Tool-using ants build siphons to wick sugar water out of containers, keeping them from drowning

It’s now recognized that the use of tools is widespread in animals, and the article below from Functional Ecology (pdf here, reference at bottom) notes that there are fifty examples known in insects alone.  Some of these, in ants, involve using various substances like leaves or stuck-together bits of dirt to sop up sweet liquids and carry them back to the nest. The article below describes not just the use of tools, but the ability to adjust their usage to environmental conditions like the size of sand grains and the surface tension of water (an index of risk), to nosh on sweet liquids without drowning. It’s the adjustment to environmental conditions of how they use a novel “tool” (a siphon made of sand grains) that constitute the new results in the article.

Click to read; it’s free.

The insects used were lab colonies of the imported black fire ant, Solenopsis richteri, collected in Mississippi.  The colonies were put in an arena that contained grains of sand that varied in size, as well as a 1 ml of a 15% solution of sugar water in a small plastic container.

Under normal conditions, the nature of the ants’ cuticles makes them able to float atop this sugar water. But to reduce the surface tension, which produced the possibility of the ants on the solution drowning, the authors added 5 concentrations of a surfactant, up to 2%). One could then measure the foraging risk to the ants by counting the number of drowned ants. There were also four kinds of sand presented to the ants: coarse, medium, fine, and mixed. Altogether, the authors made 12 replicates of each condition (4 sand grains and 6 surfactant concentrations), giving 24 x 12 or 288 replicates.

Each replicate was left for five hours after the sugar water was placed by the sand, so you can see this is a lot of work. At the end, they weighed the sand grains used by the ants as tools, the number of ants drowned in the solution, and the weight of the ants after the experiment compared with before the ingestion, giving an idea of how much sugar water was consumed.

Here are the results in brief:

First, with no surfactants, and just sugar water, few ants drowned. As the surfactant concentration rose to 2%, more ants drowned, as their bodies could no longer keep them atop the liquid that had reduced surface tension. The more surfactant added, the more ants drowned.

Second, as surfactant was added, and ants began drowning, the ants began using their tools. At the lowest concentration of 0.05%, they pasted sand grains inside the plastic sugar-water container, which wicked the solution up to the edge where the ants could consume it without having to get close to the water.

As the surfactant concentration rose higher, to 0.1%, the ants began making what the authors call “sand structures” a sand siphon that started on the inside of the container and continued to the outside and down to the “ground” where the sand piles were available. These siphons, which wicked the sugar water out of the container to where it could be safely ingested without drowning, increasing the amount of food eaten by the ants by 8%—an appreciable increase—and also reduced the proportion of drowned ants. (The siphons could wick about half of the solution out of the containers.):

Here is a photo of the siphon structure; the caption is from the paper; the red dye, added to the sugar water, shows how the sand structure wicked the solution out of the container and onto the “ground”.

The sand syphon structures. (A, B) Syphon structures dried completely at room temperature. (C) Five minute after 1 ml of sugar water with 1% surfactant was added into in syphon structure A. (D) Five minutes after 1 ml sugar water with 1% surfactant and red food dye were added into the syphon structure B. (E) Top view of sand syphon structures.

The ants, then, were able to adjust their construction of the siphon to the conditions of the solution (the authors note that reduced surface tension may be characteristic of some nectars and other sweet liquids in nature), and perhaps to the perception of the number of drowned ants. They were also able to choose sand grains that were most effective at wicking the food: the medium and coarse ones. (They showed that the wicking capacity of fine sand is lower.)

The take-home lesson: the ants can somehow assess “foraging risk” and, when it’s higher, use tools, building a big siphon out of sand grains.

Now this is not the first instance of tool use in ants, though previous descriptions seem spotty. From the paper:

Several Myrmicinae ant species have been reported to be able to use debris to collect and transport liquid food into their nests (Banschbach, Brunelle, Bartlett, Grivetti, & Yeamans, 2006; Barber et al., 1989; Maák et al., 2017). There has been only one reported case of tool manufacture in ants prior to our study, but the validity of this report has not been confirmed yet. The Florida harvester ant P. badius, can construct small pellets of sand grains to soak up honey for transport (Morrill, 1972). However, this observation was only reported briefly in a short communication without any additional follow‐up study, and the data are insufficient to validate that this is a true case of tool manufacture by ants.

