Membranes and membrane transport

Everything a neuron does, every thought a brain-computer interface will ever read, comes down to charged particles sitting on the wrong side of a thin greasy sheet. That is not a metaphor. The electrical signal an electrode records is ions moving across a membrane. So before the BCI track can record anything, you have to understand what that membrane is, why it blocks almost everything, and how the cell turns a simple concentration difference into a battery it can fire on command. This lesson builds that foundation from the one rule you already know.

A cell needs a boundary, and grease makes one for free

Recall the hydrophobic effect from S2.2. Drop a crowd of amphipathic molecules into water (molecules that are water-loving on one end and oily on the other) and water's preference for its own company forces them into the one arrangement that keeps every oily part dry and every water-loving part wet. For a phospholipid, which has a charged head that loves water and two oily tails that hate it, that arrangement is a double sheet: tails tucked inward facing each other, heads facing the water on both sides.

That double sheet is the membrane. It is about 5 nanometers thick, two molecules deep, and it wraps every one of your cells and most of the compartments inside them. Nothing assembled it. As S2.2 insisted, there is no builder here, just water and grease settling into their lowest-energy shape. Drop the pieces in and the boundary appears.

One property matters later: the sheet is a fluid, not a solid wall. The lipids are not glued in place. They drift sideways past each other constantly, so the membrane behaves more like a two-dimensional liquid than a brick, and the proteins embedded in it float around and cluster where they are needed.

Why the bilayer blocks almost everything

Now look at what the membrane keeps out, and derive it rather than memorize it. The middle of the bilayer is a layer of pure oil, the packed tails, with no charge and no water. Ask what it takes for something to cross: it has to dissolve into that oily core and out the other side.

A charged ion, say a sodium ion or a potassium ion, is the opposite of oily. It carries a full charge and drags a shell of water molecules around it (that is what hydrophilic means). Asking it to enter the greasy core is asking water and charge to move into a place that repels both. It essentially will not happen. The same hydrophobic effect that built the membrane now guards it. A bare bilayer is, for practical purposes, a perfect insulator against ions.

Small uncharged molecules are a different story. Oxygen and carbon dioxide are tiny and nonpolar, so they slip through the oily core with ease. Water itself, though polar, is small enough to leak across slowly. The rule that falls out is not "blocks everything" but selective permeability: a few small uncharged molecules cross freely, while ions and larger polar molecules are shut out unless something ferries them. That "unless" is the whole game.

The controlled crossings: channels, pumps, receptors

If the bare sheet blocks ions, then every controlled crossing is a protein embedded in the membrane. There are three kinds you need, and they do genuinely different jobs.

A channel is a protein that forms a water-lined tunnel straight through the membrane. It gives an ion a hydrophilic path so it never has to touch the grease. Channels are selective, a potassium channel passes potassium and not sodium, because the tunnel is shaped to fit one ion and not another. And channels are gated: they open and close in response to a trigger, a voltage change, a molecule binding, or a physical tug. When a channel is open, ions flow through it passively, meaning they move on their own down their gradient, from where they are crowded to where they are sparse. The channel spends no energy. It is a tap. Open it and things flow downhill, fast, up to millions of ions per second.

A pump does the opposite and it is not free. A pump grabs ions and carries them against their gradient, from where they are sparse to where they are already crowded, which is uphill and never happens on its own. To do that it burns energy, usually a molecule of ATP (you will meet ATP properly in S9.2). The famous one is the sodium-potassium pump, which spends one ATP to shove three sodium ions out of the cell and haul two potassium ions in, both against their gradients, over and over. Pumps are slower than channels because each cycle costs fuel. Their job is to build the gradients in the first place.

A receptor is the third kind, and it usually moves no ions at all. It sits in the membrane with a pocket facing outward, and when the right signal molecule from outside settles into that pocket, the receptor changes shape and passes that information to the inside of the cell. It is a sensor, not a door. We will spend all of S9.3 on what happens next. Here, just file it: the membrane is not only a barrier and a set of gates, it is also the cell's sensory surface.

Here is a programmer's way to hold the three. The membrane is a firewall, a default-deny boundary. Channels are ports that are open or closed by a gating rule, and traffic flows through an open one on its own. Pumps are gateways that actively carry payloads the wrong way up a hill and must burn fuel to do it. Receptors are the inbound listeners that sense a message without letting the sender across.

Now the failure edge, because an analogy without its limit is a bug. A firewall is a policy you configure and can flip with a keystroke, and a network port does not care which way bytes "want" to flow, there is no downhill for a packet. The membrane is nothing like that. Its default-deny is not a setting, it is the physics of grease and water, and it cannot be reconfigured, only drilled by proteins that evolution shaped over eons. And direction here is not arbitrary policy: through an open channel, an ion can only ever flow down its gradient, never up, unless a pump pays for the climb. The rules are thermodynamics and an energy budget, not a config file.

