Thursday, 23 January 2014

Can plants learn?

So a while ago I wrote a post on the question of whether insects can feel pain and at the time I thought that was a reasonably unusual question to ask. Now I'm asking whether plants can learn.

Feed me, Seymour...

The reason why I'm asking this question isn't because I've gone off the deep end and I now think plants talk to me, but rather because there has been a pretty interesting paper published called: Experience teaches plants to learn faster and forget slower in environments where it matters:
The nervous system of animals serves the acquisition, memorization and recollection of information. Like animals, plants also acquire a huge amount of information from their environment, yet their capacity to memorize and organize learned behavioral responses has not been demonstrated. InMimosa pudica—the sensitive plant—the defensive leaf-folding behaviour in response to repeated physical disturbance exhibits clear habituation, suggesting some elementary form of learning. Applying the theory and the analytical methods usually employed in animal learning research, we show that leaf-folding habituation is more pronounced and persistent for plants growing in energetically costly environments. Astonishingly, Mimosa can display the learned response even when left undisturbed in a more favourable environment for a month. This relatively long-lasting learned behavioural change as a result of previous experience matches the persistence of habituation effects observed in many animals.
Is this the beginning of the Day of the Triffids? Invasion of the Body Snatchers? Probably not, but it's still interesting.


To understand why researchers are even asking if it's possible for plants to learn it is worthwhile to look at some of the discussion that has taken place on the topic in the years leading up to this. Although there have been attempts to research more fringe topics of plant consciousness in the history of science, some of the more respectable ideas came more recently and were quite well-described by Anthony Trewavas in his seminal paper: "Aspects of Plant Intelligence".

I recommend reading the whole paper if you have time but Trewavas basically starts with Stenhouse's definition of intelligence, which is "adaptively variable behavior within the lifetime of an individual", and then attempts to collate research in plants under this framework of 'intelligence'. In my opinion, Stenhouse's definition of "intelligence" is actually closer to a decent definition of "learning" and so even though learning requires some form of intelligence, it would be more meaningful to reserve "intelligence" for more higher order behavior, like concept formation, theory of mind, or at least something like operant conditioning.

Regardless, one of the interesting points Trewavas makes early on is that we tend to think of plants as inanimate things and so it seems odd to think of them as 'behaving', nevermind learning or acting intelligently, but he urges us to instead think of them as still behaving - just on a slower timescale than what we're used to. And this makes sense, when you watch a timelapse video of a plant it does look something like trial and error learning in response to the environment.

This line of reasoning led to some people arguing that we should be talking of ' plant neurobiology' and although the concept initially sounds a little wacky, the main criticism from other biologists isn't that such a classification is wrong but simply that it doesn't add to our understanding in any meaningful way. This is an unfair conclusion to reach, in my opinion, as Trewavas and the advocates for plant neurobiology are at the core of it just arguing for the position that our current cut-off for understanding learning, intelligence, and neurobiology, is potentially arbitrary and misleading. 

If they're right then the re-classification isn't meaningless but instead would add greatly to our understanding of those concepts, as currently we are necessarily ignoring vast amounts of possible data.


This brings us to our current study which asks the question of whether plants are capable of classical conditioning. I assume the initial attempts to get them to salivate upon hearing the sound of the bell proved to be less than successful, so instead they studied the Mimosa plant which has a reflexive response to mechanical interference where it closes up its leaves (theorised to be an adaptive response to protect from predation). 

What the researchers did was to look at how these plants responded to water drops (which elicited the leaf retraction response) in different light conditions, as the response has been recognised to be context dependent (with high light conditions eliciting less of a response due to a lower predation risk). They found that after repeated exposure to this dripping stimulus the plants appeared to become habituated to it and showed less and less responding, which was carried over to different environments (switching light conditions) and persisted several weeks after trials. 

So by approaching the subject as if the plants were animals, they seem to have found that an 'automatic' response by the Mimosa plants could be modified over time in response to its consequences. As the authors propose (if we allow a semi-mentalistic explanation), this suggests that the plants were essentially learning that there was no need to to react to the water drops any more as they did not present a threat. 


To be honest, I love an underdog so I am tentatively impressed by these results and would like to believe that this is some good preliminary data on learning in plants. At the very least, as discussed by the authors, this research and that of Trewavas should cause us to question how we arrived upon our current understanding of neurobiology and learning - are our positions based on the fact that they are objectively true or are they remnants of a scientific history which was inherently tied up with anthropomorphic interests?

In summary, the really interesting part of this research isn't the specifics of the claims themselves but rather the implications given that these results are true. The idea that plants might be capable of learning could open up the possibility of understanding behavior in an entirely new way, it could even cause us to question the possibility of pain in plants and unearth a lot of complicating ethical issues. 

Most shockingly of all, what if we were to find out that Soylent Green
is people... Okay I promise, no more movie references.


  1. I don't think this is the first plant conditioning thing. Now I have to poke around the interwebs to see what I can find.

    I highly object, however, to the notion of plant neurobiology. I know it sounds sexy (as paradoxical statements often do), but the BIG point should be that plants can do it without neurons.

    1. Hi Eric - yeah I don't doubt that there have been previous conditioning studies done with plants, I just really enjoyed this one and it's a recent study so seemed like a good point to start from.

      I get your point about the term 'plant neurobiology', I'm a bit torn myself. On one hand, I appreciate the proponents' attempt to return to an older, more broad concept of "neuron" that covers aspects of plant biology but you do make a good point about an appropriate emphasis being on not needing neurons.

      I suppose the position your coming from is similar to the criticism of some of the heavier "neuroscientific" approaches to things like pain, where they will attempt to rule out an organism feeling pain because (for example) it doesn't have a developed enough CNS. In that comparison situation, I'd say you'd be right in saying that we shouldn't be trying to redefine the CNS to include these other animals but rather we should be pointing out something like the behavioral data to refute behavioral claims.