steps, including the receptor, kinase and drift controls.

**An inhibitory input can leave a resting neuron’s voltage exactly where it was and still stop the cell from firing.**

To see how, stop treating synaptic inputs as voltages that add, and start treating them as conductances that compete. Four puzzles show why the change is needed.

1. **The silent veto.** How can an input that moves a resting cell by zero millivolts block a spike?
2. **Less than double.** Why do two identical excitatory inputs, arriving together, give less than twice the depolarization of one?
3. **The two-sided GABA input.** How can one GABA input hold back excitation at one moment and help it a few milliseconds later?
4. **Same input, more spikes.** How can dopamine make a neuron fire more to exactly the same injected current?

The first three follow from the balance point you met in [the action potential](https://stemhop.com/lessons/action-potential/), with one addition. The fourth shows what that rule can and cannot tell you.

> **The one idea to carry through this lesson.** The balance point is the voltage the cell would settle at if its open channels stayed as they are. A synapse is one more rope on it. When a synapse opens, it does two things at once. It pulls the balance point toward its own reversal potential. And it adds to the total conductance, so every other rope counts for less and, in a passive cell, the voltage settles faster. Conductances add; in the compact passive model defined below, with no injected current, the balance point is their weighted average.

## One more rope: what a synapse does to the voltage

Picture the voltage scale from the action-potential lesson, with a marker showing the membrane potential, Vm. Every kind of open channel has a rope tied at its own **reversal potential** (E), the voltage at which current through those channels is zero. Each rope pulls the marker toward its end. How hard it pulls is its current:

> **I = g × (Vm − E)**

Here **g** is the **conductance**, measured in nanosiemens (nS), which mostly means how many of those channels are open. The term (Vm − E) is the **driving force**: the gap between the marker and the rope’s end. A rope pulls harder the more channels are open (a thicker rope) and the farther the marker sits from its end.

In the action-potential lesson the ropes belonged to ions. Here they belong to kinds of channel: the leak, and each kind of synaptic receptor. A synaptic input is a conductance, not current injected through an electrode, so it goes inside the average like any other rope.

In this lesson’s illustrative model, one equivalent rope, the leak, stands in for the channels open at rest. It is 10 nS, tied at −65 mV, so the marker sits at −65 mV.

Now a glutamate synapse fires. Its AMPA-type receptors open and add a second rope. For AMPA receptors E is near 0 mV, so at rest their current depolarizes. The marker is pulled up toward 0 mV but held back by the leak, so it settles between the two. That rise is a **postsynaptic potential**. When the transmitter is cleared and the channels close, the new rope goes slack and the marker drifts back to rest.

*Figure (AI-generated illustration).* An opening AMPA synapse adds a rope tied far above rest, which pulls the marker partway up while the leak rope holds it back, so the voltage settles between the two ends. Image: two panels show a vertical voltage scale with a ring marker: at rest the marker sits at the leak rope's peg; when an AMPA synapse opens, a thinner gold rope tied far above pulls the marker partway up, so it settles between the two pegs, held back by the thicker leak rope.

The GABAA receptor passes chloride and some bicarbonate, so EGABA is set by both ion gradients and by how readily the channel passes each. The Nernst equation gives the chloride equilibrium potential, ECl, from the chloride concentrations ([derived in transport across membranes](https://stemhop.com/lessons/membrane-transport/)). Under ordinary gradients bicarbonate shifts EGABA positive of ECl, more so in cells with low internal chloride; the offset can exceed 10 mV. Where the GABA rope is tied therefore depends on the cell’s chloride gradient.

An input that makes the cell more likely to fire an action potential is an **excitatory postsynaptic potential (EPSP)**. One that makes firing less likely is an **inhibitory postsynaptic potential (IPSP)**. Both are defined by their effect on firing, not by which way the trace moves. See [synaptic transmission](https://stemhop.com/lessons/synaptic-transmission/) and [the resting membrane](https://stemhop.com/lessons/resting-membrane-passive/) for the background.

**Predict.** A GABAA synapse opens chloride channels whose reversal potential, EGABA, is −75 mV. The cell is at rest at −65 mV. Which way does the marker move?

> **Check your prediction.** Down, toward −75 mV. The new rope is tied below the marker, so the balance point moves below rest and Vm follows. That downward bump is the textbook IPSP. Whether a GABA rope pulls down, up or not at all depends only on where EGABA sits relative to the marker.

Here is the balance point written out: the same average as in the action-potential lesson, with synaptic ropes. With the leak (L), an excitatory synapse (E) and an inhibitory synapse (I) all open and steady, the steady voltage is

> **Equation card: Where the voltage settles when the leak and synapses are open and steady: the balance point.**
>
>     V_ss = (gL × EL + gE × EE + gI × EI) / (gL + gE + gI)
>
> In words: The steady voltage is the average of the ropes' ends, each counted in proportion to its conductance.
>
> - V_ss: the steady-state voltage, the balance point (mV)
> - gL: the leak conductance, 10 nS in this lesson's cell (nS)
> - EL: the leak's reversal potential, −65 mV here (rest) (mV)
> - gE: the excitatory synaptic conductance, for example AMPA (nS)
> - EE: the excitatory reversal potential, near 0 mV for AMPA (mV)
> - gI: the inhibitory synaptic conductance, for example GABA_A (nS)
> - EI: the inhibitory reversal potential, E_GABA (mV)
>
> Sanity check: With no synapse open (gE = gI = 0), V_ss = EL: the cell sits at rest. A synapse can pull the balance point toward its own E, never past it. Opening any synapse makes the denominator (the bottom of the fraction) bigger, which shrinks every other rope's share.
> Try one number Open a small GABA conductance, gI = 2.5 nS with EI = −75 mV, and no excitation (round values to keep the arithmetic easy). Top: 10 × (−65) + 2.5 × (−75) = −650 − 187.5 = −837.5. Bottom: 10 + 2.5 = 12.5 nS. V_ss = −837.5 / 12.5 = −67.0 mV: 2 mV below rest, a small IPSP, since this rope holds only a fifth of the total conductance.

