Foresight, intelligence, decisiveness, a good memory. All admirable qualities. Just what you鈥檇 look for in a friend, colleague, doctor or pot plant.
Hang on a moment鈥urely intelligent life does not have leaves? Mention the word 鈥渃abbage鈥� or 鈥渧egetable鈥� and genius isn鈥檛 the first thing that springs to mind. And if someone said I had the foresight of a geranium, I don鈥檛 think I鈥檇 take it as a compliment. But speak to Tony Trewavas of Edinburgh University and he鈥檒l tell you that plants have been seriously underestimated. Plants have the power to compute, they show foresight and they remember what happens to them, he says. 鈥淚鈥檓 not being silly,鈥� he insists. 鈥淭hese powers underpin a novel form of intelligence.鈥�
Trewavas thinks the only real difference between us animals and our distant green relatives is how mobile we are. We鈥檙e used to judging intelligence by actions, he says. It鈥檚 what we do and say that reveal what goes on inside our minds. So plants, silent and rooted to the spot, naturally don鈥檛 appear too bright. But they do move and they do react to the world around them, he says, and they do so in intelligent ways. Plants can assimilate information, calculate outcomes and respond using a complex series of molecular signalling pathways that are remarkably like those in our own brains. 鈥淭he computational capacity of a plant is probably as good as many animals,鈥� he says, provocatively.
Advertisement
It鈥檚 true that plants can do many of the things we use our brains for. For a start, they have senses (快猫短视频, 26 September 1998, p 24). They can detect and react to light, sounds, chemicals, vibrations and touch, not to mention water, gravity and temperature. Usually their response is to change their pattern of growth, but in more varied and complex ways than you might think. Over the past few years, more and more botanists and ecologists have been using terms like foraging, competing and predator evasion when talking about plants. 鈥淲e use the word 鈥榖ehaviour鈥� routinely now,鈥� says Michael Hutchings, a plant ecologist at Sussex University. 鈥淧lants are far from passive. Like animals, they respond to environmental cues.鈥�
Most scientists see such responses as predetermined, reflex outcomes: the elegant result of millions of years of evolutionary adaptation. But Trewavas sees them as flexible and adaptable. 鈥淧lasticity is foresight,鈥� he says. Plants adjust their growth and development to maximise their fitness in a variable environment. And, he says, this fits exactly the definition of intelligence devised by David Stenhouse, a philosopher and psychologist in New Zealand, who wrote a book in the 1970s about its evolution.
Stenhouse spoke of intelligence as 鈥渁daptively variable behaviour during the lifetime of the individual鈥�. So intelligent responses are not just adaptations on an evolutionary timescale, and they鈥檙e not just predetermined and predictable. By this definition, Trewavas believes that all plants are capable of intelligent behaviour.
For a start, he claims plants can anticipate future problems and make decisions about how to avoid them. Botanists have long known that growing shoots can sense neighbouring vegetation. Green leaves absorb red light but reflect infrared, and plants can spot shifts in the ratio of red to infrared light that indicate nearby greenery. Trewavas says plants predict the consequences of such a presence, mapping out where they are most likely to meet competition and shade in the future, then take evasive action if necessary. The plant will alter its whole shape, the number and size of its leaves and the structure of its stem to gain the best possible position in the sun. Not exactly the work of a mastermind, but an adaptively variable response all the same.
Plants can also take dramatic avoidance action. The stilt palm has a stem raised on prop roots, so that it stands up above the soil. When neighbours encroach on its light or nutrient supply, it takes very obvious evasive action 鈥� the whole plant moves back into full sunlight by growing new prop roots towards the sunny side, while those in the shade die off. 鈥淭hat this is intentional behaviour is very clear,鈥� says Trewavas. And it鈥檚 certainly adaptive. Could this be another sign of intelligence?
Perhaps not surprisingly, not everyone thinks so. Most plant responses are more like our own reflex actions, instincts or phobias, says Andrew Goldsworthy, a plant scientist from Imperial College, London. When plants outgrow their neighbours because they 鈥渟ee鈥� them, or when they sort out a compromise growth pattern to deal with conflicting signals such as wind vibration, which normally stunts growth, and infrared light, which normally stimulates it, they may look like clever, sophisticated decision makers. But it鈥檚 really just a pre-programmed rote response, he says.