A gradient is stored energy: the membrane as a charged capacitor

Now the idea the entire BCI track rests on. When a pump spends ATP to pile sodium up outside and potassium up inside, it is not just sorting ions. It is storing energy. A concentration difference held across a barrier is potential energy, in the same way water pumped uphill behind a dam is potential energy. The pumping cost something. The difference, once built, sits there ready to do work the instant a path opens.

For a charged particle the storage is even richer, because the ions carry charge, so a gradient of ions is also a separation of charge. Picture the membrane: two salty, conductive solutions (inside the cell and outside) separated by a thin insulating sheet of oil. That is the exact recipe for a capacitor, two conductors with a thin insulator between them. The membrane is not merely like a capacitor. Electrically it is one, and electrophysiologists measure its capacitance as a real number, roughly 1 microfarad per square centimeter. The separated ions are the stored charge. The insulating oil core is the dielectric.

So a gradient stores energy in two coupled forms at once: a chemical part (the concentration difference) and an electrical part (the charge difference, which shows up as a voltage across the membrane). Together they are called the electrochemical gradient. Build it with pumps, and you have charged a capacitor.

The payoff is what a channel does to a charged capacitor. Open a channel and you open a conductive path. Ions rush down their gradient through it, and moving charge is electric current. That current is work being drawn from the store the pump filled. A neuron does not generate its signal from nothing at the moment it fires. It spends, in a fraction of a millisecond, a charge that its pumps spent minutes patiently building. Charging costs ATP. Discharging is free and fast. That asymmetry is the whole trick.

The capacitor analogy is strong, so mark where it slips. A textbook capacitor separates electrons across a solid dielectric and no matter actually crosses. Here the "charge" is whole ions, real chemical particles, and a chemical concentration term rides alongside the electrical one, which a plain capacitor does not have. So the membrane is a capacitor and a chemical battery fused into one object, together the electrochemical gradient. Keep the capacitor picture for the intuition (thin insulator, separated charge, current when a path opens) and remember there is chemistry stacked on the physics.

From a gradient to a signal (the tease)

Here is where it is heading, kept deliberately short. A resting neuron sits with its inside about 70 millivolts negative relative to its outside, written -70 millivolts. That voltage is not magic. It exists because the sodium-potassium pump built ion gradients (potassium concentrated inside, sodium outside, each by more than tenfold) and a handful of channels let the membrane "feel" those gradients as a steady voltage. The capacitor is charged and idling.

When the neuron fires, voltage-gated channels snap open in a rush, ions pour down their gradients, and the membrane voltage swings hard and fast before pumps and other channels reset it. That swing, traveling down the cell, is the action potential. It is the unit of information in the nervous system, and it is nothing but the choreographed discharge of the membrane capacitor you just built. S9.3 and the BCI track pick up the story from here.

The diagram below is that whole pipeline in one line, from the action potential you just met to a decoded command. Trace it left to right and notice how much real work sits between a neuron and an output, and that the front end (teal) is pure membrane physics while the back end (violet) is compute.

Each stage is real work: the front end (teal) senses and conditions a tiny analog signal, and the back end (violet) reduces it to features and a trained decoder that emits a command. The decoder is a classifier, not mind reading.

Key terms

phospholipid bilayer

A double sheet of amphipathic lipids that self-assembles in water with oily tails hidden inside and water-loving heads facing out, forming the boundary of every cell.

selective permeability

The property of a membrane that lets a few small uncharged molecules cross freely while blocking ions and most polar molecules unless a protein carries them.

ion channel

A membrane protein forming a gated, water-lined pore that lets a specific ion flow passively down its gradient when the gate is open, spending no energy.

pump (active transport)

A membrane protein that spends energy, usually from ATP, to move ions against their gradient, which is how the gradients get built and maintained.

electrochemical gradient

The combined concentration difference and charge difference of an ion across a membrane, which together store usable potential energy.

membrane potential

The voltage across a cell membrane, about -70 millivolts inside a resting neuron, produced by ion gradients acting through selective channels.

receptor

A membrane protein that detects a signal molecule on the outside and passes information inward without necessarily letting that molecule cross.

Why the pump moves an unequal 3 for 2, and why that is not the main point

The sodium-potassium pump exports three sodium ions and imports only two potassium ions per ATP, so each cycle nets one positive charge carried out of the cell. That unequal trade contributes a small direct amount to the negative resting voltage, and it is a real effect. But do not overweight it. The bulk of the resting potential comes from the gradients the pump builds combined with the membrane being far more permeable to potassium than to sodium at rest, so potassium leaking down its gradient sets most of the voltage. The pump's headline job is not that little 3-for-2 charge imbalance, it is being the tireless charger that keeps the gradients topped up against constant leak. Stop the pump and the gradients slowly bleed away, and with them the voltage. The battery only holds charge because something keeps paying to refill it.