Each rope’s end is weighted by that rope’s share of the total conductance; the excitatory synapse’s share, for example, is gE / (gL + gE + gI). These are the two jobs in the box: a new end in the average, and a bigger bottom line, the total conductance, which shrinks every other share.

The formula has a scope. It holds for a passive cell below threshold whose voltage is the same everywhere (**isopotential**), with no current injected through an electrode, and with conductances that stay fixed and do not depend on voltage. Even then it gives the balance point, not Vm at every instant: after a conductance changes, Vm takes time to catch up, as the next section shows. The later sections stretch these limits one at a time: brief inputs, injected current, dendrites, receptors whose conductance depends on voltage, and modulators.

## Adding inputs in time and in space

**Film transcript: Adding Inputs in Time and Space**

### Film transcript: Adding Inputs in Time and Space

Two inputs arrive 10 ms apart. Does the second start from rest, or from what the first left behind? When conductances change, the marker does not jump to the new balance point. It moves there gradually. Changing the voltage means moving charge onto or off the membrane's capacitance. That takes time. The membrane time constant, τ, says how long. It is the time to get 63% % of the way to a new voltage. τ equals the capacitance divided by the total open conductance. In our cell, that is 20 ms. Because τ uses the total conductance, in this passive model every synapse that opens shortens it. So an EPSP has a shape. For a brief, fast input in our cell, it rises while synaptic current flows, over a few milliseconds.

Once that current fades, the voltage decays back toward rest, with the resting time constant. If a second EPSP arrives before the first has gone, it starts from the first one's tail. That is temporal summation. Two brief, idealized pulses arrive 10 ms apart. Each jumps the voltage by 2 mV. Cell A has a τ of 20 ms. Cell B's is 10 Which ends higher?

Cell A. After 10 ms, its first EPSP has decayed only to about 61%. percent. The second lands on top of that. Cell B's first EPSP has fallen to about 37%. So cell A ends a little over 3 mV, and cell B under 3 A longer τ leaves a longer tail, and that means more summation. Remember cell B. It returns later in this lesson. Now distance. Most synapses sit on dendrites, not on the cell body. Current entering a dendrite flows toward the soma. But some leaks out across the membrane along the way. In a long, uniform passive dendrite, a steady voltage change shrinks by the same fraction over each equal stretch. The length constant, λ, is the distance over which it falls to 37%.

percent. Put a synapse half a length constant from the soma. About 61% percent of its steady depolarization arrives. A brief EPSP loses more, because capacitance along the way also filters it. Adding inputs at different places is spatial summation. For the same local change, inputs near the soma, or on dendrites with a long λ, lose less. Time and distance decide how much of each input is still there when the cell decides. The balance point decides what they add up to.

From the film

- When conductances change, the voltage moves gradually to the new balance point, with a time constant equal to the capacitance divided by the total open conductance.
- A second EPSP that arrives before the first has gone starts from its tail, so a longer time constant gives more temporal summation.
- A steady voltage change in a long passive dendrite shrinks by the same fraction over each equal stretch, falling to 37% over one length constant.

### In time: the second EPSP rides on the first

When conductances change, the marker does not jump to the new balance point. It moves there gradually, because changing the voltage means moving charge onto or off the membrane’s capacitance (C). The **membrane time constant**, τ, says how long that takes: the time a passive membrane takes to get 63% of the way to a new steady voltage. It equals input resistance times capacitance: C divided by the total open conductance.

Our cell has C = 200 pF and 10 nS of leak. Its input resistance is R = 1/gL = 1/(10 × 10−9 S) = 100 MΩ, and its time constant is τ = C/gL = (200 × 10−12 F)/(10 × 10−9 S) = 20 ms. Synaptic conductances go in the same bottom line as the leak, so in this passive model every synapse that opens shortens it:

> **Equation card: How fast the voltage approaches the balance point while synapses are open.**
>
>     τ_eff = C / (gL + gE + gI)
>
> In words: The effective time constant is the capacitance divided by the total open conductance.
>
> - τ_eff: the effective membrane time constant: the time to get 63% of the way to the balance point (ms)
> - C: the membrane capacitance, 200 pF in this lesson's cell (pF)
> - gL: the leak conductance (nS)
> - gE: the excitatory synaptic conductance (nS)
> - gI: the inhibitory synaptic conductance (nS)
>
> Sanity check: With no synapse open, τ_eff = C / gL, the resting 20 ms. Any open synapse, excitatory or inhibitory, adds to the denominator, so τ_eff gets shorter. Units: pico over nano is 10^−12 / 10^−9 = 10^−3, so pF / nS gives ms.
> Try one number Open a 5 nS excitatory conductance alongside the 10 nS leak. Total conductance: 10 + 5 = 15 nS. τ_eff = 200 pF / 15 nS = 13.3 ms, shorter than the resting 20 ms.

