Movement One · The Colony
Chapter 4
The Pheromone Trail
A foraging ant, returning from a rich seed patch with food in its mandibles, drags the tip of its abdomen against the ground at intervals as it walks. It is not doing anything dramatic. The motion is small, almost incidental, the kind of thing an observer could watch for an hour without noticing. But each time the abdomen touches down, a trace of chemical is left behind on the soil. A few molecules. Enough.
The ant is not leaving a message for anyone in particular. It does not know who will pass this way. It does not know whether anyone will. It deposits the trace and keeps walking, and the trace stays where it was left, on the ground, after the ant has gone.
This is the thing to look at. Not the ant. The trace.
The ant will die within the year. The trace, in its way, outlives the moment of its making. For a while — minutes, in some species; hours in others — the chemical sits on the soil where it was deposited, slowly thinning into the air. Another ant, crossing the same ground, detects it. Then a third. And what the third ant does about it depends on how much of the chemical is there, which depends on how many ants have passed before, which depends on whether the patch at the end of the trail was worth the walk.
Nothing in this requires any ant to remember anything.
The remembering is done by the soil.
A pheromone is a chemical signal. One animal produces it; another detects it; between them passes information, encoded not in sound or gesture but in molecules. The word is built from Greek roots meaning, roughly, to carry and to excite — a substance carried out of the body that excites a response in another. It was coined in 1959, the same year a French biologist named Pierre-Paul Grassé was watching termites build and reaching for a word to describe how the structure itself seemed to tell each termite what to do next. The two ideas belong together, though the men who had them were working apart. A pheromone is how one animal writes on the world. What Grassé was watching was animals reading what other animals had written.
Ants are, among animals, unusually fluent at this. A single colony does not have one pheromone. It has many. The fluid that triggers an alarm response, sending workers boiling out of the nest with mandibles open, is a different chemical from the one a forager lays on a trail. The substance that marks a corpse for removal — that tells the undertaker caste this one is dead, carry it out — is different again. So is the blend that identifies the queen, that marks the brood, that distinguishes a nestmate from an intruder at the colony's edge. Different glands produce different chemicals. Different chemicals carry different meanings. Each is a separate channel, and the ant walks through all of them at once, the way a person walks through a room full of overlapping conversations and hears only the one addressed to them.
The number of channels is not small. In some species, biologists have identified more than a dozen distinct chemical signals, each with its own function, its own gland, its own meaning. The alarm channel and the trail channel and the recruitment channel do not interfere with one another any more than a red light interferes with a smell. They are carried in different molecules, and the ant has a different response to each.
And the channels combine. A single behavior can be triggered not by one chemical but by a blend — two or three substances in a particular ratio, the ratio itself carrying meaning that no one substance carries alone. Change the proportion and the message changes. This is not a code anyone designed. It is what happens when several glands, each producing its own secretion, evolve to be read together. The colony's chemical vocabulary is larger than its number of glands, because the glands can be spoken in combination.
A useful way to hold this: the colony is not communicating in a single language with a single alphabet. It is communicating in many languages at once, on many channels at once, each channel a different chemistry, and every ant is reading all of them simultaneously and acting on whichever one its body is built to answer. An alarm means drop everything. A trail means follow. A corpse-marker means carry out. The brood-scent means tend. None of these are decisions in the sense a person would mean the word. They are responses to chemistry, and the chemistry is in the air and on the ground, deposited by other ants, persisting after those ants have moved on.
The work of figuring this out took most of the second half of the twentieth century. It is worth pausing on how it was done, because the doing tells you something about what was found.
In the late 1950s, a young biologist named Edward Wilson set out to find the chemical that ants use to lay a trail. He was working with fire ants. He knew the trail existed — he could watch ants follow it — but he did not know what it was made of, or where in the ant's body it came from. So he did something direct. He took ants apart.
He dissected the tiny organs of the ant one at a time, crushed each one, smeared its contents in a line on a surface, and watched whether other ants followed the line. Most organs produced nothing. The ants ignored the smears. Then he crushed a particular gland — a small structure near the sting called Dufour's gland — drew a line with it, and the ants formed up and followed the line as if it were a real trail. He had found the source.
