Short and long-term memory in cephalopods

          I've heard the assertion that octopuses have short- and long-term memories several times in the past few days, mostly in discussions of the ethics of eating octopuses prompted by ethical questions raised about Paul, the famous German octopod.  It's interesting to me what these people don't say - that they think that having a multiphasic memory process makes octopuses worth not eating (because, well, people have multiphasic memories, and you wouldn't eat them, would you?!?  Sicko.)  While I don't think that memory capacity of an animal is associated in an uncomplicated way with its ability to suffer or its moral status, it seems to me like a nonetheless interesting question.  I'm almost sure that most of the people who use (read: copy and paste) this bit of information to support their beliefs have very little idea of what sort of research is behind it.  Let's face it: developing a working knowledge of behavioral research on cephalopods is something that just isn't on most of the public's mind.  In fact, until I began writing this blog, I had very little knowledge of the subject.  I plan to set the record straight, so that internet users need never make an unfounded or unqualified statement about memory processes in cephalopods again (a lofty goal, huh?)

          If you don't know octopus neuroanatomy very well (and who does?) you might want to check out the figures in this post.  I'll be talking about the vertical and superior frontal lobes of the octopus brain, and I know it sometimes helps to be able to visualize things like that when you're reading about them.  Just so that it's clear: the term "biphasic memory" means that the memory system in question has two discrete parts or processes (ie. short-term and long-term memory.)  A monophasic memory would have only one process, so that memories would last for a certain amount of time and then fade similarly in all circumstances.  A multiphasic memory system (which could be biphasic, triphasic, or more) is a general term to describe memory systems that are clearly more than monophasic, but are not completely characterized yet - and no memory system is.  Now, on to the research!

          J. Z. Young, that demigod of cephalopod neurobehavioral research, published one of the few papers I could find on this topic back in 1970, following up on his earlier work on the subject.  In it, he investigated the development of short and long term memory in O. vulgaris (I assume - he doesn't actually mention what species he uses in this paper, but he almost always used O. vulgaris) as well as the role of two brain areas in memory, the median superior frontal lobe (MSF) and the vertical lobe (VL).  To do so, he performed surgeries to remove one of these two areas of octopuses' brains and put them through a learning task.  In this task, octopuses were trained to either attack a rectangle (rewarded with a piece of fish) or withhold attacking a crab (which was punished with electric shock.)

          It turned out that octopuses whose vertical lobes had been removed were greatly impaired in learning to attack the rectangle.  Young explains this by claiming that the vertical lobe is involved in short-term memory, and that the acquisition of stable behavior day-to-day was impaired because the animals without vertical lobes could not remember events long enough for the training to be effective.  The animals without median superior frontal lobes, however, learned the task just fine, but were impaired in their long-term retention of it., suggesting that the MSF lobe might have some role in retaining learned information.  Interestingly, Young also found (in other experiments) that removing the vertical lobe after a task was learned resulted in a greater retention of the task.  These results suggest that the vertical lobe plays a role in the updating of memory stores, but is not absolutely essential for the recall of memories.

          His results from the attack-withholding task were less clear, but they suggest that animals with lesions, especially those with vertical lobe lesions, were less consistent than intact animals in learning not to attack a crab after being shocked each time they attacked it.

          Basically, Young argues (on the basis of this and some of his other experiments) that octopuses have a memory system that can be disrupted in more than one way; that is, it is possible to dissociate memory acquisition from long term retention, just like in vertebrates.  For the most part, more current research has agreed with his position, as we'll see in this next paper.

          Moving forward (past a lot of great research that I'll skip over for the sake of brevity) to 2008, Shomrat et al. used electrophysiological methods to test this hypothesis.  Before we get into their methods, let's look a bit more closely at the system that we are talking about (this figure is from Shomrat et al. (2008)):

          On the left is a sagittal slice of the supraoesophageal (over-the-oesophagus) mass of the octopus brain.  On the right is a diagram of the memory system in question.  Sensory information flows into the MSF from the arms and eyes before being sent along to the VL.  The VL neurons in turn send out information encoding attack.  It's been established that long-term potentiation (LTP) can occur in this area of the octopus brain, and this is a likely mechanism for the formation of memories in octopus (I blogged about this here - check it out if you need a little more background.)

