Is our bacteria learning?

This is a cool story, but not for the reason the authors are attributing. Researchers at Princeton showed that bacteria can evolve to anticipate future environmental changes. Here is the coverage in Science:

Researchers already know that microbes can mount simple responses to changes in their environment, such as acidity fluctuations, by altering their internal workings. If the changes are regular enough, bacteria can respond ahead of time. But systems biologist Saeed Tavazoie of Princeton University wondered if microbes were capable of more sophisticated reasoning. Could they, for example, learn to match a signal that didn't occur regularly to a probable future event? If so, the bacterium could improve its chances of survival by turning on a preemptive response to that event.

Tavazoie and colleagues first ran a computer simulation to determine if a simple system could evolve such behavior. They created an environment inhabited by evolving virtual bugs. The organisms garnered more energy if they could "learn" that certain signals preceded the arrival of food and launch a preemptive metabolic response. Even when the signal combinations grew more complex, the population was able to evolve the correct responses, the team reports online this week in Science.

The researchers then looked for evidence of this ability in the bacterium Escherichia coli. Because E. coli gets warmer when it enters a human mouth--ferried in on some old meatloaf, perhaps--and then must soon contend with low oxygen levels as it passes into the large intestine, the team reasoned that the bacterium might use temperature as a cue to prepare for the upcoming lack of oxygen. Indeed, when the researchers turned up the heat in a dish of E. coli, the bugs dialed down activity in genes that normally operate in high-oxygen conditions. But the true test came when the team flipped the normal association, growing the bacteria in conditions in which high oxygen levels followed temperature increases. Less than 100 generations later, the bacteria stopped turning on their low-oxygen response after exposure to high temperatures, suggesting that they had evolved to break the association. (Emphasis mine. Link in original.)

The paper itself is here.

Here's the deal. This is definitely cool for two reasons. First, it shows that cross-over experiments using the predictions from computational models can identify interesting results in the lab. Very rarely in my experience do labs employ both computers and bench work in the same paper. That is impressive. Second, it also suggests that bacteria can evolve some incredibly elaborate behaviors to respond optimally to their environment.

However, the key word in that sentence is evolve. This is evolution, not learning.

The OED defines learning:

1. The action of the vb. LEARN. a. The action of receiving instruction or acquiring knowledge; spec. in Psychol., a process which leads to the modification of behaviour or the acquisition of new abilities or responses, and which is additional to natural development by growth or maturation; (freq. opp. insight).

From the perspective of a behavioral neuroscientist, learning is a persistent, measurable change in behavior resulting from previous experiences. The important part is that it must happen during the lifetime of an individual organism. Surely, there are examples of evolved behaviors; the entire subject of ethology is devoted to understanding evolved behavior. However, all of those behaviors are present when the animal is born; they are by definition unlearned. (Incidentally, yes, I do know that an animal can become more proficient at ethological behaviors. That does constitute learning. However, they are becoming more proficient at a behavior present at birth.)

By virtue of the fact that no individual E. coli lived long enough to match its behavior to the new environment, learning did not take place in this experiment. A very cool evolutionary trick happened, but not learning.

Maybe I am just being unnecessarily nitpicky, but I really have a problem with that word choice, even it is in quotes.