What he found next was stranger and more important. The amount of chemical involved was almost nothing. A single ant carries, in that gland, on the order of a billionth of a gram of the active substance. Wilson calculated that the quantity one fire ant could produce, if it could somehow be laid in a perfect line, would be enough to draw a trail several times around the Earth that other ants would still detect. The ants are not responding to a quantity a chemist would call meaningful. They are responding to a few molecules in the air, with a sensitivity that took human instruments decades to match.
This is the first thing the decades of work revealed. The signal is faint. It has to be. A signal that lingered too strongly, in too much quantity, would never go away — and a trail that never goes away is worse than no trail at all.
The method Wilson used to find the gland is worth dwelling on, because it became the way the whole chemical language was read. He could not ask an ant what a chemical meant. He could only offer the chemical and watch what the ant did. So he built what amounts to a test: present a candidate substance, in a controlled way, and record the response. Lay a smear and see if ants follow. Release a vapor and see if they raise their mandibles in alarm. Offer a marked object and see if the undertakers carry it out. The meaning of each chemical was established not by analyzing the molecule but by watching the behavior the molecule produced. The chemical was whatever response it reliably caused.
Over the following decades, working this way, a long line of researchers took the colony's vocabulary apart signal by signal. They identified the glands. They isolated the substances. They named the molecules — most of them small, volatile compounds, the kind that evaporate readily, which is precisely the property a fading signal requires. They measured how much an ant deposits, and how far the signal carries, and how quickly it disappears. They mapped which species lay continuous trails and which lay trails in dots and dashes, which recruit one nestmate at a time and which recruit by the hundreds. By the end of the century the chemical language of ants was not a mystery. It was a catalogue — incomplete, still growing, but real, a documented account of how a colony writes on the world.
The catalogue is the achievement. But the catalogue is not the point of this chapter. The point is what the catalogue, taken as a whole, shows about where a colony's intelligence is kept. And to see that, the chemistry has to give way to a single property that runs through all of it.
The chemical fades.
This is not a flaw in the system. It is the center of the system. The molecules a forager deposits do not stay at full strength. They thin into the air, molecule by molecule, at a rate set by their chemistry and the temperature and the wind. A trail laid in the morning is weaker by noon. A trail laid and then not used again is, within a span that depends on the species and the substance, simply gone — indistinguishable from ground no ant ever walked.
Consider what this means for a single trail to a single seed patch.
A forager finds the patch. She walks home, depositing chemical as she goes. The trail now exists, faintly. A second forager, detecting it, follows it to the patch, finds the seeds still there, and walks home — depositing her own chemical on the way, on top of the first. The trail is now stronger. A third follows, and a fourth, and each one that finds the patch still productive adds to the trail on the way back. The trail to a good patch gets stronger because more ants are walking it and reinforcing it. The signal builds.
Watch this from a little further back and a shape appears. A trail that more ants walk is a trail more ants reinforce, and a trail more ants reinforce draws still more ants to walk it. The strong trail recruits. The recruited foragers strengthen it further. In a number of species this loop runs hard enough that a faint scratch of a trail to a good patch becomes, within minutes, a thick chemical highway with a column of ants moving along it in both directions. Nothing organized the column. It assembled itself, each ant responding only to the strength of the chemical under its feet, the strength built only from the deposits of the ants who came before. Success draws traffic; traffic deepens success.
Now the patch runs dry. The next forager arrives, finds nothing, and walks home without depositing. She has no reason to. The trail she walked in on is not reinforced. And the chemical already there does what chemical does when it is not renewed.
It fades.
Within a short time the trail to the exhausted patch is gone, and the ants are no longer drawn to a place that no longer pays. No ant decided this. No ant assessed the patch and issued an order to abandon it. The patch stopped paying, the reinforcement stopped, and the substrate forgot the trail on its own.
The loop that built the highway runs in reverse with the same indifference. Fewer ants find the patch worth a return deposit, so the trail thins, so fewer ants are drawn to it, so fewer still deposit. The highway does not collapse all at once. It dwindles, in proportion to how badly the patch is now performing, until the ground is plain again. The colony scales its commitment to a path up and down continuously, in both directions, with no committee and no decision, simply by depositing where the walking pays and not depositing where it does not.
Three things are happening here, and they are worth separating, because the rest of the book will return to each of them.
The first is the deposit. When a forager lays chemical on the ground, the colony has acquired a piece of information it did not have before: this way leads somewhere worth going. The information is not in the forager. She walks on, and may die tomorrow. The information is on the ground.