          The authors' procedure went as so: O. vulgaris who had already been trained to attack a white ball either had their MSF tract cut (at the dashed line in each image,) severing the sensory input to the vertical lobe, or this tract was stimulated, causing LTP at the synapses indicated in the figure.  Shortly after the procedure, the animals were trained to avoid a red ball through electric shock.  It was found that animals with severed MSF tracts were slower than controls to learn to withhold attack, while animals in whom LTP was induced were quicker.  This is all well and good - it confirms what we already thought about the role of the vertical lobe in acquiring memories in the octopus.  The really important result from this paper came when the authors tested the octopuses a day later.  It was found that both MSF tract transection and LTP induction impaired recall after 24 hours.  So even though stimulation of the MSF tract improved short-term memory (presumably by hyper-activating the memory system in the vertical lobe,) it impaired long-term memory.  This suggests that these two processes are not identical; that is, that octopuses have discrete and dissociable short- and long-term memory circuits.  This general finding has been replicated in cuttlefish (see my post on cuttlefish memory) and nautiluses (Crook and Basil, 2008).

          Unfortunately, that's just about all that we know at this point: that cephalopods appear to have biphasic memories, meaning that the behavioral evidence of short-term memories can be dissociated from that of long-term memories.  This is hardly (by itself) a basis on which we can imply any sort of consciousness or advanced cognitive capacity, as animal-rights supporters who mention this fact seem to imply.

          In interpreting these results in the context of our knowledge of cephalopods as a whole, we should keep in mind what is meant by short- and long-term memory in humans.  Short-term memory is what happens when newly learned information is bouncing around the cortex somewhere, being continually processed but not permanently encoded somewhere.  These memories will disappear if they are not rehearsed (or otherwise actively retained).  Long-term memory has been (relatively permanently) encoded into neural circuits, so that it can be retrieved after periods when it has not been actively processed in short-term (or working) memory circuits.  These processes have been studied intensely in humans, and can be precisely because we have a complex cognitive system build around them (or on top of or parallel to them, depending on who you ask) that we can access.  As of yet, we don't have the experimental techniques to assess exactly how "human-like" or "vertebrate-like" cephalopod memory systems are, because we can't study them in nearly as much detail as language-based and other cognitive tasks allow us to in humans.  Thus, making any strong conclusions about the nature of cephalopod memory other than that it appears to be multiphasic (with no implied "and-so-cephalopods-are-smart-like-people") is untenable.

          Lastly, I find it frustrating that animal rights activists use our (very primative) knowledge of cephalopod memory systems to try to support their position that eating cephalopods is wrong.  Not only is it an inconclusive (what does memory have to do with suffering and morality?) and nonspecific argument (did anybody think that ungulates, swine and birds don't have complex memory systems?), but it misses some of the big points that the animal rights movement has taught us.  First of all, it implies that cephalopods are somehow special because they are intelligent and human-like.  However, having compassion for animals explicitly demands that we not judge their worth by analogy to our own abilities - this has proved to be an attitude that encourages cruelty to animals simply because we are ignorant of them and their behavioral and cognitive capacities.  If we didn't know about cephalopod memory systems, would they still be worth defending from fishing and consumption as food?  Hopefully, the answer is yes - so why try to use this (admittedly inadequate) argument now that we conveniently have information that appeals to one's emotional predispositions?  I find this to be irresponsible and counter-productive, as it diminshes the credibility of other, more valid arguments against the consumption of cephalopods (or any animal, for that matter) that animal rights activists might use.

          Sorry if this was a bit heavy on editorial material.  Being very concerned about animal welfare myself, I get annoyed when people make the cause look stupid by saying things that are ill-informed, ill-reasoned, or just plain wrong.  Although I wish that people would stop killing cephalopods for food, spinning information to try to get people to agree with a point is dishonest, and at best a very poor strategy for debate, as there's bound to be at least one attentive person on the other side who will point out that you're not being true to the facts - and nobody will listen to you after that.

Thanks for reading!


