Observational Learning in Octopuses

ceph

Wonderpus
Staff member
Moderator (Staff)
Joined
Dec 20, 2002
Messages
226
Location
West Palm Beach, Florida
This study (http://www.sciencemag.org/cgi/content/abstract/256/5056/545) appeared in the prestigious journal Science in the early 90's and is now taken as fact by media and members of the public.

My opinion is that the experiment found something significant but did not have proper controls. The results have not been replicated. Most octopuses are solitary, not group living animals - chances to learn through observation are not common. Running a controlled experiment with cuttlefish, which do live in groups, also failed to produce evidence of observational learning. I believe that the results of the initial study are real but are not best explained by observational learning. At the very least, we should not accept or perpetuate that observational learning is an undisputed fact.
 
The infamous Fiorito and Scotto experiment... James' post reflects my opinion. At the same time, I have to admit this is my favorite ceph article and after hearing about it, I decided to study cephalopods professionally. Of great interest is a follow-up review of this article:

Social Learning in Invertebrates
Author(s): Gerald B. Biederman, Valerie A. Davey, M. D. Suboski, D. Muir, D. Hall, Graziano
Fiorito
Source: Science, New Series, Vol. 259, No. 5101 (Mar. 12, 1993), pp. 1627-1629
Published by: American Association for the Advancement of Science
Stable URL: Social Learning in Invertebrates on JSTOR

Is it social learning or is it imitation? The difference is intent.

Has anyone tried to replicate this with the same species, O. vulgaris?

Great thread, an important issue in invertebrate research.
 
I've certainly seen this referenced frequently.

Do you have a link to the full paper? And can you describe the problems with the controls? The lack of replicability is problematic, obviously.

The one at the end entitled "Response" might be interesting; I no longer have a subscription active there.
 
I agree with both James and Stavros - as I recall the authors failed to train against a strong pre-existing bias toward choosing the red (?) ball, and did not control for familiarity (since observers would presumably spend more time attending to the ball where the action was, that which was being attacked by the demonstrator) by simply presenting a ball of one colour in the demonstration tank without it being associated with any reward.

I agree with Stravros in that this was also the paper that originally sparked my interest in cephalopods.

It irks me that this paper is referenced uncritically by a number of authors whose editorial bent is toward attributing higher intelligence or consciousness to cephalopods. In several cases I have seen it presented as the most impressive among a number of skills of putative increasing impressiveness (conditioning, long memory, visual discrimination, observational learning (!!!!) etc.).

I think it is an interesting paper with considerable merits, but the interpretation of the results may have suffered from some overreaching.
 
robyn;160661 said:
I agree with both James and Stavros - as I recall the authors failed to train against a strong pre-existing bias toward choosing the red (?) ball, and did not control for familiarity (since observers would presumably spend more time attending to the ball where the action was, that which was being attacked by the demonstrator) by simply presenting a ball of one colour in the demonstration tank without it being associated with any reward.

In their rebuttal, they offer:
According to Biederman and Davey, our description of “imitation” in octopus is complicated by the observers’ preference for red. This does not interfere with observational learning, as both red and white (or dark and light) copying are obtained without significant variation for 5 days. Furthermore, as we reported [note 9 in (1)], octopuses seem to prefer red over white only after 5 days of a free-choice experiment [note 12 and table 1 in (1)], which we discussed according to an innate visual preference.

The big tell, it seems, would be the failure of others to reproduce these results. Above, ceph mentioned cuttlefish; has anyone tried the same thing with common octopuses?
 
After seeing an interesting, undisputable interaction between a seahorse and a mandarin (neither of noted intelligence, no matter how you define it) I am convinced that lower animals observe and learn from each other for survival and food location. If they only learned from direct experience, the typical food item would not exist and the differences in how animals act and react suggests that behavior is at least partially learned. Learning from each other (nothing else alive in the tank) was also clear in a pair of mercatoris I raised. One was very shy where the other was very interactive. The shy one would mimic the interactive one over time (but never initiated its own interaction). In both my hobbiest observations, obtaining food was the driving force. Distinct intelligence was not part of the equation (the octopuses were O. mercatoris, not one of the more sophisticated octos) so I find it peculiar that the paper suggested this was an intelligence test (or that it is used as such) and that there would be much challenge to animals observing and learning how to obtain food. The most puzzling part to me is that they don't seem to be able to replicate the observed learning behavior.
 
I just returned from the European Conference on Behavioral Biology in Ferrara at which Fiorito presented another version of the observational learning study. In this one the O. vulgarus observes another animal open one of two drawers to get to a food reward. It was a short video clip, so I can't say much about the controls, but I suspect we will see more published on this.

Roy
 
dwhatley;160685 said:
...If they only learned from direct experience, the typical food item would not exist and the differences in how animals act and react suggests that behavior is at least partially learned.
What do you mean by "the typical food item would not exist"? I'm not following.

dwhatley;160685 said:
...the differences in how animals act and react suggests that behavior is at least partially learned.
Do you mean differences between how different individuals act/react, or between how a given individual acts/reacts now and how that same individual behaved in the past?
 
Any small animal/fish would be what I would call a typical food item. In the underwater world you really only get one chance so learning from your own experience is not an option. I believe observing interaction and repeated attacks by preditors is what allows many of the animals to survive.

I was referring to individuals acting, individually and "discovering" and learning to live in their explicit environment vs instinct that would be passed to all individuals.
 
From a scienfific point of view, observations of interactions of individual animals are invaluable as they provide ideas and insights. I think observations are critically important and the method that scientists actually use isn’t the linear scientific method taught in schools. The challenge is that an observed behavior can have many different plausible explanations. The “art” in science is designing a controlled experiment that eliminates alternative explanations. This is often harder than it sounds.

