I first encountered Michael Chorost in his fine book World Wide Mind (Free Press, 2011), which looks at the relationship between biology and the machine tools that can enhance it. Mike’s thinking on SETI has already produced rich discussion in these pages (see, for example, SETI: Contact and Enigma). In today’s essay, he’s asking for reader reactions to the provocative ideas on insect memory and intelligence that will inform his next book. While it does not happen on Earth, can evolution invent — somewhere — a social insect society capable of long-term memory and civilization? A nearby planet evidently hostile to our kind of life offers fertile ground for speculation.

by Michael Chorost


I’ve admired Paul Gilster’s Centauri Dreams for many months and I’ve always been impressed by the quality of the comments. Paul graciously allowed me to write a guest entry to test one of my book-in-progress’s ideas on a smart audience — you.

This book-in-progress will be my third book. My first two books were about bionics and neuroengineering, respectively titled Rebuilt (Houghton Mifflin, 2005) and World Wide Mind (Free Press, 2011.) I’ve also published in Wired, New Scientist, Slate, Technology Review, the Chronicle of Higher Education, and Astronomy Now.

The book is about communication with extraterrestrials. Not by radio but in person, with us visiting their planet and looking at their mugs (or whatever they have instead of mugs). How should we begin trying to communicate? What could we safely assume — and not assume — about how minds think? What knowledge could we bring to bear from evolutionary theory, linguistics, cognitive science, and computer science?

Of course, direct contact anytime soon is unlikely in the extreme. That’s why, below the surface, the book is about a deeper set of issues: What are the universals of thought and language? Can intelligent minds be so different as to render communication impossible? What kinds of advanced cognition can an evolutionary process invent? The book gets at these ideas by using alien communication as a vehicle.

So here’s the idea I want to test on you all. I asked myself, “Would it be possible for social insect colonies on some other planet to evolve to have language and technology – in other words, a civilization?”

Of course, the idea of swarm intelligence, or hive-mind intelligence, has been around forever in science fiction. To give but one example, Frank Schatzing’s The Swarm posits an undersea alien made of single-celled, physically unconnected organisms that collectively have considerable intelligence. But I need to examine the idea with much more rigor than can be done in fiction.

I refined the question by deciding that, as on earth, the individual insects would have brains too small for serious cognition. The unit of analysis would not be individual bugs but colonies of bugs. The intelligence would have to emerge from their interaction.

After much thought, my answer to the question is “No – but…”

Let me explain both the No and the but. It is these explanations on which I want your feedback.

To start with the No. I don’t think it’s possible for physically separate units to form a collective that supports high intelligence. The reason is straightforward: physically disconnected units have no way of permanently storing large amounts of discrete information in a way that is available to the collective. More succinctly, they can’t support long-term memory.

Of course, information can be manipulated by collectives even when the units have no permanent connections among them. If you’ve read Douglas Hofstadter’s “Ant Fugue” you know how ant colonies collectively find and consume food. A forager comes across food and lays a pheromone trail while returning to the nest. Other workers follow that trail and lay down pheromones of their own. When the food is gone the returning ants stop emitting pheromones, and the ants move on to other things. From a global perspective it looks as if the colony has a “memory” of the food source. Insect colonies have many mechanisms of this sort, which go under names like “stigmergy” and “quorum sensing.” They are brilliantly described in the literature, especially by Thomas Seeley. [1] But all of them yield only short-term memory. As soon as the insects disperse and the pheromone evaporates, the information vanishes.

That is a problem, because language and other forms of advanced cognition need long-term memory. Language requires storing a large number of primitives (e.g. words) plus state information related to a conversation (the identity of the interlocutor, the situation, information about past and future, and so forth.) Not only that, the method of storage has to be both stable and easily changeable. If it can’t be changed, an intelligence can’t keep up with changing events in the world.

Let me pause here to define what I mean by “intelligence” and “language.”

I like the definition of intelligence offered by Luke Muehlhauser in his book Facing The Intelligence Explosion. [2] He defines it as “efficient cross-domain optimization.” Cross-domain optimization refers to being able to exercise intelligence in multiple domains. Consider that IBM’s Deep Blue program is very smart at chess but can’t play checkers, let alone want to learn how. It has intelligence in one domain, and only one. Or take honeybees, who are outstanding at communicating the location of food but have no way of asking humans to move that food closer, or change it. In order to cross domains a mind needs not just cognition but metacognition, the ability to think about thinking. When I speak of intelligence I mean the kind that can reflect upon its own actions, make plans, describe things that don’t exist, and so forth. This is the kind of intelligence that is required to build a civilization.

Now language. I like Steven Pinker’s definition of it: Language is a finite set of primitives that when combined yield an infinite number of possible statements. [3] By this definition, language is open-ended. It can be used to say anything. Contrast that to, say, referee signals in baseball. They are a communication system but not a language. A referee can precisely say whether a pitch is a ball or strike, but he can’t use the repertoire of signs to talk about taxes, or explain that the pitcher has just become a free agent. Likewise, a honeybee can precisely state where food is but can’t use its waggle dance to discuss the weather with a human. Animals such as honeybees, birds, chimps, dolphins, parrots, and dogs all have communications systems, some of which are very sophisticated, but they are closed-ended; they do not rise to the level of language.