SHOMRAT, T., ZARRELLA, I., FIORITO, G., & HOCHNER, B. (2008). The Octopus Vertical Lobe Modulates Short-Term Learning Rate and Uses LTP to Acquire Long-Term Memory Current Biology, 18 (5), 337-342 DOI: 10.1016/j.cub.2008.01.056

J. Z. Young (1970). SHORT AND LONG MEMORIES IN OCTOPUS AND THE INFLUENCE OF THE VERTICAL LOBE SYSTEM Journal of Experimental Biology (52), 385-393

Crook, R., & Basil, J. (2008). A biphasic memory curve in the chambered nautilus, Nautilus pompilius L. (Cephalopoda: Nautiloidea) Journal of Experimental Biology, 211 (12), 1992-1998 DOI: 10.1242/jeb.018531

Prawn-in-the-tube (More Cuttlefish Memory)

For several years, a group of researchers in France have been studying the neural correlates of learning in cuttlefish (recently focusing on, among other things, oxytocin-like neuropeptides in the cuttlefish CNS - I'll review this in a later post.)  I reviewed some of their work in an earlier post.  Although this is a fascinating concept, their method has been criticized because they use a single learning task to elicit what they claim are learning-induced neural changes, generally.  Importantly, it is questionable whether their method causes associative learning or habituation.  Associative learning involves the formation of a mental or neural (depending on your conceptual preference) association between some behavior and a consequence of that behavior, such as finding food or feeling pain.  This form of learning has long been thought of as one of the hallmarks of adaptive behavior, and it is certainly central to any claims about cephalopod intelligence - if we could not demonstrate associative learning in cuttlefish, we would have very little ground on which to call them intelligent.  Habituation occurs when we are exposed to some stimulus for long enough that we just stop responding to it.  In the case of habituation, we haven't learned much about the stimulus - simply that it is generally unrelated to any reward or punishment we might get.

So what is this controversial procedure?  The group has given it the obscure name of the prawn-in-the-tube procedure.  It is essentially what is sounds like.  A cuttlefish is presented with a prawn enclosed in a clear plastic tube.  Cuttlefish, being visual predators, will attack the prawn, but their tentacles will hit the tube, and their attack will fail.  Over subsequent presentations, they learn not to attack the tube.  The difficulty is that it is hard to tell whether the cuttlefish are simply habituating to the prawn-in-the-tube stimulus, or whether some sort of sensory feedback from failed attacks is causing them to suppress their attacks - a type of associative learning known as passive avoidance learning.

In this group's research on cuttlefish learning (as well as in an older line of research by J. B. Messenger that used the same procedure) it is vital to know what sort of learning they are inducing in order to interpret their results.  Specifically, they work under the assumption that their procedure induces passive avoidance learning.  This is a pretty big assumption.  As such, they decided to settle this problem with a series of experiments, which they published as The “prawn-in-the-tube” procedure in the cuttlefish: Habituation or passive avoidance learning? (2006) by Agin, Chichery, Dickel, and Chichery. 

This study uses two techniques.  The first is called dishabituation.  In these experiments, a strong competing stimulus is presented alternatively with the "habituated" stimulus.  If this elicits a greater response, the it is likely to be the case that the animal has habituated rather than learned by association.  If the response is still suppressed after the novel stimulus is presented, it must be that the familiar stimulus is repressing behavior, and that passive avoidance learning has taken place.  The logic is that the effects of habituation will decrease if the animal becomes generally aroused by some other stimulus.  Their results show, however, that this is not the case.  Novel stimuli did not dishabituate the cuttlefish to the prawn-in-the-tube assembly.  Strike on against the habituation theory.

The second test that they used involved showing the cuttlefish a piece of bait (a live prawn,) and then removing it from the tank as the cuttlefish attacked, preventing them from ever catching it.  In this test, the cuttlefish never received any sort of tactile feedback when they attacked.  If the prawn-in-the-tube procedure causes habituation, we would expect attacks to decrease in this condition, as there is no reward or punishment to shape the behavior.  If the prawn-in-the-tube procedure works mainly by passive avoidance learning, we would expect that, as there is no negative sensory feedback following unsuccessful strikes, the cuttlefish would not change their response at all during this version of the procedure.  As it turns out, the procedure was almost completely ineffective in inducing any sort of learning in this condition.  The cuttlefish continued to strike regularly at the prawn, and their latency to strike actually decreased.  This experiment clearly does not support the habituation hypothesis.  Strike two!

