We're all familiar with Pavlov's conditioning experiments with dogs. Dogs were treated to an unconditioned stimulus — something to which they would normally respond with a specific behavior, in this case, meat juice which would cause them to drool. Then they were simultaneously exposed to the unconditioned stimulus and a new stimulus, the conditioned stimulus, that they would learn to associate with the tasty, drool-worthy stimulus — a bell. Afterwards, ringing a bell alone would cause the dogs to make the drooling response. The ability to make such an association is a measure of the learning ability of the animal.
Now…how do we carry out such an experiment on a cephalopod? And can it be done on a cephalopod with a reputation (perhaps undeserved, as we shall see) as a more primitive, less intelligent member of the clade?
The nautilus, Nautilus pompilius despite being a beautiful animal in its own right, is generally regarded as the simplest of the cephalopods, with a small brain lacking the more specialized areas associated with learning and memory. It's a relatively slow moving beast, drifting up and down through the water column to forage for food. It has primitive eyes, which to visual animals like ourselves seems to be a mark of less sophisticated sensory processing, but it has an elaborate array of tentacles and rhinophores which it uses to probe for food by touch and smell/taste. Compared to big-eyed, swift squid, a nautilus just seems a little sluggish and slow.
So let's look and see how good a nautilus's memory might be. First, we need a response to stimuli that we can recognize and measure, equivalent to the drooling of Pavlov's dogs. While they don't measurably salivate, the nautilus does have a reaction to the hint of something tasty in the water — it will extend its tentacles and rhinophores, as seen below, in a quantifiable metric called the tentacle extension response, or TER.
The scoring system for tentacle extension response (TER) in chambered nautilus. TER was graded every 5 s from a minimum score of 0 to a
maximum score of 3. Each level corresponds to a range of percentage extension relative to the length of the animalʼs hood. Zero is recorded when all
tentacles are retracted into their sheaths. A score of 1 corresponds to an extension of <33% of the hood length. A score of 2 corresponds to extension
between 34% and 66%, and 3 is given when tentacles are extended beyond 67% of hood length.
The nautilus will extend those tentacles when treated to an unconditioned stimulus: a spritz of fish juice. Now we just need an unconditioned stimulus. A bell won't do, but those simple eyes can respond to light, so investigators use a blue light, one similar in wavelength to the glow of luminescent bacteria.
Now we have a nice clean experiment: strap a nautilus to a bit of pipe in a tank. Shine a blue light at it, then give it a squirt of fish juice, which prompts it to splay out those tentacles. Do that ten times at an interval of a few minutes, and then let them rest for a period of time. Then the test: shine the blue light on the animal, without the fish juice, and see if it extends its tentacles.
The answer is yes, it does. The nautilus learns fairly quickly to associate a light pulse with the prospect of an imminent food stimulus, as measured in two different ways, the tentacle extension response and the ventilation rate (the animal breathes faster if it thinks it's going to get a whiff of fish.)
N. pompilius expresses a bi-phasic memory curve. There was expression of two distinct memory peaks in animals conditioned with a light pulse plus food odour association (CS+, black bars; N=6) for both
behaviours we measured: (A) tentacle extension response (TER), and (B)
ventilation rate. The short-term memory peak was expressed at 3 and
30 min after training, whereas a long-term memory peak was apparent at 6
and 12 h after training, and persisted for less than 24 h. The control group
(CS-, white bars; N=6) received a light pulse combined with home-tank
water during training and showed little change in behaviour over the six
test periods. Graphs show the mean ± 1 s.e.m. Asterisks above the bars
denote a significant difference (P≤0.05) between CS+ and CS- measures
at each time interval. The linking bars below the x-axis show significant
differences (P≤0.05) within the CS+ group compared across test
procedures at each retention interval. There were no significant differences
in the CS- group across test periods.
Those graphs illustrate another interesting phenomenon; what's being varied is the period between the training with the combination of the conditioned and unconditioned stimulus, and the presentation of a series of blue lights, the conditioned stimulus, without the reward. The response is biphasic. They remember the association well in a window of time of less than an hour after training, and then they seem to forget … and then they respond strongly again 6-12 hours after the training. This looks like a two-step memory, with short term and long term components; there is rapid assimilation of an association, and a different process that leads to long term consolidation of the lesson.
