Octopus brains

Once upon a time, as a young undergraduate, I took a course in neurobiology (which turned out to be rather influential in my life, but that’s another story). The professor, Johnny Palka, took pains at the beginning to explain to his class full of pre-meds and other such riff-raff that the course was going to study how the brain works, and that we were going to be looking at invertebrates almost exclusively—and he had to carefully reassure them that flies and squid actually did have brains, very good brains, and that he almost took it as a personal offense when his students implied that they didn’t. The lesson was that if you wanted to learn how your brain worked, often the most fruitful approach was an indirect one, using comparative studies to work out the commonalities and differences in organization, and try to correlate those with differences and similarities in function.

At about that time, I also discovered the work of the great physiologist, JZ Young, who had done a great deal of influential work on the octopus as a preparation for studying brain and behavior. (Young, by the way, went by the informal name “Jay-Zed”, and there you have another clue to my affectation of using my first and middle initial as if it were a proper name.) It was around then that I was developing that peculiar coleoideal fascination a few of the readers here might have noticed—it was born out of an appreciation of comparative biology and the recognition that cephalopods represented a lineage that independently acquired a large brain and complex behavior from the vertebrates. To understand ourselves, we must embrace the alien.

Young’s attempts to understand mechanisms of learning in memory in the octopus were premature, unfortunately—they have very complex brains, and we made much faster progress using simple invertebrates, like Aplysia, to work out the basics first—but it’s still the subject of ongoing research. I was very pleased to run across a general overview of the octopus brain in The Biological Bulletin.

We know that the octopus is amazingly smart. They are capable of associative and observational learning, they are curious and adaptive, and can invent new solutions to problems. They have a large brain relative to their body size, containing about 500 million cells, and they have condensed the classically distributed invertebrate central nervous system into a dense, discrete brain. One of the problems that stymied Young was that, rather than retaining the very large and accessible identifiable neurons we associate with invertebrates, the cephalopods have paralleled the vertebrates, microminiaturizing neurons to pack more cells into a given space. They’ve also built layered structures into their brains, and thrown the tissue up into folds that increase surface area, much as the vertebrate cortex has.

So what does an octopus brain look like? Superficially, nothing like ours.

(click for larger image)

The slice preparation and the basic circuitry of the MSF-VL system. A sagittal section in the central brain of octopus showing the sub- and supraesophageal masses. Note the location of the vertical and median superior frontal lobes.

Right away, you should notice one major peculiarity: the gut runs through the middle of it, separating the brain into a supra- and sub-esophogeal ganglion. Look at it again, though: there are lobes and nuclei and tracts, and even without knowing anything about function, you can tell that this is a modular brain of considerable complexity. Details are different, but the key thing is that we see specialized subunits emerging—processing centers that carry out dedicated tasks.

This paper focuses on one particular area, the Vertical Lobe, or VL, which is at the top of this image. Here’s a closeup:

(Top) An image of a slice used in the physiological experiments. A sagittal slice from the medial part of the supraesophageal brain mass showing the vertical lobe (VL) and median superior frontal lobe (MSF) located dorsally to the median inferior frontal (MIF) and subvertical (SV) lobes. (Bottom) The area within the white rectangle in top figure with a superimposed circuitry schema. MSF neurons (blue) innervating the VL via the MSF tract are shown schematically, as are the amacrine cells (yellow), which synapse onto the large efferent cells (red).

What does the VL do? Octopuses are tough and resilient, and as it turns out you can do some fairly invasive surgeries, stitch them up, and they recover just fine. The VL can actually be extirpated, and VLless octopuses, once they’ve got over the surgery, seem perfectly normal in swimming, feeding, and other ordinary behavioral functions. Deficits show up, though, when they are tested on learning and memory tasks: long term memory function is lost, and learning is greatly impaired. The VL and the median superior frontal lobe (MSF) together form a structure that is functionally analogous to the vertebrate hippocampus.