Our brains have evolved to connect the dots and form explanations. Generally an observation of a single animal or interaction is not sufficient to form a conclusion and explanation although it can disprove a theory. For example, the theory that life does not exist in the deep-sea or all life uses energy from sunlight can be disproved by a single observation of a deep-sea vent community. Most of the time, it takes carefully controlled experiments to explain a perceived pattern.

Take a close look at the cuttlefish observational learning experiment below. Lets break it down into parts.

When first presented with live crab prey, naive cuttlefish typically approached from the front and were often pinched. In subsequent trials, this initial group rapidly improved their prey capture techniques and attacked from above or behind the crab. Naive cuttlefish that first watched experienced conspecifics prey on crabs captured crabs without getting pinched.

What if we stopped there? What if the experiment was only run with naive cuttlefish and those that observed other cuttlefish? Does this sound like observational learning to you? It does to me. . . But keep reading. . .

However, naive cuttlefish that first watched non-attacking cuttlefish in the same tank with crabs also avoided pinches, as did naive cuttlefish that were exposed only to crab odor. All three experimental groups were as successful on their first predation as the initial group was on its second predation. . .

Would you still conclude that observational learning had occurred? If so, how do you explain the improvement based on olfaction? Is your explanation different once you consider the results of the controls? How important are those controls to how you interpret the results? What if Boal et al hadn’t run those controls?

Animals do amazing things and cephalopods are especially intriguing. However, scientists have to be careful of how we connect the dots. We build on each others work and challenge each other. Science is an evidence based human endeavor, it is not infallible or perfect, not always popular (e.g. Galileo) but it is one way to explain the natural world. Often controlled experiments with a number of different animals are needed to eliminate alternative explanations.

Here is the full reference and abstract which I quoted above:


Observational learning does not explain improvement in predation tactics by cuttlefish (Mollusca: Cephalopoda)

J. G. Boal, , K. M. Wittenberg and R. T. Hanlon
Marine Biological Laboratory, Woods Hole, MA 02543-1015, USA
Received 17 August 1999; revised 10 August 2000; accepted 18 August 2000. Available online 26 January 2001.
Abstract

When first presented with live crab prey, naive cuttlefish typically approached from the front and were often pinched. In subsequent trials, this initial group rapidly improved their prey capture techniques and attacked from above or behind the crab. Naive cuttlefish that first watched experienced conspecifics prey on crabs captured crabs without getting pinched. However, naive cuttlefish that first watched non-attacking cuttlefish in the same tank with crabs also avoided pinches, as did naive cuttlefish that were exposed only to crab odor. All three experimental groups were as successful on their first predation as the initial group was on its second predation, but the attack techniques they used were not as well developed as those of the initial group on their fifth trial. Results suggest that odor may serve as a primer for cuttlefish predatory attack behavior, perhaps by enhancing food arousal and improving attention. Practice was required for further improvements in predation techniques. We found no evidence that cuttlefish improved their predation techniques by observing conspecifics.
 
For as much as it might count, the authors reproduce their own results in a different group of animals in "Lesions of the vertical lobe impair visual discrimination learning by observation in Octopus vulgaris." Lesions of the vertical lobe impair visual discrimination learning by observation in Octopus vulgaris - PubMed

While it doesn't address the problems in this line of behavioral research, they found that their "observational learning" was affected by vertical lobe lesions in the same way that sensory discrimination learning is, suggesting that the effect is dependent on the memory system to some extent (and perhaps not just an attentional bias caused by the experimental conditions - the presence of the other octopus or of food items.) - Maybe. It's still an under-explored topic.

I also get annoyed when people try to use that result to call octopuses "intelligent", as if we need to convince the world that octopuses are like people to make them worth caring about.
 
To some (and there are a lot of these people) the ocean is a mysterious place to be avoided and to avoid anything that moves that lives there. I have seen this lack of interest and fear and never can help feeling at a loss for words to try to get them to see things differently. We were visiting our local aquarium a couple of years ago and going through the under aquarium tunnel. A child in front of me stopped and excitedly pointed to one of the large creatures that had come to the viewing window. The mother grabbed him by the arm (obviously anxious to be out of the tunnel) and said, "THAT is why we don't go into the ocean" (of course my thinking was, NO! THAT is WHY we go into the ocean but I said nothing). It was a candid moment because my family and friends all IMMEDIATELY turned to look at ME to see how and if I would react.
 
The full article, addressing the actual performance of the groups of cuttlefish and the experimental setup, is here:

http://www.millersville.edu/biology/faculty/boal-pdf/12.boal_et_al_behav_proc_202000.pdf

Clearly something is going on -- but it isn't at all obvious what it is. Cuttlefish that smell a crab, but cannot immediately see it or go after it, apparently learn from considering the aroma to avoid the claws when they later see one. Bizarre.

They perform almost (but not quite) as well as cuttlefish that see experienced ones catching crabs from behind, to avoid being pinched.
 
There was another study that was curious about cuttles and visual learning. I don't have references but could probably dig them up if there is interest. The researchers found that cuttles in the egg, exposed visually to a type of food would forever prefer that food. Paradox tried experimenting a little with the idea to see if he could get new hatch bandensis to eat sooner but did not note any improvement.
 

Shop Amazon

Shop Amazon
Shop Amazon; support TONMO!
Shop Amazon
We are a participant in the Amazon Services LLC Associates Program, an affiliate program designed to provide a means for us to earn fees by linking to Amazon and affiliated sites.
Back
Top