A Boundary Is Something You Keep Doing, Not Something You Have

How is the resting membrane potential maintained?

We picture a boundary as a wall. Build it once and it stands: the membrane around a cell, the skin around a body, the line between inside and outside. For a class of boundaries that matters a great deal in biology, this picture is wrong in an instructive way. The boundary is not a thing the system owns. It is a thing the system keeps doing, moment to moment, at a cost, and it disappears the instant the doing stops.

Take the clearest case, the electrical boundary of a living cell. A cell holds a steep difference across its membrane: potassium concentrated inside, sodium outside, a voltage of a few tens of millivolts. That difference is a real boundary, a sharp line between two chemical worlds a few nanometers apart. It is also leaky. Ions slip back down their gradients through channels in the membrane, and on their own the two worlds would mix until the difference vanished. They do not mix, because a molecular pump runs continuously against the leak, throwing sodium out and hauling potassium in, paying for each cycle with a molecule of ATP. Jens Skou identified this pump in 1957 and shared a Nobel Prize for it. Cut the cell's energy supply and the pump halts, the leak wins, and the gradient collapses toward equilibrium. That collapse is one of the ways we define the moment of death.

So this boundary has three features a wall does not. It costs energy every second it exists. It depends on that cost, so knocking out the pump dissolves it. And when it dissolves it does so on a clock set by how fast the leak runs. The boundary is a process with a price, not a structure you can walk away from.

Is the boundary of a cell a static barrier or an active process?

The point generalizes past chemistry. Karl Friston (2013) describes the boundary of any living system in statistical terms, as a Markov blanket: a set of states that screens what is inside from what is outside, so that the interior can only feel the world through the blanket. The blanket is the system's boundary in the most general sense. And it too is not handed over once. It is held in place by the system's own ongoing activity, the same activity that keeps the system in its viable states. Stop that activity and the statistical separation blurs. The interior stops being an interior.

Why does the cell gradient dissolve at death?

Underneath both cases sits a law. A difference held against its surroundings is order, and order kept away from equilibrium is exactly what the second law of thermodynamics works to erase. You are allowed to keep it, but only by paying continuously, by pushing energy through and dumping the resulting disorder into the environment. Schrödinger said in 1944 that life stays alive by feeding on "negative entropy," and later corrected himself to say the real currency is free energy. Prigogine built the physics of such maintained structures, the ones that exist only while energy flows through them. A living boundary is one of these. For a leaky difference held far from equilibrium like this one, the fact that it persists means work is still being done.

There is a tempting next step, and I will only gesture at it. Boundaries that are not chemical gradients seem to behave the same way: the felt edge of your own body, the line an immune system draws between self and foreign, the boundary of a group or an institution. They too look maintained rather than given, and they fray when whatever maintains them is disrupted. Whether they are active boundaries in the same precise sense, with a real cost and a dissolution clock, or whether some are just stored cheaply and statically, I leave open. Nothing above depends on it.

The claim is testable where it counts. If a boundary is actively maintained, then taking away the maintenance should dissolve it, on a schedule set by the leak, and the cost of holding it should be measurable as the throughput required. A boundary that just sits there forever after you remove its upkeep, at no energetic cost, is a static wall, and for that one the whole framing is idle and should be dropped. The membrane gradient is not such a boundary. It is held, at a price, against a leak, and it goes when the price stops being paid.