These results are comparable to similar results obtained in the cuttlefish, except that long term memory in the nautilus is relatively shorter. That's impressive, since the nautilus lacks the brain structures (the vertical lobe complex) that are considered important for learning and memory in other cephalopods. That suggests that this is a mechanism that may have evolved independently in the nautilus, or that the brain structures in more visual cephalopods are refinements and elaborations of a primitive ability.
I'm also impressed with their ability relative to certain vertebrates. I've supervised several student projects that have tried to assess learning and memory in zebrafish, and they are much more difficult. We have the complication that these fish lack a simply assessed unconditioned response; they have to be trained, for instance, to swim to a particular area in a tank to show their interest in a stimulus, and their behavior is complicated by schooling and other confounding social behaviors — it usually takes a few weeks of training to get a measurable response to something like a light stimulus. I'm a bit envious of an animal and a protocol that can show a response to a conditioned stimulus within minutes after a relatively short training period; fish are such flaky flibbertigibbets when it comes to this sort of thing.
Crook R, Basil J (2008) A biphasic memory curve in the chambered nautilus, Nautilus pompilius L. (Cephalopoda: Nautiloidea). J Exp Biol 211:1992-1998.
- Log in to post comments
(.)(.)
This seems like operant conditioning rather than Pavlov's classical conditioning. I do sometimes wonder whether there is really much of a difference between them, and think it would be interesting to use Skiner's schedules of reinforcement using classical conditions. I suspect you would get the same results. Maybe it has been done?
You mean Pavlov's experiment wasn't done just to laminate the floors with dog drool?
But anyway, if flies can do it, I'd hardly be surprised that nautilus's can:
Glen Davidson
http://tinyurl.com/2kxyc7
OT as can be, but I'm sure people would like to know:
Glen D
http://tinyurl.com/2kxyc7
So it seems that Pavlov-like experiments can prove conclusively enough that a learning process can be associated with even simpler brain developments.
Now I'm only curious as to how undeveloped must the brain be to not have a learning ability (more precisely, I am asking when did this learning process start developing into entities with a nervous system, or a compatible replacement, and why).
Also, there's a small mistake in the text:
Second one should be conditioned
By the way, another question, related to the bi-phase memory of the nautilus.
Since the experiment points towards two separate components of memory, is it possible that the nautilus is actually a transitional "not-yet-fossil" between development of short-term memory (which, for reasons I cannot yet explain, I currently presume to have developed first) and a more advanced memory like, say, ours ?
By the way, if I am wrong in my presumption, please point me the right way.
Makes me curious as to whether the short term/long term memory division is an example of convergent evolution or of descent from a common ancestor that had this same divide.
If a Christian can be conditioned, of course a 'mere' cephalopod can! Although Christian conditioning is a bit more sophisticated. Show them a cross with an invisible man and spout nonsense, they will follow and throw money. You can even exchange the invisible man for an invisible elephant and they will still follow and throw money. :)
Your entries have made me fall in love with Cephalapods!
Nautiluses (Nautili ?) are pretty cool.
Growing up, I used to see these nifty animals in the "prehistorical seas" pictures in stereotypical dinosaur books and had assumed that they were extinct. Then, several years ago, I was visiting the Monterrey Bay Aquarium and was floored when I saw a group of them floating around in a tank. That was a seriously heavy moment when I discovered that they still existed.
-Richard
The ability to learn seems to be a useful feature that has been around for a very long time and has not yet been selected out by the environment.
I find this article to be very interesting, but not entirely surprising. There is so much more to learn about intelligence in all animal species (Are plants next? I dare to wonder.).
Seeing investigators set puzzles for octopi to solve gave me pause as I contemplated yet another example of how easy it is to challenge our notions of what other organisms are capable of intellectually.
Tool use used to be considered the hallmark of human divergence from other species. Now, no longer. In fact, I once saw a blue heron (I think that is what it was) take a piece of bread, toss it in the water and wait for a fish to get the bread.