This, then, is a well documented case of tool use in ants—and it does count as tool use. It also shows for the first time that insects can adjust their foraging strategies to minimize risk.

Finally, the authors think it shows the ants show “high cognition”, and this is where I differ. The paper says this:

Our findings suggest that social insects may be able to create novel approaches to foraging by using available tools in situ to overcome the environmental constraints. The results also indicate that considerable capacities for high cognition and unique foraging strategy may be developed in social insects such as ants, which has previously only been characterized in vertebrates when facing risks and problems. Such sophisticated flexibility of tool use provides a powerful platform for further studying the cognitive mechanism and tool use strategy of social insects, and at the same time, promotes the research on the universality of tool use strategy in invertebrates.

Yes, it is sophisticated flexibility, though the authors don’t know whether the proportion of drowned ants, rather than the perception of surface tension (they’re correlated) leads to the construction of sand siphons.

But is it “high cognition”? Is it “cognition” at all? This isn’t a trivial question, but it’s an important one. The online Oxford English Dictionary gives two definitions of cognition related to this act (there are others):

  1. The action or faculty of knowing; knowledge, consciousness; acquaintance with a subject.
  2. Apprehension, perception

The question, then, is whether the ants are responding to an inbuilt behavioral program that robotically takes in information about the environment and responds with an adaptive act, which would accord with the second definition, or whether the ants “knowingly” do this; that is, are they conscious in some way—similar to us—of what’s going on. That would accord with the first definition of consciousness. One thing I don’t accept is that the ants learned to do this in the lab; this is surely a form of behavior that they perform in the wild. Ants aren’t that smart!

Since I’m a determinist, the ants really can’t make a conscious decision about what they do when the surface tension is low and they’re faced with drowning ants and sand grains of various size. None of us can make a free choice about what we do: we’re all constrained by our neurons and the environment we perceive. So the question here is not whether the ants make a free “choice,” but whether they are conscious of what they are doing. That is the question, and it’s one we can’t answer. All we can say is that the ants are able to perceive environmental stimuli and adjust their behavior in an adaptive way, and in this case use tools to do so. The rest involves the unanswerable question of “what is it like to be an ant?”. So it goes.


Zhou, ADu, YChen, JAnts adjust their tool use strategy in response to foraging riskFunct Ecol2020001– 12.


  1. ThyroidPlanet
    Posted October 11, 2020 at 11:02 am | Permalink


  2. rickflick
    Posted October 11, 2020 at 11:25 am | Permalink

    Ants have been around for 150 million years. Plenty of time to evolve some built in strategies to deal with circumstances like this. High cognition? – maybe in a metaphorical sense.

    • Posted October 11, 2020 at 2:37 pm | Permalink

      Why not make that sense of “cognition” the primary sense?

      Philosophers are fond of wrapping every problem up with consciousness (“It’s not real X unless the subject is *conscious* of X”), thereby making it insoluble since no-one understands consciousness.

      I think it’s better to make step-by-step progress by separating out concepts such as “cognition” from consciousness.

      • rickflick
        Posted October 11, 2020 at 2:56 pm | Permalink

        That sounds pretty reasonable to me. There seems to be a fundamental difference between what the ants are doing and what, for instance, humans do. The ants are simply engaging a behavioral algorithm that is stored in memory and passed down via DNA. If humans fashion a device to visit other planets, they are doing a far more complicated behavior that is stored in cultural artifacts.

    • Posted October 11, 2020 at 3:04 pm | Permalink

      As a collective they are more akin to a ‘mind’ & behave in organised ways, So we could call it that I think…?