Where's the third strike?  Oh, yeah, Purdy et al found similar results using a variation of this procedure in their paper Prawn-in-a-Tube Procedure: Habituation or Associative Learning in Cuttlefish? (2006).  Strike three, and the habituation hypothesis is out!

Actually, these results could presumably be overturned by some more sensitive or definitive test in the future.  For the moment, however, these studies allow the cuttlefish memory research community to investigate the neural bases of memory in the cuttlefish with a reasonable amount of certainty that they are studying associative learning.  They also make a nice general point about the sort of fine-grained analysis that's needed in order to study complex psychological processes like learning and memory, as well as emphasizing the importance of being critical of the assays that one uses to study these things.

Thanks for reading!

 (Sepia apama.  Photo by Nick Hobgood, used under a Creative Commons license.)

Cuttlefish Memory

Image by David Sim, used under a Creative Commons License.

Being, up to this point, rather octopus-centric in my posts, I decided to dig into the literature on cuttlefish behavior.  Cuttlefish are (relatively) easily kept in aquaria, and have been extensively studied in terms of their complex color-changing behavior (see A review of cuttlefish camouflage and object recognition and evidence for depth perception by Kelman, Osorio, and Baddeley for more information on this area of study) as well as their predatory behavior learning abilities, and possible social behavior.

The first paper I want to talk about in this post is Evidence for a specific short-term memory in the cuttlefish by Agin, Dickel, Chichery, and Chichery (1998).

Agin et al's experiment ran as follows: cuttlefish of various ages were shown a prawn or shrimp sealed inside of a glass tube during a training phase.  As one might expect, the cuttlefish attacked the glass tube, but learned to suppress their attacking as they learned that the prey was inaccessable.  Then, after a 5 minute break, the cuttlefish were tested to see how well they could remember not to attack the prey in the tube.  There were 3 groups - the first got 20 minutes of training, the second got 5 minutes of training, and the last was exposed to the empty glass jar for 20 minutes as a control group.

In a nutshell, here are their results:  Juvenile cuttlefish (15-30 days old) learned to suppress attack in the 20-minute condition only, while for older cuttlefish, a 5 minute exposure was enough for them to learn the lesson and almost completely suppress attack.

This is all well and good by itself.  We would more or less expect that cuttlefish should be able to retain this simple memory, knowing how good cephalopods are at that sort of thing.  What makes this story more interesting is this groups follow-up paper.

In Effects of learning on cytochrome oxidase activity in cuttlefish brain by Agin, Chichery, and Chichery (2001), the group subjected cuttlefish to the same learning task and then measured cytochrome oxidase levels in their brains at different time points after learning.  Cytochrome oxidase expression is increased in metabolically active cells, putatively indicating (in neurons, at least) the overall activity level of the cell (with rather poor spatial resolution, of course.)  What this experiment revealed was that the pattern of cytochrome oxidase staining changed depending on how long it had been since the cuttlefish did the learning task.  In other words, it looks rather like different populations of neurons are involved in the early stages of memory than the late stages of memory.  A similar effect was found using a marker of acetylcholine synthesis in this study, Central acetylcholine synthesis and catabolism activities in the cuttlefish during aging by Bellanger et al .

This is notable because it suggests that cephalopods, like birds and mammals, may have dissociable short- and long-term memory systems.  According to the generally accepted theory of human memory, the short-term and long-term memory systems are comprised of distinct but overlapping neural circuits.  It would be very interesting if the same dissociation had developed in cephalopods (in fact, this same group has gathered more behavioral evidence supporting this hypothesis since these studies were published, ie. their study Developmental study of multiple memory stages in the cuttlefish, Sepia officinalis, 2006 .)  While it seems that the study of cephalopod memory processes is still in its infancy (compared, at least, to the study of human memory processes) and no strong assertions about this system can be made yet, it's a tantalizing thought that we might be able to find and understand such a striking example of the convergent evolution of neural systems as this.