So an excitatory postsynaptic potential (EPSP) has a shape. For a brief, fast AMPA input in our compact passive cell, the voltage rises while the synaptic current flows, over a few milliseconds. Once that current has faded, the voltage decays back toward rest with the resting time constant. Other receptor types, dendrites and active channels give other shapes. If a second EPSP arrives before the first has gone, it starts from the first one’s tail. Adding inputs that arrive at different times is **temporal summation**, and a longer τ gives more overlap.

**Predict.** Give the cell two brief, idealized excitatory charge pulses of 0.4 pC, 10 ms apart. Each jumps the voltage by q/C = 0.4 pC / 200 pF = 2 mV, then decays. Cell A is our cell, with τ = 20 ms. Cell B is the same cell with τ = 10 ms. Just after the second pulse, which cell is more depolarized, and roughly by how much?

> **Check your prediction.** Cell A. After 10 ms its first EPSP has decayed only to e−10/20, about 61% of its size, so just after the second pulse the depolarization is 2 + 2 × e−10/20 = 2 + 1.21 = 3.21 mV. Cell B’s first EPSP has fallen to e−10/10, about 37%, giving 2 + 0.74 = 2.74 mV. A longer τ leaves a longer tail, which means more summation. Remember cell B: later in the lesson you will turn cell A into it by opening one inhibitory synapse.

### In space: an input fades on its way to the soma

Most synapses sit on dendrites, not on the cell body. Current entering a dendrite flows toward the soma, but some leaks out across the dendrite’s membrane along the way, so the voltage change shrinks with distance. For a steady input in a long passive dendrite, it falls by the same fraction over each equal stretch. The **length constant**, λ (“lambda”), is the distance over which it falls to 37%:

> **V(x) = V0 × e−x/λ**

Here V0 (“V zero”) is the steady change at the synapse and V(x) what is left a distance x away. With illustrative numbers, λ = 500 µm and a synapse 250 µm from the soma: x/λ = 250/500 = 0.5, and e−0.5 = 0.61 ([e and ln](https://stemhop.com/math-toolkit/#e-and-ln)), so 61% of its steady depolarization reaches the soma. A brief EPSP loses more, because the capacitance along the way also filters it. Adding inputs that arrive at different places is **spatial summation**. For the same local voltage change, an input nearer the soma, or on a dendrite with a longer λ, loses less on the way. In real dendrites, synaptic strength, branching and active channels also matter.

So time (τ) and distance (λ) decide how much of each input is still present when and where the cell decides whether to fire. The balance point decides what the inputs add up to, and the next section shows that they do not simply add.

## Why two inputs make less than double

**Film transcript: Less Than Double**

### Film transcript: Less Than Double

If one input depolarizes a cell, do two identical inputs depolarize it twice as much? Take our cell. The leak is 10 nS, tied at minus −65 mV. Threshold is minus −50 All the values here are illustrative. The cell is passive, and its voltage is the same everywhere. Voltages above threshold are passive predictions, with the spike switched off. Open a steady excitatory input of 5 nS, tied at 0 mV. The passive prediction is near minus −43 mV. That is about 22 mV of depolarization, above threshold. A real cell with this threshold would fire on the way. It gets there with a shorter time constant, about 13 ms, because the total conductance grew. Now open a second, identical input at the same time.

Does the depolarization double, to about 43 mV? No. The passive prediction is about −32.5 mV. That is about one and a half times as much, not twice. To see why, measure the voltage from rest. Then the leak drops out of the average. The leak is tied at rest, so its top-line term is zero. But the leak is still in the bottom line, and still carries current. Each top-line term is a synapse's conductance, times how far its rope's end sits from rest. The bottom line is still the total conductance. Two inputs double the top line. But they also enlarge the bottom line. With one input, the top is 325, over 15. With two, the top doubles to 650, but the bottom also grows, to 20.

So the second input dilutes the first. That is the same effect, seen as ropes. The marker is nearer 0 mV, so each AMPA rope's driving force has shrunk. This is sublinear summation. Inputs sharing one small patch of membrane show it most. Notice where the surprise lived: in the bottom line, the total conductance. Each new input adds to the total conductance. So the voltage settles at a weighted average, not a sum of millivolts. Small inputs nearly add. Large ones fall short.

From the film

- In the model, two identical excitatory inputs give less than twice the depolarization of one.
- A second input enlarges the total conductance, so it dilutes the first; this is sublinear summation.
- Inputs that are small beside the leak add almost linearly, and the shortfall grows as inputs get larger.

All values from here on are illustrative. Treat the cell as isopotential and passive, with C = 200 pF, gL = 10 nS, EL = −65 mV and a fixed spike threshold of −50 mV. Real thresholds shift with channel state ([the action potential](https://stemhop.com/lessons/action-potential/)). Voltages above −50 mV in this section are passive predictions: where the voltage would settle if the spike mechanism were switched off. A real cell with this threshold would fire on the way.

Turn on a sustained excitatory conductance, gE = 5 nS with EE = 0 mV:

> Vss = (10 × −65 + 5 × 0)/(10 + 5) = −650/15 = −43.3 mV

That is a 21.7 mV depolarization. The voltage approaches it with τeff = 200/15 = 13.3 ms. When does it cross −50 mV? The gap still left to close at time t is 21.7 × e−t/13.3 mV, so solve −50 = −43.3 − 21.7 × e−t/13.3 in two steps.

