Plant Intelligence?

Today I read a particularly thought-provoking paper—Aspects of Plant Intelligence by Anthony Trewavas—arguing that we should recognise plants as exhibiting intelligent behaviour. I'm not convinced, but there are some very thought-provoking things in the paper, so I thought I'd summarise my reactions here and see if anyone else has a response.

The general message of the paper was that while plants don't locomote like animals do, their growth patterns are highly selective and adaptive to their environment, and they do exhibit some temporal behaviours in response to environmental changes (such as the opening and closing of stomata). Trewavas argues that these are just as validly described as intelligent as the animal behaviours we normally describe as such; they just happen on a slower timescale. The paper reviews a wide range of evidence for this argument, some parts of which I found more persuasive than others.

[note: I started writing this last week, but took a while to finish the post]

Unconvincing evidence

At various points in the paper, Trewavas points out similarities between the underlying mechanisms of signal transmission within a plant and within an animal's nervous system. I had two problems with this, though the more obvious one is also easily dismissed.

I am a firm believer in my advisor's argument that the underlying mechanisms are absolutely irrelevant to whether a given behaviour deserves a label like 'intelligent' or 'learning', making arguments from similarity of underlying mechanisms irrelevant. On reflection, however, papers from our lab often get criticised by reviewers who don't share this viewpoint, precisely because we don't make the connection they want us to make between underlying mechanisms. So perhaps the references to similar mechanisms were a wise to sop to a group of people who account for quite a large proportion of biologists.

That makes the second problem—that the resemblances between plant and animal mechanisms don't amount to much—more important. The issue is that all biological systems draw from a surprisingly limited toolkit of basic forms and structures. All the diversity of plant and animal forms is based on a surprisingly small set of basic molecular parts, each of which can only function in a limited set of ways (this idea is well developed in How The Leopard Changed Its Spots). So when Trewavas argues that Quite remarkably (p6) plant cells and animal neurons use a few of the same ions and chemicals to transmit information, that's a bit like saying how remarkable it is that my ethernet connection and the first telegraph both use electrons in wires to send signals. It doesn't really tell us anything about the capabilities or interestingness of either, and not even very much about how they work.

In fairness, a lot more of the paper is devoted to examples of behaviour than to discussion of the underlying mechanisms, and some of the behavioural examples are more impressive than others. I'll describe a couple of those I was less impressed by before moving on to the parts that really gave me pause for thought.

Several of the behavioural examples—while clearly adaptive—struck me as too simple and inflexible to count as intelligent. Examples at different timescales include things like phototropism and the opening and closing of stomata (which is triggered by changes in humidity levels and/or gas concentrations in the air). I felt that both of these examples have more in common with reflexes than any animal behaviour that one would normally refer to as intelligent: there is a stimulus and a response, but no flexibility in how the stimulus is responded to.

More impressive evidence, and caveats

The paper does, however, contain some evidence that did give me serious pause for thought, and some evidence that was provocative enough that I need to read the referenced work before I can decide what it amounts to. The parts that impressed me the most were all aspects of drought resistance behaviour. It seems that a plant's initial response to drought is a set of reflex-like adaptations (for instance growth slows, and then speeds up when the drought is over), but there are also longer-term responses suggestive of learning. I was struck by the passages about resistance to future droughts being potentiated by prior exposure to milder droughts (p4), and changes in the structure of new leaves after a period of drought (p5). The persistence of these changes in behaviour after the stimulus is removed suggests to me that there is some kind of learning going on, rather than a simple reflex arc; something along the lines of the habituation observed in Aplysia.

The other piece of evidence that really made me stop and wonder was about inter-plant communication. Trewavas referenced a paper that I have to dig up and read now (Petterson et al. 1999), claiming that when a plant is attacked by aphids it releases chemical signals that lead other individuals to act pre-emptively.

There is still something in the Aplysia behaviour that doesn't seem to have an equivalent in plants: a variety of possible responses to a given stimuli. That Aplysia paper includes examples of individuals being habituated to particular species of seaweed, while their response to other species is unaffected. By contrast, all of the behaviours Trewavas describes in plants are fixed in relation to stimuli, at least as far as I can tell. I need to read the inter-plant communication paper, because that seems like the most likely candidate for learned flexibility of response.

Intelligence is in the eye of the beholder

In the end, the strongest message I took from this paper was about the extreme difficulty of defining "intelligence" in a hard-and-fast way. Trewavas presented all this evidence because he clearly found it matched his criteria for intelligent behaviour, but I'm less convinced because my criteria are a little different. So the onus is on me to state what my criteria are for accepting that a given behaviour is intelligent.

I feel that there are at least two important criteria that distinguish intelligent behaviour from the 'merely reactive': anticipation and flexibility. While plants do seem to show some sorts of anticipatory behaviour—responding to a signal in a way that prepares them for what will follow that signal—I see the lack of flexibility as crucial. If I see evidence of plants being able to modify which stimulus is associated with which response, in ways that are not pre-set at the beginning of development, then I would have to concede that plants can reasonably be described as intelligent.

Until then, all the cool things that plants do in response to environmental variability strike me as more in the realm of "the genius of evolution" than showing intelligence in the organisms.


Trackback URL for this entry is:


You are certainly right about the definitions part. Plants are not intelligent if the definition is strict, and vice versa.

Two aspects of intelligence seem to be strictly animal:
(1) The ability to form abstractions based on other abstractions, i.e. thinking.
(2) The ability to recall specific events, i.e. memory as we know it - as opposed to conditioning.

Apart from these three things plants do have most of the other attributes of intelligence.

Posted: November 11, 2006 02:42 PM

Post a comment