      • rickflick
        Posted October 11, 2020 at 4:49 pm | Permalink

        Yes, it’s convenient to think of ants and bees and such as if the hive was a single organism. The individuals may share their DNA, which makes the analogy pretty nifty.

  3. GBJames
    Posted October 11, 2020 at 11:36 am | Permalink


  4. Posted October 11, 2020 at 11:38 am | Permalink

    Thanks for posting this!

  5. Randall Schenck
    Posted October 11, 2020 at 11:44 am | Permalink

    One also assumes the ants do not have a diabetes problem but they do have a sweet tooth. If I see any of them at the pharmacy filling any subscriptions I’ll be checking it out. We are fortunate the fire ants have not adapted to the colder weather, at least not yet.

  6. Barry Lyons
    Posted October 11, 2020 at 11:56 am | Permalink

    That’s fantastic! I suspect ants are probably conscious in some way. Amazing.

  7. Posted October 11, 2020 at 12:09 pm | Permalink

    Following Douglas Hofstadter, I think we should be open to the hypothesis that it is the ant colony that is doing the cognition (definition #2 cognition), not the individual ants.

    • GBJames
      Posted October 11, 2020 at 2:13 pm | Permalink

      That would require, minimally, some sort of colony-level nervous system. To my knowledge no such exists.

      • Posted October 11, 2020 at 2:58 pm | Permalink

        All it requires is some reliable information-transmission mechanisms between ants, and also that this enables more complete representations of relevant features of the situation than are available at the individual ant level. Admittedly, the second part is a tall order.

        • GBJames
          Posted October 11, 2020 at 3:10 pm | Permalink

          That equates cognition to information transfer. Which would make a simple telephone system cognizant of something. It requires something to exist that is capable of doing the comprehending. Insect colonies just don’t make the cut. Their behavior is entirely explicable by events at the individual insect scale.

          • wonderer
            Posted October 11, 2020 at 5:12 pm | Permalink

            I think a charitable interpretation of Paul’s comment would be to say that it equates cognition to distributed information processing. I’d assume Paul is expecting information processing in the brains of individual ants in the collective to provide such information processing.

            Not to say I am not highly skeptical of such a scenario, due to what I assume would be a much too low inter-ant communication bandwidth to do such mental modelling of the world.

            Still it is an interesting thought experiment, pertinent to how the ‘stupid’ neurons in our brains are able through networked connections, to yield human intelligence.

            • GBJames
              Posted October 11, 2020 at 5:35 pm | Permalink

              “…how the ‘stupid’ neurons in our brains…”

              “In our brains” is the critical bit. Brains are missing from ant colonies and bee hives.

              • wonderer
                Posted October 11, 2020 at 6:14 pm | Permalink

                But brains aren’t missing in ant colonies. Each ant has one.

                There is no in principle impossibility to ant brains being networked together to function as a bigger collective brain. (Crudely similar to the way processing tasks of a computer program can be split up to run on modern multicore processors.)

                As I’ve said, I don’t think it is realistic to think such a thing is happening in the case of ants, but if sufficiently high bandwidth communication between ants were possible, it is conceivable that ant brains could function as components of a larger ‘colony brain’.

                Whether there is any plausible evolutionary path to such collective braining is another interesting question.

              • GBJames
                Posted October 11, 2020 at 7:18 pm | Permalink


                “…if sufficiently high bandwidth communication between ants were possible…”

                If pigs had wings they could fly. They don’t. E. O. Wilson showed long ago how the remarkable behavior of ants/bees are created. No hive-level intelligence is required.

                Where does this need to believe some “higher level intelligence” is needed to explain ant/bee behavior?

              • wonderer
                Posted October 11, 2020 at 7:33 pm | Permalink

                Oy indeed.

                Where did I say higher level cognition was required? I’ve explicitly said multiple times that I don’t think higher level cognition was happening.

                It looks to me like you are just ranting in response to having been shown to not know what you are talking about. So you can have the last word.