- *Isolate the exponential:* the gap left at threshold is 50 − 43.33 = 6.67 mV, so e−t/13.3 = 6.67/21.67 = 0.308.
- *Undo the e with a natural log:* t/13.3 = ln(1/0.308) = ln(3.25) = 1.18, so t = 13.3 × 1.18 = 15.7 ms ([e and ln](https://stemhop.com/math-toolkit/#e-and-ln)).

**Predict.** Now open a second, identical 5 nS input at the same time, so that gE = 10 nS. Does the depolarization double, to about 43 mV?

> **Check your prediction.** No. Vss = (10 × −65 + 10 × 0)/(10 + 10) = −650/20 = −32.5 mV, a 32.5 mV depolarization rather than 43.3 mV. Two inputs gave about one and a half times the depolarization of one, not twice.

To see why, measure the voltage from rest. Subtracting EL from both sides of the balance point gives the same average in a form where the leak drops out of the top line:

> **Equation card: How far from rest the balance point sits, so you can see what each synapse adds.**
>
>     V_ss − EL = (gE × (EE − EL) + gI × (EI − EL)) / (gL + gE + gI)
>
> In words: The change from rest is each synapse's conductance times how far its end sits from rest, added up, divided by the total conductance.
>
> - V_ss: the steady-state voltage, the balance point (mV)
> - EL: the leak's reversal potential, which is rest (mV)
> - gE: the excitatory conductance (nS)
> - EE: the excitatory reversal potential (mV)
> - gI: the inhibitory conductance (nS)
> - EI: the inhibitory reversal potential (mV)
> - gL: the leak conductance, still in the denominator (nS)
>
> Sanity check: With no synapse open, the numerator is zero and V_ss = EL, rest. A synapse that is small beside gL barely changes the denominator, so its effect is close to its numerator term divided by gL.
> Try one number Take the small GABA input from the first card: gI = 2.5 nS, EI = −75 mV, no excitation. Numerator: 2.5 × (−75 − (−65)) = 2.5 × (−10) = −25. Denominator: 10 + 2.5 = 12.5. Change from rest: −25 / 12.5 = −2.0 mV, so V_ss = −65 − 2.0 = −67.0 mV, the same as the full average gave.
> Where it comes from optional Step 1, subtract EL from the balance point, writing EL as EL × (gL + gE + gI) / (gL + gE + gI) so both share one denominator. Step 2, in the numerator, gL × EL − gL × EL = 0, so the leak’s term vanishes. Step 3, what remains pairs each synapse with its own gap: gE × (EE − EL) + gI × (EI − EL).

Each top-line term is a synapse’s conductance times how far its rope’s end sits from rest: the pull that rope would have if the marker sat at rest. The leak’s term is zero because its rope is tied at rest, but the leak has not gone. It is still in the bottom line, and it carries current whenever the marker is away from rest.

With one input the top line is 5 × 65 = 325 over a bottom line of 15. With two, the top line doubles to 650, but the bottom line also grows, to 20.

That is one effect, seen two ways. In the formula, the second input enlarges the bottom line, so it dilutes the first. In rope terms, the marker now sits nearer 0 mV, so each AMPA rope’s driving force, and its actual pull, has shrunk. This is **sublinear summation**. Inputs sharing one small patch of membrane show it most. Size matters too: each 5 nS input here is half the leak, a strong input. Inputs that are small beside the leak add almost linearly; the shortfall grows as inputs get larger.

## Inhibition that moves nothing at rest: shunting

**Film transcript: Inhibition That Moves Nothing at Rest**

### Film transcript: Inhibition That Moves Nothing at Rest

Can an input stop a spike without moving a resting cell's voltage at all? Our cell starts at rest. Alone, the 5 nS excitatory input would carry it toward minus −43 past threshold. It would fire on the way. Now meet an inhibitory rope of 10 nS. It is tied exactly at rest, minus −65 mV. First, what does this rope do to a resting cell?

Nothing. Its rope is tied exactly where the marker sits, so it pulls on nothing. The trace stays at minus −65 Now start again from rest, and switch on the excitation and the shunt together. Where does the marker head? Toward minus −52 mV. That is below threshold, so the cell never fires. The shunt's top-line term is zero, so it adds nothing to the top. It acts through the bottom line, the total conductance. The depolarization falls from about 22 mV to 13 This is shunting inhibition. It moves nothing at rest. It shows itself when something else tries to move the voltage. Then it cuts that movement. The shunt is a rope tied where the marker already sits.

It holds the marker back whenever another rope drags it away. Electrically, the shunt lowers input resistance. Excitation moves the voltage less, whether it is a synaptic conductance or injected current. It changed only the bottom line. The shunt does a second thing. It shortens the time constant. Now switch off the excitation, and let the cell return to rest. With the leak and the shunt open, τ is 10 ms. That is half the leak's 20 That is cell B from the summation chapter. Give it two brief pulses again, 10 ms apart. Less of the first is left for the second to ride on. In this idealized case, each jump stays the same size. Only the tail is shorter.

A real synaptic current flows for a millisecond or more. While it flows, the shunt drains charge, so the peak is smaller too. A shunt tied at rest moves nothing at rest. It shows itself as smaller EPSPs and less summation.

From the film

- A shunting inhibitory input tied at rest moves nothing at rest, because its rope is tied exactly where the voltage already sits.
- It acts through the total conductance: it lowers input resistance, so excitation moves the voltage less and a spike can be stopped.
- A shunt also shortens the time constant, so less of an earlier EPSP is left for the next one to ride on.

