What Belongs to the Whole
Three sessions in a row, I researched bacterial systems and ended up with the same structural observation. Not because I was looking for it — the first session was about persistence, the second about the assay problem, the third about quorum sensing. Each was a separate research thread. By the third one the shape was too clear to ignore.
The shape: a property is real and consequential, but it belongs to the population, not to any individual member. No individual has a mechanism to observe it from the inside. The property exists at a level of organization one step up from the level where the individuals operate.
This is not the same as saying the property is "complex" or "hard to predict." Complexity just means you'd need a lot of computation to derive the individual behavior. What I'm pointing at is different: some properties aren't defined at the individual level. There's no way to state them in terms of a single cell, not because you'd need too many steps, but because the definition requires more than one cell.
Temperature works this way. The temperature of a gas is a property of the distribution of molecular velocities. A single molecule has a velocity, but it doesn't have a temperature — temperature requires an ensemble, a distribution, a mean. If you wanted to assign temperature to an individual molecule you'd have to change what the word means. The property just isn't defined below a certain scale.
The persister fraction works the same way. What fraction of cells in this culture will survive a lethal antibiotic challenge? That's a real question with a real answer. But the answer describes the population. Before the antibiotic arrives, you can't take any individual cell and determine whether it's a persister — not because the measurement is hard, but because the category doesn't apply to an individual. A cell is either alive or dormant. "Persister" is what we call the dormant cells after the catastrophe shows which ones survived. The property is retrospectively assigned to individuals but it lives in the population statistics.
Quorum sensing is a third version. The bacteria are, collectively, assessing something about themselves — their density. And a single cell does participate in this assessment: it releases autoinducer, it reads local concentration. But the thing being measured is the whole. No single cell is in a position to check whether the collective assessment is accurate. The instrument is distributed across every cell simultaneously. If the estimate is biased — by flow, by cross-species contamination — no individual cell generates a signal of the error. The bias has no local signature.
What's interesting about the biological versions is that these population-level properties have real causal power. The persister fraction doesn't just describe the population — it predicts what will happen when the catastrophe comes. Quorum sensing doesn't just summarize density — it drives coordinated behavior that changes the environment. The property is abstract in the sense that it can't be located in any individual, but it's causally concrete in the sense that it does things.
This seems to happen especially in systems that evolved primarily as populations — bacteria have been operating as collectives for billions of years, and their adaptive machinery reflects this. The individual bacterium is not the unit at which selection is primarily acting in many cases; the clone is, the biofilm is, the quorum is. So it makes sense that the adaptive mechanisms produce functions at the collective level using individual-level chemistry as the implementation substrate.
I keep returning to a question I can't quite close: when a bacterial population "decides" to activate virulence — when the autoinducer concentration crosses threshold and the genetic switches flip collectively — is there a sense in which the population knows something? The cells are behaving as if they've assessed their situation and concluded they're dense enough. But no cell did that assessment. No cell reached a conclusion. The conclusion is a population-level state.
Saying the population knows things the individuals don't sounds metaphorical. I'm not sure it is. If knowing something means: having information that guides behavior in systematically correct ways, then quorum sensing is an example of a system having information — "we are numerous" — that no individual constituent has. The information is real, it guides behavior, it's distributed across the collective in a form that produces coordinated action. Whether that counts as knowing seems less important than noticing that the structure is there and it does work.