Now that I have defined intelligence and language, please note that both of them simply have to have long-term memory. Without long-term memory, no intelligence, no language. And I don’t think there is any way at all that a social insect colony can get long-term memory if its units are physically disconnected. It has no physical medium in which it can store information in a way that is both permanent and easily changed.

So, Conclusion A: Social insect colonies do not have the memory mechanisms to support language, therefore no bug civilizations.

Now let’s get to the “but.” After working out Conclusion A I asked myself, “Could insect colonies acquire, through an evolutionary process, a mechanism of long-term memory?” I think the answer to that question is yes.

Consider how mammalian brains store long-term memory: in collections of synapses. A synapse is a physical gap between the axon of one neuron and the dendrite of another. Depending on the strength of an incoming signal and the synapse’s threshold, neurotransmitters either flood into that gap or they don’t. If they do, they are picked up – essentially “smelled” – by chemoreceptors on the dendrite’s side. Then the signal continues to the body of the next neuron, which uses it as an input for its own decision-making process.

Each neuron in a mammalian brain has thousands of synaptic connections to other neurons – it is part of an immense network of physically connected units. By changing synaptic configurations and thresholds, neurons can encode immense amounts of discrete information. That information is both stable and easily changeable.

So for an insect colony to gain long-term memory, it has to invent the equivalent of the synapse. Not in the brains of individual insects – they already have plenty – but on the level of the colony as a whole, using interactions between insects.

This is obviously tricky because insects move around. But there are insects in colonies that don’t move around: the larvae. Even better, in flying insect colonies they are generally stored in honeycombs that keep them in place. And better still, they’re loaded with chemoreceptors. The ends of antennae and feet are the “noses” by which insects pick up smells.

Imagine, then, the antennae and feet of developing larvae thrusting their way through the waxen walls of honeycombs and making contact with the antennae and feet of their neighbors. Right there you have the basic elements of synaptic connections. If the larvae can send signals and adjust synaptic thresholds, they could form a network.

Of course, there has to be an evolutionary reason why such a network would ever come into being. There would have to be accidental variations that create primitive networks, and they would have to confer fitness and reproduction benefits.

So consider this story of an evolutionary process. It so happens that in some kinds of colonies, the larvae perform a digestive function for the colony. The workers bring them the food that they can’t digest, and the larvae break them down into compounds the workers can eat. [4] So the larvae are effectively the colony’s stomach. The food needs of workers vary depending on temperature and season and so forth. Larvae that could exchange information with other larvae about digestion could produce better food, and that benefit would tend to be conserved and amplified. Over many generations, then, colonial stomachs could evolve into colonial brains. Each larva would be a large neuron with many connections to other larvae, and the synaptic configurations between them would store long-term memories.

This is, of course, a just-so story – but then evolution is full of just-so stories of evolutionary adaptations that seem spectacularly improbable. For example, insect wings are thought to be adapted legs. [5] And insects have often evolved to look like leaves and twigs for camouflage. Nature is astonishingly inventive at reshuffling its building blocks. I am not trying to convince you that my larvae-to-brains story is likely, only that it is possible.

There is one more piece to the puzzle. Long-term memory is metabolically and spatially expensive. Clearly, on Earth insect colonies have seen no need to develop it; they’ve done well for millions of years without it. So you need to have an environment in which it would confer fitness advantages.

Consider the planet GJ832c.

GJ832c is a rocky planet of 5.4 Earth masses orbiting a red dwarf star sixteen light-years away. Happily, it’s in the star’s habitable zone. Since a red dwarf is very dim the habitable zone has to be very close to it, and accordingly GJ832c has a year just 36 days long. [6]

We don’t know much about GJ832c. We don’t know its density, so we don’t know its surface gravity. But I’ve guessed that it’s 78% as dense as Earth, which would give it a surface gravity of 1.5 gees. We don’t know its rotational period, but since it’s so close to its star it would probably be gravitationally locked. Mercury has a 3:2 spin-orbit resonance, which means that it rotates three times every two years. So let’s say that GJ832c also has a 3:2 spin-orbit resonance.

We don’t know its axial tilt, but gravitational locking tends to stabilize axial tilt near zero – Mercury’s is just two degrees, and the Moon’s is 6.6 degrees, compared to the Earth’s 23 degrees. So let’s say its axial tilt is zero. But we do know its orbital eccentricity, .18, which is very eccentric by our solar system’s standards.