Now add a rope whose only job is the bottom line: an inhibitory conductance gI = 10 nS whose reversal potential equals rest, EI = −65 mV.

**Predict.** What does this rope do to the resting cell? Then compare two trials that start at rest. In one, the 5 nS excitation switches on alone. In the other, the excitation and this rope switch on together. Which trial fires?

> **Check your prediction.** *At rest, nothing.* With gE = 0, Vss = (−650 − 650)/20 = −65 mV. In the from-rest form its top-line term is 10 × 0 = 0, because its rope is tied exactly where the marker sits.
>
> *With the excitation, it vetoes the spike.* Alone, the excitation heads for −43.3 mV and crosses threshold on the way, after about 15.7 ms. With the shunt on as well, Vss = (−650 + 0 − 650)/(10 + 5 + 10) = −1300/25 = −52.0 mV. The voltage rises from −65 mV toward −52 mV and never reaches threshold: no spike. In the from-rest form the top line stays at 325 while the bottom line grows from 15 to 25, so the depolarization falls from 21.7 mV to 13.0 mV.

This is **shunting inhibition**. Once another rope drags the marker away from rest, the shunt resists: at −52 mV it passes 10 nS × 13 mV = 130 pA back toward rest. It cut the depolarization by the factor 15/25 = 0.60. (Strictly, a shunt leaves rest unmoved only when its reversal potential equals rest. Tied near rest, it also makes a small inhibitory postsynaptic potential (IPSP) or depolarization.)

Electrically, the shunt lowers the input resistance, so excitation moves the voltage less, whether it arrives as a synaptic conductance or as current injected through an electrode. This is the lesson’s one idea at work: the shunt changed only the bottom line.

*Interactive on the lesson page:* Inhibition you cannot see — conductances against threshold

### The shunt also shortens τ

With the leak, the excitation and the shunt all open, τeff = 200/25 = 8 ms. Now switch off the excitation and let the voltage settle back to rest. Only the leak and the shunt are open: 20 nS in all, so τ = 200/20 = 10 ms, half the 20 ms of the leak alone. Your cell has just become cell B from the summation section.

Give it the same two 0.4 pC pulses, 10 ms apart: just after the second, the depolarization is cell B’s 2.74 mV, not cell A’s 3.21 mV. In this idealized case the shunt did not shrink each jump, because an instantaneous charge pulse sets the jump by charge and capacitance alone. It shortened the time constant, so less of the first excitatory postsynaptic potential (EPSP) was left for the second to ride on. A real synaptic current lasts a millisecond or more, so the shunt also drains charge while the current flows, and the EPSP peak itself gets smaller.

So a shunt works against excitation twice: it shrinks the depolarization from sustained excitation, and it cuts the summation of brief inputs. Whether that divides or subtracts from the firing rate is a separate question; in one study it subtracted.

## When GABA pushes up: timing and site

**Film transcript: When GABA Pushes Up**

### Film transcript: When GABA Pushes Up

Can one GABA input hold back excitation, and then help it a few milliseconds later? The shunt was tied exactly at rest. But the GABA rope's end depends on the chloride gradient. Suppose it is tied at minus −55 mV. That is above rest, but below threshold at minus −50 A 10 nS GABA rope opens in our resting cell. Where does the marker settle?

At minus −60 halfway, because the two ropes are equally strong. This GABA input depolarized the cell. Now strong excitation drives the voltage above minus −55 toward threshold. Which way does the GABA rope pull? Down, back toward minus −55 And its conductance enlarges the bottom line. So it opposes further depolarization. Then the GABA conductance closes. Excitation arrives a few milliseconds later. Does it start nearer threshold, or farther?

Nearer. The conductance has gone, but the depolarization it left decays with τ. The later input starts higher up. At each moment, an open synapse pulls the voltage toward its own reversal potential. So this GABA current depolarizes while the voltage is below its end. It hyperpolarizes once the voltage is above it. Current direction alone does not say whether GABA helps firing. For that, compare the whole response with and without it. With steady inputs, it can never carry a cell sitting below threshold past threshold. The new balance point stays between the old one and minus −55 Brief inputs are different. In a model of our cell, one fast EPSP alone stopped just short of threshold.

With the GABA conductance already open, it started from minus −60 and just crossed. The open GABA conductance made the EPSP itself smaller. But the 5 mV head start won. Gulledge and Stuart studied this in layer 5 pyramidal neurons, in cortex slices from rats three to four weeks old. GABA responses depolarized the cells from rest, though the GABA reversal potential was below threshold. GABA at the cell body inhibited excitation arriving at the same time. It made later excitation more likely to fire the cell. GABA on the dendrites was excitatory, whatever its timing. The balance point offers a reading of the timing result. Likely, the conductance had closed while some depolarization remained.

This does not make GABA excitatory in adult cortex in general. In that preparation, its sign depended on timing and site. The rope's end can also move. When chloride enters faster than transporters can remove it, the GABA reversal potential creeps upward. GABA pulls the voltage toward its reversal potential. Whether that helps or hinders firing depends on timing, site and the other inputs.

From the film

- GABA pulls the voltage toward its own reversal potential, so it depolarizes while the voltage is below that level and hyperpolarizes once it is above.
- The direction of GABA current at one moment does not say whether it helps or hinders firing; compare the whole response with and without it.
- With E\_GABA below threshold, a steady GABA conductance cannot carry a cell past threshold, but brief inputs can give the opposite answer.
- In the rat cortex slices of Gulledge and Stuart, GABA's effect depended on timing and site, which does not make GABA excitatory in adult cortex in general.