On the lawsuit issue, here's the rest of the story:
Glen Davidson
http://tinyurl.com/2kxyc7
A google search of "can plants learn" revealed this paper:
http://www.pnas.org/cgi/content/full/96/8/4216
I also found a pdf which mentioned the above and a few other papers:
http://eprints.iisc.ernet.in/archive/00005656/01/Do_plants.pdf
I'm completely unqualified to comment on any of it, but it at least appears that people are researching plants' ability to learn.
#8: "Although Christian conditioning is a bit more sophisticated. Show them a cross with an invisible man and spout nonsense, they will follow and throw money. You can even exchange the invisible man for an invisible elephant and they will still follow and throw money. :)"
"fish are such flaky flibbertigibbets when it comes to this sort of thing."
I suspect the Christians are more like fish... flibbertigibbets.
I did aversion training on flatworm groups as a freshman. Planaria have such small processors that they are called ganglia rather than brains, and they can be taught to not go to food if the food is on a quarter.
Embarrassingly, some of the flatworms found a way to the food that did not require going on the quarter to get it, and an approach that even bigger-brained animals rarely use spontaneously - they went away from the food, up the dish side, across the surface tension of the water, and tried to reach the food from above. Even I hadn't anticipated that route (but freshmen only have slightly bigger ganglia than flatworms).
Take all measures of animal intelligence with skepticism - they is not so dumb.
sailor @#2,
This is definitely Pavlovian (Classical) conditioning, not operant. Simple presentation of the light contemporaneous with the food, and then see the food response when only the light is presented.
Operant conditioning would be waiting for the nautilus to do something specific, and then present a "reinforcer" in an attempt to increase the incidence of the target behavior.
Pavlov, incidentally, was not interested in conditioning when he started his experiments, he was interested in saliva production.
I do wonder, though, if your fish experimenters are relying on food as stimuli for the classical conditioning experiment. It doesn't have to be food -- any response you can elicit could serve to show classical conditioning. Personally, I would go with a startle response with fish, I would think. Shine the light, then disturb the surface of the water, and then later see if you can get the fish to scatter (or do whatever they do) just by shining the light.
Yes, I was in psychology, but focused as purely as I could on experimental psych.
Great, i always like to see a experience
that simple with useful findings... where we
arrive to that many more questions...
Inspire me to my tesis to come... do
something awesome with little money.
Mmm, we need more research in that brain,
if there is a different estructure involved
that would be excelent...
Darby (#15),
Is it true that you can chop up a clever flatworn, feed it to it's fellows and then observe them sharing the late one's memory? Or am I thinking of something else entirely?
I though it wasn't possible to bread nautiluses in captivity, so these are 'freshly caught' so to speak, right?
While I do realise that the unconditioned critters don't show any statistical difference in their responce, I can't help but think that it looks like they do have a long term memory of the of the light in terms of their 'stress' level. They look sorta agitated in the first hour (they breathe more), but after six hours they're pretty blazé about the light only to become stressed again later.
Or is that too fanciful an interpretation?
Richard: They used to have a sibling clade called "ammonites", apparently similar except that their shells were straight (long cones). Those are (AFAIK) extinct.
Some were straight, many were coiled.
Ebay has lots of the coiled shells for sale.
All of the ammonites went extinct at the Cretaceous-Tertiary boundary.
Glen D
http://tinyurl.com/2kxyc7
This is a very interesting study. I'm a learning theorist (trained in both but study mainly operant or operant/pavlovian interactions) and can confirm that this is a traditional respondent learning paradigm. There are procedural aspects that need to be clarified to better identify the learning mechanism (such as the timing of the blue only tests - if they were preditable, then stimulus control could be involved; if they were unpredictable but blue tests occurred for an extended duration, then extinction/spontaneous recovery could be involved; habituation also would need to be ruled out). Remembering is probably easiest to study in a purely operant paradigm such (delayed) matching to sample which I would imagine could even be used with these creatures (I've seen clear evidence of operant conditioning in flies that live no more than 24 hours). However, operant learning is not studied solely by waiting for a response to happen and delivering a reinforcer. In fairness to Crypticlife, this was Skinner's original preparation and probably the most commonly referenced operant learning paradigm. This is, however, just one way of studying free operant behavior. Restricted operants such as learning in a match-to-sample task or schedule performance in a drug discrimination paradigm involve signals that evoke responses - that's evoke not elicit. Elicit is properly reserved for responses solely caused by antecedent stimuli and are insensitive to consequences (stimuli that follow responding). I mainly work with children and prompting responding (and minimizing the likelihood of errors) is much more efficient than waiting for it to occur and reinforcing. Not to mention that what will effectively reinforce is often dynamic.