They also exhibit a microstructure of a sort we see over and over again in the vertebrate brain: a matrix of parallel fibers produced by many small cells, a set of input fibers (the MSF tract) crossing them orthogonally, and a small set of large output neurons upon which all this arrayed activity converges. It’s an elegant way to set up a huge number of synaptic connections and sample from a large field of possibilities, independently evolved in the octopus, and therefore may represent an optimal mechanism for generating behavioral flexibility. Again, there are differences in the details; octopus neurons, like many other invertebrate neurons, do not involve the cell body in generating electrical activity (very much unlike our neurons), and there are certainly substantial differences in the molecular biology and pharmacology of the channels involved. The striking features, though, are the convergences, which tell us what properties of a neuronal array might be necessary for learning and memory.

The findings emerging from recent electrophysiological studies in the octopus suggest that a convergent evolutionary process has led to the selection of similar networks and synaptic plasticity in evolutionarily very remote species that evolved to similar behaviors and modes of life. These evolutionary considerations substantiate the importance of these cellular and morphological properties for neural systems that mediate complex forms of learning and memory. In particular, the similarity in the architecture and physiological connectivity of the octopus MSF-VL system to the mammalian hippocampus and the extremely high number of small interneurons in its areas of learning and memory suggest the importance of a large number of units that independently, by en passant innervation, form a high redundancy of connections. As these features are found in both the octopus MSF-VL system and the hippocampus, it would appear that they are needed to create a large capacity for memory associations.

Even something as specific as the neurobiology of learning and memory benefits from the evolutionary approach. By comparing different systems, we can identify the commonalities linked to function, and thereby come closer to finding generalizable rules and principles rather than the usual welter of details.

Hochner B, Shomrat T, Fiorito G (2006) The Octopus: A Model for a Comparative Analysis of the Evolution of Learning and Memory Mechanisms. Biol. Bull. 210: 308-317.


  1. says

    I haven’t completely absorbed this article, but I’m beginning to share your fascination with octopuses. What’s interesting about them is that they so smart, and yet so different from the smart creatures that we are most familiar with. The other smart critters we know about are all vertebrates, so they are much more closely related to us humans than octopuses are.

    So, is your nickname pronounced “Pee Zed”?

  2. Russell says

    It would be an interesting comparison to see how the microarchitecture of these neuron fields compare with how neural nets have evolved in computing. When I studied this in graduate school, there was lots of handwaving over the parallels, but the microarchitecture in the biological case was not well understood. This strikes me as an area still rife for cross-fertilization.

    BTW, it’s not at all surprising — actually, necessary — that as animals evolve more complex brains, the neurons in the areas that serve higher functions shrink in size. Smaller neurons work faster, and consume less energy. If the neurons in an octopus’s vertical lobe were ten times larger, it’s brain would be more bulky, the octopus would need more vascularization to supply it, and by the time the octopus solved the typical kind of problem it faces, it would already be food for some other critter.

  3. Uniqueuponhim says

    Speaking of academics and papers, I’ve just tried doing a search on ISI Web of Knowledge for articles written by you, PZ, and the latest one I’ve found is an abstract in Developmental Biology in 2001 in which you get a bunch of zebrafish all liquored up. In fact, the search for Myers PZ only turned up 11 articles in total from 1984-2001. Are you still publishing papers? And if you are, is it under a name other than PZ Myers? Although I probably wouldn’t understand most of it (I’m in physics, not neurobiology), it still would be cool to take a look at them.

  4. cm says

    PZ said:

    octopus neurons, like many other invertebrate neurons, do not involve the cell body in generating electrical activity (very much unlike our neurons)

    Technical question: Do you mean that action potentials are only generated at the axon hillock for these invertebrates? Is that definitely known? Or what do you mean? Is there some good reference on which invertebrates have this and which don’t?

  5. says

    Above you mention the massive parallelism and redundancy in the microstructure of the brain which is responsible for the plasticity in both mammalian and octopus brains. In essence, two perpendicular arrays. Perhaps it is only a random association, but it reminds me of a story I read just recently regarding bacteria which are sensitive to chemical gradients. We discovered that the chemical sensors are arranged in arrays, and that it is this arrayed structure which makes possible an amplification of sensitivity to the bacteria’s environment.

    Then there was the story regarding bacteria with magnetosomes. These, too, are arranged in arrays. The investigators had thought that a particular locus was a factor in the formation of the magnetosomes, but when they knocked out the gene, the magnetosomes still formed, but not in an arrayed structure. Consequently, the sensitivity to the earth’s magnetic field was severely diminished although not eliminated.