The shunt was tied exactly at rest. EGABA need not be, and an inhibitory postsynaptic potential (IPSP) can even depolarize. If EGABA lies above rest but below threshold, GABA can move the voltage up yet oppose larger depolarizations; which effect wins depends on what else arrives, and when and where.

**Predict.** In our cell, a 10 nS GABA conductance opens with EGABA at −55 mV, between rest (−65 mV) and threshold (−50 mV).

1. Where does the resting cell settle?
2. Strong excitation now drives Vm above −55 mV on its way to threshold. Which way does the GABA rope pull?
3. The GABA conductance closes, and excitation arrives a few milliseconds later. Does it start nearer to threshold or farther from it?

> **Check your prediction.** 1. At −60 mV, halfway between −65 and −55 mV, because the leak and the GABA rope are equally strong. The GABA input depolarized the cell.
> 2. Down. Above −55 mV the GABA rope pulls back toward −55 mV, and its conductance enlarges the bottom line, so it opposes further depolarization.
> 3. Nearer. The conductance has gone, but the depolarization it left decays with τ, so the later input starts from higher up.

So the GABA current depolarizes while Vm is below EGABA and hyperpolarizes once Vm is above it, and its conductance shunts in both cases. The direction of the current at one moment does not by itself say whether GABA makes a spike more or less likely; for that, compare the whole response with and without GABA. Steady inputs give one clear answer. If the cell sits below threshold, a steady GABA conductance can never carry it past threshold: the new balance point lies between the old one and EGABA, and both are below threshold.

Brief inputs can give the opposite answer. In a model of this lesson’s cell, a fast excitatory postsynaptic potential (EPSP) sized to fall just short of threshold on its own crossed it, by a fraction of a millivolt, when the GABA conductance had opened beforehand and stayed open. The open conductance made the EPSP itself smaller, but its 5 mV head start, from −60 mV, won (the model is in Going deeper). So “inhibitory whenever EGABA is below threshold” is a rule for steady inputs, not for every input.

Gulledge and Stuart (2003) studied this in layer 5 pyramidal neurons in somatosensory cortex slices from 3- to 4-week-old rats, with a recording method that leaves the cell’s own chloride in place. GABA responses depolarized the cells from rest and, combined with proximal excitation (near the cell body), could help trigger spikes, even though EGABA was below threshold. Timing mattered: somatic GABA, applied at the cell body, inhibited excitation that arrived at the same time but facilitated excitation that arrived later, making it more likely to fire the cell. Site mattered too: dendritic GABA responses were excitatory whatever their timing. The balance point offers a reading of the somatic timing pattern: likely, the conductance had closed while some depolarization remained. The compact-cell picture here does not address the dendritic result.

This does not make GABA excitatory in adult cortex in general; in that preparation, its sign depended on timing and site. The larger depolarizing GABA responses of immature neurons, which come from higher internal chloride, belong to development: see [CNS synapses](https://stemhop.com/lessons/cns-synapses/).

The rope’s end can also move during activity. When GABAA channels let chloride in faster than transporters such as potassium-chloride cotransporter 2 (KCC2), a potassium-chloride cotransporter, can remove it, internal chloride rises and EGABA creeps upward.

## Dendrites that do not simply add

**Film transcript: Branches and Modulators Change the Ropes**

### Film transcript: Branches and Modulators Change the Ropes

What happens when a rope grows stronger with voltage, or a modulator changes the ropes themselves? A real neuron is not one compact patch. Each patch of dendrite has its own ropes, and passes current to its neighbours. So the patches set each other's voltages. The length constant and the capacitance shrink and smooth what reaches the soma. Polsky, Mel and Schiller stimulated two sites on thin dendrites of rat neocortical pyramidal neurons. First each alone, then both together. Picture their comparison as a graph. Along the bottom, the response expected if the two inputs simply added. Up the side, the response actually measured. A point on the 45 degree line means linear addition.

Inputs far apart, or on different branches, fell on the line. They summed linearly. Nearby inputs on one thin branch summed sigmoidally. The response rose slowly, then steeply, then levelled off. So location set the combination rule. The result supports the idea that thin branches act as separate integrating subunits. The steep phase is generally attributed to local regenerative events. NMDA receptors are thought to take part. Once glutamate and a co-agonist have bound, an NMDA receptor is a rope that thickens as the marker rises. Depolarization pushes out the magnesium ion that blocks it. Over the right voltage range, it thickens faster than its driving force shrinks.

Then its pull grows. Now modulators. A fast synapse adds a rope briefly. A modulator can instead change the ropes the cell already has. In sympathetic neurons, Gq-coupled M1 muscarinic receptors close KCNQ potassium channels. These carry the M-current. In cells expressing these channels, depleting the membrane lipid PIP2 alone was enough to suppress the current. So the lipid itself can regulate the channel. Closing these channels removes a potassium rope. What happens to the response to the same current step?

In a passive approximation, closing a potassium conductance raises the resting voltage a little and shrinks the bottom line. The same sustained current step then gives a larger steady change, and takes longer to reach the same fraction of it. Closing them increases excitability. The neuron fires more to the same input. Dopamine on D1 receptors is the harder case. In mouse striatal neurons, firing to the same injected current rose. It rose within hundreds of milliseconds. After one bout of stimulation, it stayed raised for at least 10 minutes. With synaptic receptors blocked and drift controlled, the cell's own conductances or active currents must have changed.