Awesome. I learn so much about cephalopods coming here...and now I know where to start building my animal army of global conquest!
Moo hoo ha ha ha, etc.
I've been conditioned by your blog, PZ. When ever I see that image (http://bpr3.org/images/rbicons/ResearchBlogging-Large-Trans.png), I respond, "Ooh! Science! Yay, there's gonna be science!" If you put that pic next to a post about what you ate for breakfast, I'd be all exited, then confused, then disappointed: "No science? D: "
Aww, I was hoping it would be octopus Grigori.
Another fine write-up, though!
Sili, yes, you can feed trained to untrained and pass the training. As I recall, they did that with successively smaller bits until it was determined that just extracting the RNA from the ganglia could do it - started a search for RNA-based memory in mammals that never came to anything.
As a freshman, I knew I couldn't recreate the full breadth of the studies, so I thought I would just study the training technique itself.
Slime is slime. Right?
Whether from peops or snails
The game must be played
Cool, at any rate.
Has anyone else noticed that PZ looks like colonel Sanders of KFC fame?
Darby #25- so RNA was transmitting the knowledge?
What exactly did this knowledge consist of? Might have just been triggering behaviors based on the presence of that RNA, as if certain genes had been activated.
IIRC the knowledge was of how to pass through a very simple maze.
Makes sense that organisms without a centralized neural system relies on "cellular intelligence".
That was pretty funny, when you started talking about students and zebrafish, I (for a moment) thought you were going to complain that your students were harder to teach/train/condition than the cephalopods :-)
Thank you, PZ, for another nice biology lesson, exemplary for those who might associate big human brains with superior learning capability -- although it WAS humans designing experiments on cephalopods and not the other way around. (Oops, will that get me banned from this blog?)
Is the theme of Christian conditioning, introduced by mothra and carried on by others in this thread, more than just a humorous note? Don't the conditioned Christians resemble fish in a school? Fish schooling is innate; social grouping in humans is as well. Repetition of the cherished beliefs of family and community is rewarded and reinforced in many ways from childhood on up. Conditioning experiments isolate and quantify a mechanism of learning that in populations, not only human communities, amounts to cultural conditioning.
If conditioning as an isolated psychology experiment was designed and controlled by a scientist, who or what designs and controls religious conditioning? For nautilus or dog or human, conditioned learning can be adaptive to environmental challenge, except humans are at risk of being influenced and controlled by others of their own kind thru conditioning.
An apparent difference between fish schooling and religions is that fish don't have obvious leaders or exploiters. Are belief systems simply self-perpetuating, whether or not there are teachers? The principles of science are culture. We know that it's very hard to see beyond one's culture.
Science and critical thinking are pretty good antidotes to older, cherished-but-outmoded belief systems, But science is an evolving cultural phenomenon -- better, of course, but not so different from previous belief systems, like the Christian one.
John, I don't remember whether anyone took the research in that direction, and it was long enough ago that I don't think anyone reduced the RNA component to just what bit of RNA might be doing it.
The knowledge itself were stimulus-response type things, similar to what I described as my little replication.
The ability to make such an association is a measure of the learning ability of the animal
Nonsense. It's a measure of a dog's stupidity! He can't eat a bell, and he sure as hell can't eat the sound of a bell.
Now if meat was frozen into the shape of a bell, and the dogs would learn to defrost it, shape it into patties and broil it, then I'd be impressed, if they didn't over-cook it.
So what, these Commie dogs could droll at a ringing bell. I know dogs who drool at the drop of a hat, and that hardly makes a sound.