Trewavas disagrees. Many plants show a behavioural flexibility that goes far beyond mere reflex or programming, he says. Roots can follow mineral or moisture gradients in the soil, but they don鈥檛 always take the simple route. Hutchings and his colleagues have studied the foraging behaviour of a creeping herb called Glechoma (快猫短视频, 27 May 2000, p 28). When they鈥檙e in good soil these herbs grow more branches, shoots and leaves. They also form clumps of root much faster to fully exploit the patch. But when they鈥檙e on poorer territory they spread farther and faster, almost as if they鈥檙e escaping, and their rhizomes are generally thinner and branch less frequently.
This means that new shoots develop further from the parent plant and actively search for new resource-rich patches. And the amount of growth is not related just to the absolute quality of a patch, but to how good it is relative to surrounding soils. Not only that, but experiments have shown that related plants can sense the presence of competitors鈥� roots and head off to other areas 鈥� even when there鈥檚 still lots of food around.
One of Trewavas鈥檚 favourite examples of flexibility and foresight is the remarkable foraging strategy of a parasitic plant called dodder, or Cuscuta, as studied in the early 1990s by Colleen Kelly, now at the University of Southampton. Dodder doesn鈥檛 photosynthesise. Instead, it coils around a host plant, piercing it with shoots to draw out nutrients and water. The parasite鈥檚 intelligence, according to Trewavas, comes in anticipating how much energy a host will yield, and deciding how much effort to expend in exploiting it.
It takes about four days after contact before dodder starts collecting any nutrients, but long before that the parasite has somehow anticipated how fruitful this host will be, and grown more or fewer coils around it accordingly. More coils will lead to more shoots and greater exploitation, but if the host is poor, too many coils will be a waste of energy. Kelly showed that dodder鈥檚 foraging strategies fit all the mathematical models devised to explain the economics of animal foraging, describing when you should eat and when you should move on, depending on the quality of the patch and how common other patches are. Dodder may not be the most cunning predator around, but in its foraging behaviour, it can do the maths as well as any animal.
The key proof that plants are intelligent, says Trewavas, is that they show subtle flexibility in their responses 鈥� that is, they are adaptively variable, not just adaptive. Plants are all individuals, and no two seeds turn out alike, even when they鈥檙e genetically identical or grown under seemingly identical conditions. Add to that their responses to more than 15 different sensory signals 鈥� including light, chemicals, water, gravity, the feel of the soil, and damage 鈥� which are combined and compared, so that each response depends on a complex mix of factors. Clearly plants are very flexible indeed.
Just like a simple nervous system, these signalling systems have the potential for computation and learning. As Darwin pointed out more than a hundred years ago, 鈥渋n several respects light seems to act on plants in nearly the same manner as it does on animals by means of the nervous system鈥�. But it has taken modern molecular biology to show just how similar animal nervous systems and plant signalling systems really are.
Plants use changes in voltage across their cell membranes to send electrical signals from one region to another, similar to the action potentials that propagate along our own nerves. Just like our own pain messages, these potentials can signal that part of the plant has been injured. And many of the chemicals used to pass messages within and between plant cells are exactly those used to process information within and between our own brain cells. Both animal and plant cells respond to proteins, nucleic acids, ions and hormones, glutamate, calcium, cyclic nucleotides and protein kinases 鈥� a complex signalling language shared by two different kingdoms of organisms.
The molecular underpinnings of learning and memory are also similar. When animals learn to withdraw more rapidly from a repeated threat, such as a food source that made them ill, or an electric fence they keep encountering, the speed and size of the electrical signal is increased over a matter of minutes. A system that uses calcium ions, chemicals called second messengers and a few enzymes temporarily makes the ion channels that transmit the signals a little more responsive. If the threat persists, this heightened awareness becomes permanent by changing gene expression and protein building to make more channels or more connections between cells.
When a plant senses a lack of water, exactly the same signalling molecules direct it to create more sensitive channels to close the stomata and take other measures that help control the amount of water in its cells. Over the long term, gene expression and the rates of protein synthesis change, cell walls thicken and leaves get smaller. Eventually the plant will grow more roots, and fewer shoots and leaves. 鈥淭he plant learns by trial and error when sufficient changes have taken place so that further stress and injury is minimised,鈥� says Trewavas. The plant also modifies its strategy in response to other environmental signals such as nutrients, temperature and humidity, and its own history: its age, previous disease and so on.