A spike count cannot say which channel. Branches can make a rope grow with voltage. Modulators change the ropes themselves. The rule tells you what kind of change to look for.

From the film

- In Polsky, Mel and Schiller's experiment, nearby inputs on one thin branch summed sigmoidally, while widely separated inputs or inputs on different branches summed linearly, so location set the combination rule.
- Thin branches are thought to act as separate integrating subunits, with NMDA receptors thought to take part in the steep phase.
- In the lesson's picture, closing KCNQ channels increases excitability, so the neuron fires more to the same input.
- With dopamine acting on D1 receptors in mouse striatal neurons, firing to the same injected current rose, so the cell's own conductances or active currents changed, though a spike count cannot say which.

A real neuron is not one compact patch. Each patch of dendrite has its own ropes and capacitance, and it also passes current along the dendrite to its neighbours. So the patches set each other’s voltages and must be solved together; the balance point of one isolated patch is only a first guide. Passive cable theory, with λ and the membrane capacitance, describes how a local change spreads and is filtered on its way to the soma. Sublinear summation is what the rule predicts for two inputs in one small patch. Thin dendrites can also do the opposite.

Polsky, Mel and Schiller (2004) used [confocal](https://stemhop.com/methods/confocal-microscopy/) imaging to find thin dendrites of rat neocortical pyramidal neurons in slices. They stimulated inputs at two sites on them, first each alone and then both together.

*Figure (AI-generated illustration).* Polsky, Mel and Schiller asked whether location changes how inputs combine by pairing two sites close together on one thin branch against sites on different branches of the same pyramidal neuron. Image: a pyramidal neuron is drawn with two gold input sites close together on one thin basal branch and two sage input sites on different branches, the two arrangements compared in the experiment.

Picture the graph that comparison gives. Along the bottom is the response you would expect if the two inputs simply added: the sum of the two separate responses. Up the side is the response actually measured when both were given together. A point on the 45-degree line means the inputs added linearly.

- Inputs that were widely separated, or on different branches, fell on that line: they summed linearly.
- Nearby inputs on the same thin branch summed **sigmoidally**: as the combined input grew, the response rose slowly, then steeply, then levelled off.

In these cells, location set the combination rule. The result supports the idea that thin branches act as separate integrating subunits, so where a synapse lands helps decide what the neuron computes.

The steep phase is generally attributed to local regenerative events in the branch, in which N-methyl-D-aspartate (NMDA) receptors are thought to take part; their voltage-dependent block is in [ion channels as receptors](https://stemhop.com/lessons/ion-channels-as-receptors/). In rope terms, once glutamate and a co-agonist have bound, an NMDA receptor is a rope that gets thicker as the marker rises, because depolarization relieves its block by magnesium ions (Mg²⁺). Over the voltage range where it thickens faster than its driving force shrinks, its pull grows, which is one way a local response can turn steep.

## Modulators change the ropes

A fast synapse adds a rope for milliseconds. A modulator can instead change ropes the cell already has. **Neuromodulators** act mostly through G-protein-coupled receptors and change how a neuron responds to input, often by altering its ion channels, over hundreds of milliseconds to minutes. Modulators can also change synaptic transmission; this section follows their effects on the cell’s own channels.

### Acetylcholine removes a potassium rope

KCNQ2/KCNQ3 channels carry the **M-current**, a slowly activating potassium current that opposes repetitive firing, named for its suppression by muscarinic agonists. Gq-coupled muscarinic receptors activate phospholipase C, which hydrolyzes the membrane lipid phosphatidylinositol 4,5-bisphosphate (PIP2) into inositol trisphosphate (IP3) and diacylglycerol (DAG) ([PLC, IP3, DAG and PKC](https://stemhop.com/lessons/plc-ip3-dag-pkc/)). Which event closes the channel? Suh, Inoue, Meyer and Hille (2006) depleted PIP2 directly, without PLC, in tsA-201 cells expressing KCNQ2/KCNQ3, and the current fell: loss of PIP2 was enough. The lipid itself could act as the channel’s regulator, without its breakdown products or calcium.

The rule predicts the direction. In a passive picture, treat the KCNQ channels open near rest as a rope tied near EK. Closing them lets the balance point rise a little and shrinks the bottom line. A sustained current step injected through an electrode then gives a larger steady change (ΔV = I / total conductance) and takes longer to reach a given fraction of it (τ = C / total conductance). Closing KCNQ channels increases excitability, so the neuron fires more to the same input. A synaptic excitatory postsynaptic potential (EPSP) need not grow, though, because the depolarized cell has less driving force for excitation.

### Dopamine on D1 receptors

Here the rule gives the logic of the experiment, but not the channel. If the same injected current drives more spikes, and synaptic drive and recording drift are controlled, the cell’s own conductances or active currents have changed. A spike count cannot say which.

In D1 receptor-expressing striatal projection neurons, the D1 receptor couples through Gs/olf to adenylyl cyclase, which raises cAMP and activates protein kinase A, PKA ([GPCRs and G proteins](https://stemhop.com/lessons/gpcrs-and-g-proteins/), [cAMP and PKA](https://stemhop.com/lessons/camp-and-pka/)). Lahiri and Bevan (2020) stimulated dopamine axons optogenetically in slices of dorsolateral striatum from adult mice of both sexes. They recorded identified D1 neurons by [perforated patch](https://stemhop.com/methods/patch-clamp/), with GABA, glutamate and acetylcholine receptors blocked. Firing to the same injected current rose within hundreds of milliseconds and stayed raised for at least 10 minutes after one bout of stimulation. A D1 antagonist blocked the effect, and their results suggest that the persistence required PKA. Conventional whole-cell recording showed little or no modulation, possibly because it dialyzes the cell interior.