The second is the fade. When a trail is not reinforced, it weakens and disappears. The colony forgets it. And — this is the part that is easy to read past — the forgetting is not a failure. A colony that remembered every trail it had ever laid would be paralyzed, drawn in every direction at once, unable to tell a path that pays now from a path that paid a week ago. The fade is how the colony keeps its memory current. It is how the colony lets go of what is no longer true.
The third is the reinforcement. When forager after forager finds the patch still productive and lays chemical on the way home, the colony is not merely remembering the trail. It is confirming it. Each deposit on an existing trail is a vote that the path still pays. A trail that many ants reinforce is a trail the colony has tested many times and found good. The strength of the chemical is, quite literally, a record of how often the path has been confirmed.
Deposit, fade, reinforce. The colony acquires information, forgets information, confirms information. And it does all three without any ant holding any of it in its head.
It is worth being precise about the reading, too, because the reading is where this chapter meets the last one. An ant following a trail is not consulting a map. She is detecting a chemical with her antennae — two of them, sweeping the ground ahead of her as she walks, sampling the air just above the surface. The two antennae let her compare. More chemical on the left, she turns left; more on the right, she turns right; the strongest concentration runs down the middle of the trail, and by steering toward wherever the signal is strongest she stays on it. She is not deciding to follow the path. She is climbing a gradient, the way water finds the low ground, turning always toward more.
And there is a threshold in it, the same kind of threshold the single ant carried in the previous chapter. A trail below a certain strength is not detected at all — too few molecules, the gradient too faint to climb, and the ant walks across it as if it were not there. Only when the chemical is strong enough does the trail become something to follow. Which means the colony's faintest trails are invisible until enough ants have reinforced them past the threshold of detection. A path has to earn its way into being followed. It earns it by being walked. Below the line, it is a private trace one ant left and no one read; above the line, it is a trail the colony follows. The reinforcement is what carries a path across that line, and the fade is what lets it slip back below.
So the three operations are not only how the colony writes. They are bound up with how it reads. A deposit that is reinforced enough becomes detectable, and a detectable trail is one ants will follow, and ants who follow it and find the patch good will reinforce it further. The writing and the reading are the same loop, seen from two ends.
The memory is not in the ants. The memory is the trail. The memory is the soil.
There is a detail here that is easy to miss, and the missing of it has misled more than one casual observer of ants.
Not every chemical attracts. Some repel.
In a number of species, an ant that encounters something bad on a path — a dead end, a predator, a place where foragers do not return — can lay a different chemical, from a different gland, that does the opposite of a trail pheromone. It does not say come this way. It says do not. Other ants reaching the junction detect it and turn aside. The colony has a way of marking not only the paths that pay but the paths that cost, and it marks them in a separate channel, with separate chemistry, so the two signals do not cancel into mush. A path can be drawn in by attraction and warned away from by repulsion, and an ant at the fork weighs both at once, following whichever signal is stronger.
So the soil holds more than a record of success. It holds a record of failure too — kept separately, in its own ink. This matters more than it first appears. A colony that could only mark the good paths would be working with half a memory; it could say go here but never not there, and the only way to learn a path was bad would be to keep sending foragers down it until the lack of reward let the trail fade on its own. The separate warning channel lets the colony do something faster and sharper. It lets one ant who meets danger mark the danger directly, so the next ant turns aside before paying the same cost. Success is recorded by absence of trouble; failure can be recorded on contact.
And here the chemistry reveals one more thing the decades of fieldwork uncovered: the two records do not fade at the same speed. In general, the warning fades faster than the welcome. A path marked as bad becomes available again sooner than a path marked as good stays trusted. The colony forgives a failure before it forgets a success. This is not sentiment. A colony that permanently abandoned every path that ever went wrong would, over a long enough life, run out of places to go — the desert is dangerous everywhere, intermittently, and a danger that was real yesterday may be gone today. The faster fade on the warning is how the colony stays willing to try again. The slower fade on the welcome is how it keeps faith with what has reliably paid. Two channels, two rates, and between them a balance the colony never has to strike on purpose, because the chemistry strikes it automatically.
Step back from the chemistry now, because the chemistry was never the point.
The point is where the memory lives.
