SNAILS, jellyfish and starfish have taught us that you don’t need a brain to learn. These seemingly simple creatures are capable learners, despite being completely brainless. Perhaps this is no great surprise. After all, it’s not as if they lack nerve cells. Strictly speaking, it’s neurons that enable learning – theirs are simply spread out, rather than being packed into centralised bundles.
But what if you take away the neurons?
Advertisement
Most life forms on Earth lack neurons, and yet they frequently manage to behave in complex ways. Previously, we have chalked this up to innate responses refined over generations, but it is beginning to look as if some of these humble non-neural organisms can actually learn. While that’s left some scientists scratching their heads, others are busy investigating how this ability could offer new approaches to fighting diseases and designing intelligent machines.
Take a slime mould. It certainly doesn’t look smart. This unusual creature, which is not a plant, animal or fungus, often resembles a glob of lemon curd that has fallen on the floor. Really, this manifestation is just one stage in the slime mould’s life, formed when many single cells, each with their own distinct DNA, mingle and fuse. The resulting yellow blob can grow to a few square metres, and is just one enormous cell containing thousands of nuclei.
In nature, a slime mould relies on chemical receptors on its surface to sense substances in its path as it creeps along the forest floor. If it gets a whiff of something attractive, like food, it will rapidly pulsate, pushing itself closer to the source. Toxins elicit the opposite response, causing the slime mould to slow down its rhythmic throbbing and retract from potential harm.
“Slime moulds aren’t just capable of learning, they can teach each other too”
They may be sluggish, with maximum speeds of around 4 centimetres an hour, but the fact that slime moulds are mobile has allowed researchers to get creative with their experiments. Last year, Audrey Dussutour grew some in giant Petri dishes in her lab at the University of Toulouse, France. Then she prepared a feast of blended oats and placed it strategically out of reach, save for a bridge that the slime moulds could crawl across. Their path was clear. That is, until Dussutour polluted the bridge with bitter compounds such as caffeine. Although the concentration was not high enough to harm the slime moulds, it was sufficient to stop them in their tracks for several hours. Eventually though, the lure of a meal induced them to push past.
As time went by, the slime moulds began to cross the bridge more quickly. After a few days, the caffeine was no longer a deterrent: . “Learned” is the key word here. More specifically, they had habituated – a simple form of learning where the response to an irrelevant cue weakens over time. “I was surprised because they don’t have neurons and everybody was saying that ability relies on a neural system,” says Dussutour.
If neurons weren’t enabling the slime moulds to learn, what was? Dussutour admits she doesn’t know. But one idea she hopes to test is that their experience modified their genes. In a cell’s nucleus are swarms of molecules that can stick to DNA and switch genes on or off. They don’t rewrite the genetic code, but they do temporarily alter how it’s read. This process – known as epigenetic regulation – satisfies one of the most basic requirements for memory and learning, according to neuroscientist David Glanzman at the University of California, Los Angeles. “All you need is to have some information in the cell changed.” His research suggests this might be how slime moulds learn.
Neuroscience tells us that the memories created when an animal learns something get stored in the synapses between neighbouring neurons. So when Glanzman took sea slugs and broke apart the synapses that had formed during training, he was not surprised to find that the memories vanished. However, when he gave the neurons involved a jolt of electricity, the synapses . Glanzman believes that remnants of the memories had been stored in the neurons’ DNA, in the form of epigenetic changes. There is no reason why similar changes couldn’t occur in other cell types, he says. “You don’t really need a nervous system to learn.”
And with slime moulds, it’s not only about learning. Dussutour has discovered that . In a second experiment, she allowed slime moulds habituated to salt to fuse with others that had never encountered the deterrent. These hybrids scurried across a salt-laced bridge without hesitation. After three hours, she tore the companions apart and the once-naive slime mould continued to ignore the salt when crossing the bridge, as though the habituated mould had somehow transferred its learning.
“One might speculate that what is getting exchanged during the fusion is epigenetic information,” says Glanzman. Eva Jablonka, an epigeneticist at the University of Tel Aviv, can see how a single cell such as a slime mould could learn through such modifications to its DNA. But a brainless multicellular organism is another matter. “It’s not just that there are more cells, but they also have to behave in a coordinated manner,” she says. “When you don’t have a nervous system that can integrate and coordinate, how is that happening?” That makes the research coming out of Monica Gagliano’s lab at the University of Western Australia tough to explain.
Smarty plants
A few years back, Gagliano’s team repeatedly dropped potted mimosa plants 15 centimetres to a soft landing pad on the floor. When mimosa plants are disturbed, they curl up their leaves as though hiding in fear. But after surviving about five drops unscathed, they stopped withdrawing their leaves. They had learned that they could safely ignore the fall. Even after being left undisturbed for a month, a reprise of the dropping experiment failed to elicit the slightest response. They had remembered their lesson.
Encouraged by the plants’ ability to habituate, Gagliano wanted to see if they could do more. Could they learn to associate a reward with a neutral cue, as Pavlov’s dogs had learned to link food with the sound of a bell?