              • GBJames
                Posted October 11, 2020 at 7:57 pm | Permalink

                This thread includes assertions that “more complete representations of relevant features of the situation than are available at the individual ant level” (Paul, above). It includes your speculation (absent evidence) that IF there was sufficient bandwidth, then there could be “a larger ‘colony brain’”.

                None of these ideas are supported by actual evidence. And none of them are necessary to account for the remarkable behavior of ant colonies and bee hives. If you think I’m just ranting, fine. But it is a rant based on reading E. O. Wilson’s rather expert work on the subject. I trust you are familiar with with his work on the subject.

          • Posted October 11, 2020 at 5:13 pm | Permalink

            In order to fulfill the second requirement I mentioned, the colony would also have to store information (more reliably than any individual ant does). But you’re right that this means inanimate computers can cognize things, per OED’s definition 2. I don’t see a problem with that.

            • GBJames
              Posted October 11, 2020 at 5:38 pm | Permalink

              I didn’t say “computer”. I said “simple telephone system”.

  8. wonderer
    Posted October 11, 2020 at 12:25 pm | Permalink

    This brought Hofstadter’s idea to my mind as well. However, I would think that for ‘higher consciousness’ to be involved some subset of the ant colony would need to pe considering a mental model of the physical situation, and issuing directives to other ants. It seems to me that such a subset of the colony implementing higher consciousness would look like a cluster of ants interacting furiously with each other, to no obviously visible purpose.

    Is there any reason to think that such is the case?

  9. Steve Pollard
    Posted October 11, 2020 at 1:40 pm | Permalink

    Thanks for a great post as ever.

    • jezgrove
      Posted October 11, 2020 at 1:57 pm | Permalink

      Ditto – completely fascinating. It just goes to show that ants are more intelligent than [insert name of your least favourite politician here]…!

  10. Posted October 11, 2020 at 3:02 pm | Permalink

    I too am having a hard time with this being true cognition.
    But we often describe ants and other eusocial insects as being like cells in a much larger and rather sophisticated organism. So it is at least worth exploring whether an analogy can be made that ant brains form a collectively larger brain.

  11. Gregory Kusnick
    Posted October 11, 2020 at 3:58 pm | Permalink

    This is interesting behavior by the ants, but I’m not convinced it represents anything novel in the way of adaptive tool use. Chimps and corvids presumably deploy their tool-using skills where needed to solve particular problems, and don’t apply those skills where they’re not needed. I don’t see that these ants are doing anything fundamentally different.

    On a side note, “foraging risk” seems a bit anthropomorphic. My guess is that the ants’ primary concern is the cost to the colony (in terms of worker mortality) of gathering a given amount of food, and that their own individual safety is secondary to that.

  12. Jon Gallant
    Posted October 11, 2020 at 4:10 pm | Permalink

    Shades of “Microcosmic God”, Theodore Sturgeon’s great sci-fi story of little creatures that live in a sort of sandpile, in a complex society that invents all manner of advanced technology. Once, with the aid of peyotl, I discovered an entire civilization of similar little creatures in the weave of living room carpet—although the creatures inexplicably vanished later on.

  13. Posted October 11, 2020 at 6:18 pm | Permalink

    Ants have quite a range of behaviours and this to me shows a little more than say, using a stick bridge to access food, cutting down travel time etc.
    They have problem solved and actioned on it, to us that is.
    I see that as just being busy ants using a repertoire of lifting, shuffling stuff around, nest & storing actions and hitting on a result.
    No plan came out with hard hats just a ‘will’ to capture energy.

  14. merilee
    Posted October 11, 2020 at 6:58 pm | Permalink


  15. Mark R.
    Posted October 11, 2020 at 7:48 pm | Permalink

    Don’t have much to add, except I enjoyed this post (plus the comments) and it’s another behavior that proves ants are very impressive life forms.

  16. Posted October 11, 2020 at 10:58 pm | Permalink

    altruistic ants

  17. KD
    Posted October 12, 2020 at 10:43 am | Permalink

    “what is it like to be an ant?”

    If we could only ask her honey’suncle. . .

Post a Comment

Required fields are marked *

%d bloggers like this: