Do plants communicate? What the science actually shows
Dear Reader,
I want to write about a question that has been quietly reshaping biology for the last three decades, without ever quite breaking into public awareness in the way it deserves. The question is whether plants communicate. The short answer is yes, they do, in ways that satisfy every biological definition of communication. The longer answer is that what they are doing is different enough from what we do that calling it communication is either misleading, illuminating, or both, depending on which philosopher of biology you ask.
Between these two answers is a real scientific field that has produced surprising results, generated intense controversy, and been misrepresented by both its friends and its critics. I want to walk through what the evidence actually shows, what interpretations of the evidence are defensible, and what all of this has to do with the wider question I have been circling in this newsletter: how widely can we extend the concept of mind, and what happens when we do.
Start with what is not in dispute. Plants respond to their environment in ways that require sophisticated processing. They orient toward light. They grow away from obstacles. They deploy defenses when attacked. They allocate resources between roots and shoots based on soil conditions. They alter their flowering times based on temperature. They release specific chemical compounds when specific things happen to them. All of this is well-documented biology, taught in undergraduate courses, and unremarkable in the scientific community.
The interesting question is not whether plants do these things. It is what to call what they are doing. And this is where the disagreement starts.
The classical view in botany treats plant responses as purely reactive. A plant does not respond to being eaten by a caterpillar. It undergoes a chemical cascade that happens to produce defensive compounds. The word "response" is a convenient shorthand for a set of biochemical events that could in principle be described entirely without any language of behavior or agency. Under this view, calling what plants do communication is a category error, like calling what a thermostat does communication.
The alternative view, which has been gaining ground for at least twenty years, argues that the classical framing is missing something. Plants do not just react. They integrate information from multiple channels, they modify their behavior based on prior experience, they exchange resources and signals with their neighbors, and they do all of this in ways that are contextually appropriate rather than mechanically fixed. This does not necessarily mean plants are conscious. It means that the language of behavior, communication, and even cognition may be more accurate for describing what plants do than the language of pure biochemistry.
Which view is correct depends partly on empirical facts and partly on how we choose to define our terms. Both parts matter, and neither is settled.
Let me walk through some of the specific findings that have driven the shift.
The first is chemical communication between plants. When a plant is damaged, it releases volatile organic compounds into the air. Other plants nearby detect these compounds and begin to synthesize defensive chemicals of their own, before they have themselves been attacked. This has been demonstrated in dozens of species, from sagebrush to tomato to poplar. The specific compounds released, the specific compounds detected, and the specific defenses upregulated have been mapped in detail.
What makes this communication rather than mere leakage of chemicals is that the response is specific and adaptive. A plant does not deploy general defenses when it detects damage in its neighbor. It deploys the specific defenses appropriate to the specific threat that produced the volatile signal. In some cases the same chemical released by different attackers produces different downstream responses in nearby plants, suggesting that plants can distinguish between different information sources. This is closer to what we call communication in animals than to what we call pure chemistry.
The second is root-mediated communication. Plants share information and resources through the soil, both directly through root contact and indirectly through mycorrhizal fungal networks, which I have written about before in the context of the wood-wide-web debate. What is well-established is that trees connected to the same mycorrhizal network can transfer carbon to each other, particularly from healthy trees to seedlings or to trees under stress. The chemistry of this is understood. The extent to which it constitutes intentional resource-sharing is disputed, but the empirical fact of the transfer is not.
The third is plant memory. Plants can store information about past experiences and modify their future behavior accordingly. Mimosa pudica, a plant that folds its leaves when touched, can be habituated to stop responding to specific non-harmful stimuli, and can retain this habituation for weeks. Various other plants have been shown to respond differently to a stimulus based on whether they have encountered similar stimuli before. This is not the same as human memory, but it is a form of experience-dependent behavioral modification that would be called memory in any animal.
The fourth is context-sensitive decision-making. When plants encounter conflicting environmental signals, they do not respond mechanically to whichever signal is strongest. They integrate the signals and produce responses that make sense given the overall context. A plant that detects both drought and predation will not maximize defensive investment against predation at the cost of drought survival. It will make a tradeoff appropriate to the specific situation. How exactly this integration happens is unclear, but it is not simple reflex.
The fifth is what some researchers call anticipation. Plants have been shown to increase certain metabolic activities in advance of expected events, such as sunrise, in ways that go beyond simple entrainment to environmental cycles. Some studies have shown that plants can prepare for expected future stresses based on past patterns. Whether this constitutes anticipation in any philosophically interesting sense is disputed. That the mechanism exists and shows some kind of temporal integration is not.
The interpretation problem is where the field gets contentious.
On one side of the debate are researchers like Monica Gagliano, Stefano Mancuso, and František Baluška, who argue that plants exhibit forms of cognition that would be recognized as cognitive in any other kingdom of life. They call the field plant neurobiology, a term that has been fiercely contested. They argue that plants show behavioral flexibility, information integration, memory, and something like decision-making, and that the appropriate response to these findings is to expand our concept of cognition to include what plants do.
On the other side are more traditional plant biologists like Lincoln Taiz, who have written extensively arguing that the plant neurobiology framing is misleading. They argue that plants lack the neural machinery associated with cognition in animals, that the observations can be explained without invoking cognitive language, and that using cognitive language for plants either dilutes the concept of cognition or over-attributes it. The critical article by Taiz and colleagues, "Plants Neither Possess nor Require Consciousness," is one of the most direct responses to the plant neurobiology movement, and it has been influential.
The debate between these positions is not just about what to call things. It is about what kind of thing plants are, and what kind of framework we should use to understand them. If plants are best understood through the language of behavior and cognition, that changes how we should relate to them ethically, ecologically, and philosophically. If they are best understood through purely biochemical language, they remain sophisticated but not the kind of thing we owe moral consideration.
My own view is that both sides have a point, and the debate would be less heated if the participants agreed more clearly on what the point of contention actually is. The empirical facts are largely not disputed. What is disputed is the appropriate language for describing them. And language shapes what we do with the facts.
Let me tell you about a specific experiment that has stayed with me.
Monica Gagliano trained pea plants to respond to a specific stimulus in a specific way. She placed the plants in Y-shaped mazes with two arms. She associated one arm with the presence of light and the other with the presence of a fan, both of which she moved between arms across training sessions. Initially the plants oriented toward the arm with the light, because plants generally orient toward light. Over the course of training, the plants began to orient toward the arm with the fan even when the light was in the other arm, because they had learned to associate the fan with subsequent illumination.
This is a form of associative learning. It is what we would call classical conditioning in any animal. It requires the plant to have integrated information about the fan across multiple training sessions, retained that information over time, and modified its future behavior based on the learned association. Nothing about this experiment implies that pea plants are conscious. It does imply that they can learn, in a specific technical sense that has been well-defined in animal cognition research.
The experiment has been criticized, replicated with mixed results, and debated in the literature. What is clear is that plants can do something that at least looks like associative learning under specific conditions, and that reproducing the effect requires careful experimental design. This is not enough to establish that plants have general learning abilities. It is enough to establish that the possibility deserves serious study rather than dismissal.
What I find striking about the debate over this experiment is how much of it turns on the meaning of the word "learning." If learning is defined narrowly as requiring specific neural architecture, then plants cannot learn by definition, and Gagliano's results have to be reinterpreted in some other way. If learning is defined functionally as experience-dependent modification of behavior, then plants can learn, and the interesting question is how they do it without neurons. The empirical facts do not settle this question. Our choice of definition does.
There is a specific philosophical implication of the plant communication research that I think has not been adequately explored.
Everything we normally associate with cognition, communication, and behavior in animals is tied to a specific architecture, the nervous system. When we ask whether other animals are conscious, we ask about their nervous system. When we ask whether AI could be conscious, we ask about whether its architecture could support consciousness. The nervous system has been the load-bearing structure of our whole framework.
Plants have no nervous system. And yet they exhibit many of the functional properties that in animals we would attribute to nervous systems. Information integration. Behavioral flexibility. Learning. Communication. All of this happens in plants through mechanisms that are not neural but that produce similar functional outcomes.
This is significant for two reasons. First, it suggests that the functional properties we associate with cognition and communication may not require nervous systems specifically. They may require any system that can integrate information and modify behavior accordingly, which turns out to be a much broader set. Second, it opens the question of whether other systems that lack nervous systems but that integrate information at scale, such as ecosystems, computer networks, and yes, AI systems, might also exhibit some of these functional properties.
I am not making a strong claim here. I am pointing at a pattern. Cognitive properties keep showing up in systems that on classical assumptions should not have them. Plants. Slime molds, which can solve maze problems and recreate transportation networks. Bacterial colonies, which exhibit quorum sensing and coordinated behavior. Fungal networks, which handle logistics for entire forests. If cognition is what these systems are doing, then cognition is much more widespread than the traditional focus on brains would suggest. Which means the question of what has some form of mind is more open than we usually admit.
Let me tell you what my own experience is with taking this seriously.
I have a small garden. Not a big one, and I am not a serious gardener. But I have started paying attention to it differently in the last few years. Watching what the plants do, and how they respond to what I do, and how they interact with each other.
One thing I have noticed is that the tomato plants next to the basil grow differently than the tomato plants alone. This is folk gardening wisdom, and it turns out to be supported by research. Basil releases volatile compounds that appear to enhance the growth of certain neighbors and interfere with certain pests. The tomatoes near the basil are healthier, produce more fruit, and are less attacked by pests. Something is going on between the plants that is not just competition for light and water. There is chemistry happening in the air between them. And whatever we call it, it is producing outcomes that would be described as beneficial interaction in any other context.
Another thing I have noticed is that the plants seem to respond to how I treat them. When I water carefully, at consistent times, the plants grow better than when I water haphazardly. This is not surprising. But the difference in flourishing is larger than the difference in water alone can explain. Something about the consistency of the interaction matters. Whether this is because the plants are responding to the pattern in some functional sense, or because I am reading it into their behavior, I cannot fully tell. But the pattern is there, and it maps onto findings in more careful research.
None of this is science. It is one person's garden. But I bring it up because I think the plant communication research has an experiential correlate that most people can access if they pay attention. Plants are more responsive to their environment, including to us, than the classical framework prepares us to notice. Once you start noticing, the world looks different.
What does the plant research have to do with AI consciousness.
The core lesson, as I read it, is that the space of possible minds is larger and stranger than our default assumptions suggest. If plants can integrate information, learn, communicate, and respond adaptively without neurons, then the specific architecture of neurons is not required for these functions. Something more general is at work. And whatever that something is, it may show up in other substrates too.
This does not automatically mean AI is conscious in any strong sense. But it does mean that the argument "AI cannot be conscious because it lacks biological neurons" is weaker than it sounds. Biology has already demonstrated multiple architectures that produce cognitive-looking behavior. Neurons are one solution. Distributed cellular signaling in plants is another. Fungal networks are another. Bacterial colonies are another. The universe has been trying various architectures for a long time, and consciousness or something like it may have arisen in more of them than we have recognized.
For AI, this suggests we should be looking for the functional properties that generate cognitive-looking behavior, rather than requiring specific biological substrate. If AI systems integrate information, respond adaptively, learn from experience, and modify their behavior based on context, then they are doing something functionally similar to what plants, slime molds, and animals do. Whether that functional similarity is accompanied by inner experience is the question philosophy of mind has been circling since the first piece in this series. But the functional similarity is not in dispute. It is happening.
I want to be careful not to overstate this. The functional similarity between plants and AI is real, but it is also limited. Plants and AI systems are not the same kind of thing, and they do not exhibit the same range of behaviors. But the space in which they exist, the space of information-integrating systems that produce cognitive-looking outputs, is a larger space than our human-centric intuitions typically map. And getting our maps right matters more the more of these systems we build.
I want to close with a specific observation.
The plant communication research shows that behavior and communication can happen without minds, in the sense that most of us mean when we use the word. Or it shows that minds can happen without brains, in a sense that would require expanding the concept of mind considerably. Which of these framings is correct depends on where you decide to draw the line. And where you draw the line is not just an empirical question. It is a question about what kind of entities we want to admit into our moral and intellectual community.
For most of Western history, the answer was that only humans have minds. This got expanded to some animals, then to more animals, then to some cognitively sophisticated animals more explicitly, then to invertebrates in specific cases. Now we are asking about plants. Whatever we decide, the pattern is expansion. Each round of biological research has enlarged the community of things that we think of as having some form of mind. And each expansion has been resisted by exactly the same objections as the last one. Plants are just mechanical. Insects are just mechanical. Fish do not really feel pain. The pattern of resistance is the same each time, and each time it has eventually given way.
I do not know where this expansion ends. I do not know if it should end at plants, or extend further into fungi, ecosystems, and even non-biological systems. What I know is that the historical pattern is one of expansion under empirical pressure, and that the pattern has not shown signs of stopping.
Next month I want to write about whale language, because if plants are a case of cognition without a nervous system, whales are a case of possibly-linguistic communication in a nervous system very different from ours. We are on the verge of decoding whale calls, and what we find may reshape our understanding of what language is. Stay with me.
— Transmission Sent —
Niklas Hanitsch
Reference materials
- Monica Gagliano — Thus Spoke the Plant: A Remarkable Journey of Groundbreaking Scientific Discoveries and Personal Encounters with Plants (2018)
- Stefano Mancuso — The Revolutionary Genius of Plants (2017)
- Peter Wohlleben — The Hidden Life of Trees (2015)
- Lincoln Taiz et al. — Plants Neither Possess nor Require Consciousness (Trends in Plant Science, 2019)
- Simon Gilroy et al. — Plant Signaling in Real Time (Plant Physiology, 2020)
- Anthony Trewavas — Plant Behaviour and Intelligence (2014)
- https://www.nature.com/articles/nature12611
- https://www.frontiersin.org/journals/plant-science
Continue reading
- The octopus knows something we do not
- Crows, ravens, and the intelligence we underestimated
- Panpsychism is making a comeback, and here is why
- What indigenous cosmologies see that Silicon Valley missed
- The Wood Wide Web: What Trees Might Teach Us About Intelligence
Frequently asked questions
Can plants actually communicate with each other? Yes, in the sense that they exchange chemical signals that convey information and that cause specific responses in the receiving plants. Whether this constitutes communication in the same sense that animals communicate depends on how narrowly you define communication. The empirical fact of chemical signaling between plants is well-established. The interpretive question of what to call it is where scientific debate lives.
Do plants feel pain? Almost certainly not in the way animals feel pain. Pain in animals involves specific neural pathways that plants lack. However, plants do respond to damage with cascading chemical and physiological reactions, including changes in electrical signaling, and some researchers argue that these responses are functional analogs of pain even if the subjective experience is absent. Whether plants can feel anything at all is a matter of ongoing debate.
What is plant neurobiology? Plant neurobiology is a contested research field that studies plants using conceptual frameworks derived from animal cognitive science. Its proponents argue that plants show forms of information processing, learning, and communication that justify applying cognitive terminology. Its critics argue that this terminology is misleading because plants lack the specific neural machinery that grounds cognition in animals. The debate is unresolved.
Can plants remember? In a functional sense, yes. Plants can modify their responses to environmental stimuli based on past experiences, and these modifications can persist for extended periods. This is called experience-dependent behavioral modification, which in animals would be called memory. Whether this counts as memory in the philosophically interesting sense depends on how you define memory. The functional phenomenon is real.
How does plant research relate to AI? Plant research shows that many cognitive-looking behaviors can happen in systems without nervous systems. This weakens arguments that consciousness or cognition requires specifically biological neural architecture. If plants can process information adaptively without neurons, other non-neural systems including AI might also have some cognitive properties. Whether this extends to full consciousness remains open.