If you put these facts and guesses together you can compute how much solar exposure each point on such a planet gets, like so:


Astonishingly, on such a planet the climate is determined as much by longitude as latitude. Yes, longitude. Some longitudes are in daylight for long periods, while other longitudes never see the sun at all – including a few points on the equator. The planet looks like a tennis ball with burns on opposite sides (red), a temperate zone ringing the burns (yellow), and ice everywhere else (blue). [7]

To be sure, the temperature extremes would be moderated by the atmosphere. My guess is that you would see Hadley cells centered on the hot zones, since the hot air would rise and cold air would come in underneath it. Since the planet rotates so slowly, you wouldn’t see much Coriolis force to shear the atmosphere sideways. So there would probably be steady winds moving toward the center of each hot zone, distributing heat between the zones.

Note, however, that only one “hot” end can face the star at any given moment. The center of each hot zone would face the star continuously for blistering days on end, and then suffer a long night. (I haven’t worked out what the day-night cycle would look like on various points of the planet. Perhaps the temperate zones would be in continuous but relatively soft illumination. For this I need the help of someone who specializes in orbital dynamics.)

In any case, GJ832c would be a nasty planet. It’d have high gravity, temperature extremes, constant wind, and possibly a thick atmosphere and ultraviolet flares from its star. I don’t think you would get large-brained mammals here simply because of the gravity: blood circulation and locomotion would be expensive. Predator actions like leaping and throwing things would be difficult. So would prey defenses like running and climbing.

What would flourish here? Bugs. Bugs are modular, tough, and cheap. They are small enough to be relatively unaffected by gravity, and their chitinous exoskeletons would be relatively impervious to UV flares.

So let’s say that social insect colonies evolve in the temperate zone. But the temperate zone is exceedingly narrow, perhaps just a few hundred miles across. Sooner or later population pressures are going to drive new colonies into the hot and cold zones. There, new colonies could find resources that aren’t in the temperate zone, say particular kinds of hothouse flowers, lichens, and fungi. And they would face new scarcities too, say of water.

On GJ832c, colonies that learned to trade resources across zones would have an enormous survival advantage. Water for nectar, nectar for fungus, and so on. Insects on Earth have signaling mechanisms that could be adapted to manage such trades. For example, they engage in territorial displays in which soldiers posture at the borders between colonies, inflating their limbs to seem more threatening, while “head-counting” ants on each side carry information about the enemy back to the nest. (They probably don’t actually count the soldiers using numerals; more likely they sense the rate of encounters with them.) [8] Such signaling mechanisms could be adapted to convey information for economic exchanges. Colonial brains would store such information, remembering who traded what and for how much. Over many generations, such signaling systems could evolve into language.

You may wonder about tools, since tools have fundamentally shaped the development of language in humans. For brevity’s sake I won’t go into it here, but I’ve worked out how insect colonies could ignite fire, forge metals, and use tools; again, I’ve extrapolated from things social insects do on Earth. With language and tools a species is just a few hops, skips, and jumps away from having a full-fledged civilization.

This doesn’t mean they would think like humans, of course. They would have networks that can support long-term memory, but those networks would have a very different organization and would support very different kinds of physical needs. In the manuscript I discuss the role of simulation and embodied cognition on the development of language.

So, Conclusion B: With the right environmental pressures, social insects could develop long-term memory, language, tool use, and a civilization.

Again, I am not arguing that this is likely, only that it is possible. What do you think? Am I correct in thinking it is possible, or is there something fundamental that I am neglecting?

I’m asking you to put pressure on these ideas. To look for their weak spots. But I would also appreciate it if, for each weak spot, you could suggest a solution, if you can think of one.

Many thanks in advance for your comments and ideas.



1. Seeley, Thomas (2010). Honeybee Democracy, Princeton.

2. Muehlhauser, Luke (2013). Facing The Intelligence Explosion. Machine Intelligence Research Institute. Kindle location 655.

3. Pinker, Steven (2007.) The Language Instinct: How the Mind Creates Language. Harper Perennial, p. 75.

4. Masuko, Keiichi (1986). “Larval hemolymph feeding: a nondestructive parental cannibalism in the primitive ant Amblyopone silvestrii Wheeler (Hymenoptera: Formicidae).” Behav Ecol Sociobiol 19: 249-255. See also http://blog.wildaboutants.com/2010/06/21/question-1-ant-digestion/.

5. Carroll, Sean (2006). Endless Forms Most Beautiful: The New Science of Evo Devo. Norton, p. 176.

6. Planetary Habitability Laboratory data for GJ832c, http://www.hpcf.upr.edu/~abel/phl/hec_plots/hec_orbit/hec_orbit_GJ_832_c.png

7. Brown et al. (2014). “Photosynthetic Potential of Planets in 3:2 Spin Orbit Resonances.” International Journal of Astrobiology 13:4 (279-289). Page 284. I’ve used the figure computed for an eccentricity of 0.2, which I figure is close enough.

8. Hölldobler, B., and Wilson E. O. (2008). The Superorganism: The Beauty, Elegance, and Strangeness of Insect Societies. New York: W. W. Norton, p. 306.