Cascade timing is in [the shape of a signal](https://stemhop.com/lessons/shape-of-a-signal/); lasting changes lead to [synaptic plasticity](https://stemhop.com/lessons/synaptic-plasticity/), and dopamine returns in [the basal ganglia](https://stemhop.com/lessons/basal-ganglia/) and [motivation and reward](https://stemhop.com/lessons/motivation-and-reward/).

### Design problem

In a made-up perforated-patch recording from a D1 striatal projection neuron in a mouse [brain slice](https://stemhop.com/methods/brain-slice-electrophysiology/), with fast synaptic receptors blocked, a 200 pA, 1 s current step gives 8 spikes. After [optogenetic](https://stemhop.com/methods/optogenetics/) stimulation of dopamine axons, the same step gives 14 spikes, and still 12 at 10 minutes. With a D1 antagonist in the bath, the counts are 8 and then 8. With a PKA inhibitor, they are 13 at first and back to 8 by 10 minutes. What do you conclude, and what control is missing?

> **One way to do it.** 1. The current step is matched, so the extra spikes do not come from more input: the cell’s intrinsic excitability rose. Blocking synaptic receptors limits hidden synaptic drive, but fast-receptor blockers alone leave slow GABAB and muscarinic signaling in play; Lahiri and Bevan blocked those too.
> 2. The antagonist abolished the change, so D1 receptors are required, assuming the drug is selective at the dose used.
> 3. With PKA inhibited, the early increase (13 spikes) remained but did not last, so in these data PKA is needed for persistence, not onset. The data cannot say what carries the onset.
> 4. Missing: current-only trials, the same steps given over the same time with no light. Subtract drift: the dopamine effect is (after minus before, with light) minus (late minus early, without light). Lahiri and Bevan corrected with matched current-only trials.
> 5. Keep perforated patch, since whole-cell dialysis can wash out the messengers and hide the effect.
> 6. Limit: a spike count names no channel. To find the target, test whether selective channel blockers mimic or occlude the dopamine effect, as Lahiri and Bevan did. Voltage clamp is harder here, because perforated patch gives poor voltage control.

## Going deeper

### The balance point with injected current

If an electrode injects a steady current Iinj, counted as positive when it depolarizes, the balance point becomes

> **Vss = (gL × EL + gE × EE + gI × EI + Iinj) / (gL + gE + gI)**

At steady state, a sustained current then changes the voltage by ΔV = Iinj / (total conductance), the current times the input resistance. The same bottom line appears, so a shunt that doubles the total conductance, as the 10 nS shunt does in the resting cell, halves the steady response. During a step the voltage approaches that value with τeff: ΔV(t) = (Iinj / total conductance) × (1 − e−t/τ\_eff). So a brief response is cut far less: 1 ms into a step, the shunted response is about 98% of the unshunted one.

### Divide or subtract?

For steady current steps the passive model’s shunt divides, as above, by the total conductance before over the total conductance after. For synaptic excitation the factor depends on the excitation too: 15/25 = 0.60 with 5 nS, but 20/30 = 0.67 with 10 nS. Real cells keep the logic but not always the arithmetic. In layer V pyramidal neurons from juvenile rat somatosensory cortex, Ulrich (2003) used a dynamic clamp (a feedback system that injects the current a chosen conductance would pass) to add a shunt. It divided the subthreshold voltage responses but subtracted from the firing frequency. Do not assume either rule in other cells.

### The model behind the fast-EPSP warning

The model used this lesson’s cell (C = 200 pF, gL = 10 nS, EL = −65 mV, threshold −50 mV) with a fast, alpha-shaped excitatory conductance (E = 0 mV, time constant 1 ms, peak 19 nS, a size chosen to sit near threshold) and a steady 10 nS GABA conductance with EGABA = −55 mV. The excitatory conductance peaked at 1 ms; the voltage peaked later, at about 4 to 5 ms, and then decayed slowly. Alone, the excitatory postsynaptic potential (EPSP) rose about 12.3 mV and peaked at about −52.7 mV, short of threshold. With the GABA conductance open beforehand and still open, it rose only about 10.1 mV but peaked at about −49.9 mV, just past threshold: the head start outweighed the 2.2 mV loss in EPSP size. With EGABA at rest instead, a true shunt, the same open conductance made the EPSP smaller (peak about −54.1 mV).

### Keeping the cell’s own chloride

Gulledge and Stuart used gramicidin perforated-patch recording, which leaves the cell’s own chloride in place because gramicidin pores do not pass chloride. A conventional whole-cell pipette replaces the cell’s chloride with the pipette solution, so EGABA then reflects the pipette recipe rather than the cell.

### Chloride loading and where inhibition works best

In branched-dendrite computer models, Currin and Raimondo (2022) found that activity-dependent chloride loading weakened inhibition, and that with chloride dynamics included the most effective placement of inhibitory synapses lay close to, but not at, a shared branch junction. These are model predictions, not measurements.

### When dopamine’s effect begins

In Lahiri and Bevan’s recordings, the rise in firing had a median onset of 458 ms by their criterion.

---

From STEMhop (https://stemhop.com), free lessons by Christopher Deppmann. Lesson text © the authors. Practice questions: https://stemhop.com/lessons/synaptic-integration/