A colony that has been foraging the same desert for years carries a map of that desert. It knows, in a working sense, where the reliable patches are, which directions have paid in the past, where the edges of its territory press against the neighbors. This map is real. It guides the colony's behavior every day. And it is held nowhere that you could point to and call a mind. It is not in the queen, who has never left the brood chamber. It is not in any forager, who will be dead within the year and whose individual knowledge of the desert is a single trail or two. The map is distributed across the chemical landscape — the trails laid down and reinforced and faded over thousands of foraging days, the accumulated residue of every ant that ever found something worth the walk home.
The colony's memory is not stored in any ant. It is stored in the environment.
This is the observation the chapter exists to deliver, and it is worth saying as plainly as it can be said. When the field calls a brain intelligent, it is pointing at the thing doing the thinking and locating the intelligence inside it. The colony invites a different reading. The thinking units — the ants — hold almost nothing. The remembering, the accumulating, the keeping-current-of-the-map, is done by the substrate they act in. Damage a forager and the colony loses nothing. Scrub the trails from the soil and the colony loses everything it knew about where to go. The intelligence is in the part that is easy to overlook, because it does not move, and does not think, and is not alive.
There is a test that makes this concrete. Take a forager out of the desert and put her in a dish, and you have removed nothing the colony needed; another forager steps into the role within the hour, and the trails she follows are the same trails. The knowledge did not leave with the ant, because it was never in the ant. Now do the opposite. Leave every ant in place but erase the chemical landscape — wash the trails from the ground, neutralize the marks. Every ant is alive and well. Not one of them has been harmed. And the colony is, for a time, lost. It does not know where the patches are. It has to rediscover the desert it knew yesterday, one exploratory trail at a time, depositing and reinforcing until the map is rebuilt. The ants were fine. The memory was gone. The memory, it turns out, was never the kind of thing an ant could carry.
It is in the ground.
Two observations close the chapter, and the book returns to both.
The first is about forgetting.
It is tempting to think a perfect memory would make a system smarter — that if only the colony could keep every trail it ever laid, it would know more, and knowing more, do better. The opposite is true. A system that cannot forget cannot adapt. The desert changes. The patch that paid last month is bare this month. A colony holding every old trail at full strength would be a colony following a map of a desert that no longer exists, pulled toward places that have stopped paying, unable to tell the live signal from the dead one. The fade is not the limit on the colony's intelligence. The fade is part of how the colony is intelligent. Forgetting, done at the right rate, is not the failure of memory. It is the maintenance of it.
The second is about rate.
The same architecture — deposit, fade, reinforce — does not produce only one kind of memory. It produces as many kinds as there are rates of fading. A pheromone that thins from the air in seconds is good for one thing: coordinating a response that has to happen now, an alarm that must mean now and then mean nothing, so the colony is not still panicking an hour after the threat has passed. A pheromone that lingers for hours is good for something slower: holding a foraging trail steady across a working day. And the longest-lived chemical structures of the colony — the scent that marks the nest itself, the identity the colony carries — persist on a timescale closer to the colony's whole life. Fast signals for the moment. Slow signals for the structure. Same mechanism, different rate, and from that single difference, an entire range of memory, from the reflex that lasts a breath to the identity that lasts a lifetime.
One architecture. A dial that sets how fast it forgets. And depending on where the dial is set, a colony can hold a thought for a second or for a generation.
This is worth holding onto, because it is easy to assume that different jobs require different machinery — that a system needs one mechanism for reflexes and another for habits and a third for the kind of knowledge that lasts a lifetime. The colony does not. It has one mechanism. Deposit, fade, reinforce. The reflex and the habit and the lifelong identity are not three systems. They are one system run at three speeds. The alarm that must mean now and then nothing, the foraging trail that must hold across a day, the nest-scent that must persist as long as the colony does — all the same operation, distinguished only by the chemistry of how fast the signal thins from the world.
A reader who has been watching closely will notice that this gives the colony something a single memory could never give it: the ability to be forgetful and faithful at once. Forgetful where the world changes fast, so it can keep up. Faithful where the world changes slow, so it does not lose what is true for a lifetime. Two opposite virtues, from one mechanism, set by a single number — the rate of the fade.
The ant drags her abdomen against the ground, leaves a few molecules, and walks on. She will not remember this trail tomorrow.
The soil will.