A plant will naturally grow towards light. So to train her pea plant seedlings, Gagliano put them in the dark and then shone a light (the reward) towards them from one direction and blew a fan from the opposite direction. She also switched it up for some seedlings by blasting the light and the air from the same side. Once their training was done, she removed the light and hit all the plants with a fan. Those that were accustomed to light and wind originating from the same side grew towards the fan, while those that had experienced the light and wind from different directions grew away from it. The plants seemed to be seeking their reward. They had .
Other scientists have greeted the tantalising results with a healthy dose of scepticism. A leading critic is plant physiologist Lincoln Taiz, now retired from the University of California Santa Cruz. He thinks Gagliano’s initial evidence for habituation was tenuous and she should have shored it up before building on it. He also points out that in associative learning experiments with animals you would expect around 90 per cent of subjects to respond, whereas just over 60 per cent of the pea plants did. However, the brunt of Taiz’s argument is rooted in language. According to neurobiology, “both learning and memory are mental processes carried out by the mind, which is centred in the brain,” he says. “By this definition, plants are incapable of learning and memory.”
Not everyone subscribes to such a strict definition. For his part, Glanzman is open to the idea of plants learning by association, but wonders about the mechanism. He notes that animals link events together using molecules found in nerve cells called NMDA receptors, which help strengthen connections between neurons that are repeatedly stimulated at the same time. A similar “associative molecule” would need to be operating in plants, he says.
Gagliano is equally mystified. “There must be some system that allows this memory to be recorded and literally etched into the organism, and those triggers then get recalled and those are memories,” she says.
The mechanism may be elusive, but simply realising that brainless organisms are capable of learning could have some practical payoffs. “There are lots of unicellular organisms that are very harmful for humans, like those that cause malaria,” says Dussutour. “They belong to the same group as slime moulds, and we never thought about these organisms as being able to learn.” She suggests that knowing whether and how pathogens learn could help guide new strategies for combating them.
Considering epigenetic learning could also help computer scientists improve artificial neural networks, which model biological learning. Current neural network models are based on the Hebbian theory of learning – the idea that a synapse becomes stronger when the neurons on either side fire in synchrony. In other words, neurons that fire together wire together. Jablonka believes that incorporating epigenetic memory into those models would enrich them.
“Plants can learn to associate a reward with a cue, like Pavlov’s dogs”
There’s also the provocative idea of memory transfer. If one slime mould can teach another by fusing with it, might something similar take place in animals? Experiments done by James McConnell at the University of Michigan over half a century ago suggest it might. He trained freshwater flatworms to fear light by repeatedly pairing it with electric shocks. Then he ground them up and fed them to untrained flatworms, which proceeded to twitch whenever a light flashed.
McConnell believed that the memories of the trained flatworms were encoded in small molecules that the naive flatworms then ingested. Unfortunately, his findings proved impossible to replicate. But the idea that small strands of RNA – one type of epigenetic molecule – could mobilise memories resonates with many scientists today, including Glanzman. “In principle, there should be no reason why you couldn’t transfer some aspects of memory by transferring RNA from the brain of one animal to another,” he says. The implications are mind-boggling.
Jablonka stops short of suggesting that memories could ever be transplanted, but she can imagine that it might be possible to transmit something like an increased sensitivity to a stimulus. “Once upon a time I would have told you I don’t agree with this kind of thing at all… but I think the more we’re learning, the more flexible we should be in our thinking,” she says. “There are very curious things in this world.”
Who needs brains?
Organisms with tiny brains, or no brain at all, are capable of amazing feats
Slime mould: 0 neurons
When their food is scattered in a pattern matching the layout of the cities around Tokyo, that closely resembles Japan’s highly efficient rail system.
Pea plant: 0 neurons
When allowed to grow their roots into either a pot with a steady food supply or one with a boom-or-bust supply, pea plants prefer the former if food is plentiful, but gamble on the latter if they are starved, .
Box jellyfish: ˜13,000 neurons
Box jellyfish use four of their 24 eyes to peer up through the water’s surface at tree canopies, which they use to help them .
Freshwater snail: ˜20,000 neurons
freshwater snails decide whether or not to eat. The controller neuron signals the presence of food, and the motivator neuron lets the brain know whether the snail is hungry.
Fruit fly: ˜250,000 neurons
Fruit flies take longer to distinguish between very similar concentrations of odours than very different ones, suggesting that .
Bumblebee: ˜1,000,000 neurons
Bumblebees can by observing another bee performing the task. They can also be trained to move a tiny ball to a target.
The power of association
Pavlov’s dogs learned to associate the sound of a bell with the imminent arrival of food. It’s a simple form of learning, but .
Take chimps with a knack for cracking nuts with stone tools. This precision behaviour is considered one of the most sophisticated observed in wild animals, but it might be learned as a sequence of small associative steps in a process called backward chaining. First the chimp might steal a shelled nut from its mother, so it learns to associate nuts with a tasty reward. Then, when it strikes a nutshell with a stone, that act becomes associated with the reward. Handling stones then becomes rewarding, and so on until the chimp is a proficient tool user. Establishing the sequence requires very little in the way of reasoning, but once the chaining is complete it adds up to an advanced skill.
This idea could have far-reaching implications for simple organisms. Even some plants seem capable of learning through association (see main story). So, in principle, they are equipped to acquire more complex behaviours via backward chaining. And all without even a single neuron.
This article appeared in print under the headline “It’s a no-brainer”