It鈥檚 here, at the very smallest scale of the signalling systems, where Trewavas sees the real brains of a plant. In plants, just as in animals, calcium ions are the major intermediaries that turn sensory information into a common internal language where the different signals can be combined. Virtually every sensory signal that a plant can pick up produces a transient rise in calcium levels, says Trewavas. He has also found that the calcium blip comes with slightly different timings depending on the sensory signal, potentially helping the plant to distinguish between different sensations. Perhaps the calcium system is where computations and decisions are made and memories stored, he suggests.
Experiments have shown that calcium ions don鈥檛 move very far by themselves. They diffuse only a small distance from stores inside cells through calcium channels in the surrounding membrane. But when attached to a variety of other chemicals and enzymes they can encourage neighbouring calcium channels to open, eventually allowing signals to be passed over much greater distances. It sounds a tortuous way to send a signal 鈥� rather like a slow Mexican wave 鈥� but having several steps allows great flexibility and complexity.
Each channel, suggests Trewavas, is a bit like a node in a neural network. Each one is a switch, but a switch whose ease of opening can be manipulated. Signals pass through particular pathways of switches, which are effectively wired together by their layout on the cell membrane. The switches direct the flow of information along particular pathways and can be set to pass or block signals that arrive at the same time, just like the 鈥淎ND鈥� and 鈥淥R鈥� logic gates used in computer circuits. In addition, they can be made more or less sensitive depending on past signals 鈥� a form of learning. Trewavas鈥檚 experiments have shown that different sources of information can be combined, prioritised or weighted, or even ignored. On a larger scale, calcium movements around the network can become coordinated into waves or oscillations. All are properties you鈥檇 expect to find in a neural network. 鈥淐alcium forms the basis of the intelligent system controlling plasticity,鈥� suggests Trewavas.
Clearly what plants do is amazingly complex and elegant, and we have much still to learn. But can the movements of calcium ions within plant cells really produce intelligence? Rolf Pfeifer, an AI researcher at the University of Zurich and author of Understanding Intelligence, points out that there are many ways of defining intelligence. Trewavas鈥檚 version is a perfectly plausible one, but there are many others 鈥� the ability to converse in a human-like way, for example. Ultimately, says Pfeifer, it becomes pointless trying to define intelligence and decide which behaviours we should call intelligent and which not. What鈥檚 interesting is taking individual behaviours and finding out how they work.
Goldsworthy is similarly sceptical of his definition. It seems reasonable, he agrees, but why stop at plants? An individual cell such as an amoeba also makes 鈥渋ntelligent鈥� decisions in its natural environment. The cells in our own bodies also respond 鈥渋ntelligently鈥� to many factors 鈥� electrical, chemical, tactile, and so on 鈥� in order to grow and differentiate in a coordinated way. 鈥淚f [intelligence] is seen just as a means to respond to complex stimuli in a way that promotes survival, it is very widespread indeed,鈥� he says.
But when Trewavas calls plants intelligent, he鈥檚 not just provoking headlines and producing good PR for botany. His point is that people have overlooked a huge amount of complexity in the way plants perceive and respond to the world. If using the term 鈥渋ntelligence鈥� stirs up controversy or a debate about how complex plants are, all the better for our final understanding.
The fact that 99 per cent of Earth鈥檚 biomass is plants suggests that they are extremely good at dealing with their local environment, Trewavas reckons. He thinks it鈥檚 remarkable 鈥渉ow plants can compute their environment on a whole-plant basis without the benefit of a brain鈥�. Even if this 鈥渕indless mastery鈥�, as he calls it, is not intelligence, plants are clearly doing something right. Maybe this will even prompt a rethink about what sort of complexity and calculation our own bodies are capable of, even without the help of our central nervous system.
Above all, Trewavas is throwing down the gauntlet to plant scientists. We鈥檙e only just beginning to expose the wonderful variety of plant responses, he says. But we鈥檒l only find them in the wild. Only when there鈥檚 real variety in the environment 鈥� not the static, simplified conditions of a greenhouse 鈥� will real intelligent behaviours reveal themselves. 鈥淭he last place to find intelligence is in the lab,鈥� says Trewavas.