    The need for some sort of array-like structure for a spatial awareness or sensitivity to the environment seems obvious, once you think about it: “awareness” becomes primarily a function of the object and not the organ with which it is “sensed.” But for increased sensitivity, one would expect a secondary, perpendicular array so that the elements of the primary array may work in concert. Whether this secondary array exists within bacteria is unknown to me — although it would seem that it must. However, once you have this increased sensitivity, this implies amplification — which greatly increases the potential for complex and perhaps even chaotic behavior.

  6. says

    Most invertebrate neurons are monopolar: the cell body extends a neurite into the neuropil, where it arborizes into complex dendrite and an axon. Action potential generation occurs out there and doesn’t involve the soma.

    All the invertebrates I’ve worked with (insects, crustaceans, molluscs) have that kind of organization, but there could be some significant exceptions. A source might be Bullock’s The Nervous Systems of Invertebrates (I don’t have a copy myself to check. Sorry.)

  7. lancelot_gobbo says

    1976, first year medicine at University College London: J Z Young (emeritus at that time) taught a course entitled ‘An Introduction to the Study of Man’ based on his book of the same name. My, that was boring to us in our callow youth. I’d love to do it over again now, and really pump him for more information than the tidbits he fed us. Funny how we change.

  8. says

    You have not addressed the most important question of all: How close are we to cross-breeding octopi with bonobos? It is _vital_ that humanity have a worthy successor species, and what more better heir could their possibly be? The sooner you sciency types can create the Octobo, the sooner it can Rule The Galaxy With An Iron (Yet “Playful”) Tentacle. Get crackin’!

  9. typekey pseudonym says

    Pee-Zed: You wrote – “a lineage that independently acquired a large brain and complex behavior from the vertebrates”.

    What other lineages acquired these attributes from the vertebrates?

  10. Loren Petrich says

    It would be VERY difficult to do a bonobo-octopus cross that is anything more than undifferentiated cells. That’s because differentiation mechanisms have diverged too much.

    I think that some people seem to have a beanbag-genetics view of genes to macroscopic features. While beanbag genetics words fine with proteins, it doesn’t work nearly as well for macroscopic features, which are produced by complicated interplays of genes and their transcription and translation products.

    But even what’s known at the genetic level of development does not match very well. Consider the lineup of Hox genes here:


    Vertebrates have many more Hox genes than molluscs do — for starters.

  11. Bruce Thompson says

    “… the gut runs through the middle of it…”

    I saw that movie.
    Vertebrates keep theirs encased in bone but cephalopods squish theirs around, mashing it into all kinds of shapes when squeezing through tight spaces. I would think that would lead to all kinds of damage to those little neuronal connections?

  12. jaimito says

    Brains are useful for all lifeforms for survival. Aliens must have them too. If convergence is so strong a force, we may be able to talk to them. Hola!

  13. Torbjörn Larsson says

    Re squashing brains this discussion come up before. Googling a bit one can find out that brain surgery and head impacts move around neurons in humans as much as in squids without “all kinds of damage”.

  14. Eric Irvine says

    “I would think that would lead to all kinds of damage to those little neuronal connections?”

    Perhaps that’s why we can “do some fairly invasive surgeries, stitch them up, and they recover just fine”.

  15. Torbjörn Larsson says

    “Googling a bit one can find out that brain surgery and head impacts move around neurons in humans as much as in squids without “all kinds of damage”.”

    I should have checked too :-), I misstated. What the discussion was that time was about the squashing octopus brains take when large chunks of food from the beaks passes the brain opening. A reasonable assumption could perhaps be that the properties of octopus tissue keeps the squashing on the same order when squeezing spaces, but it is merely an assumption.

    An answer informed from independent information is that they may do this squeezing regularly and still seem to behave fine. I wouldn’t worry too much.

  16. John Emerson says

    Human brains can be significantly rearranged without major damage. I guy in Oregon had an arrow shot into his eye during a drinking game, and surgeons had to drill a hole in the back of his skull to push out the barbed head because pulling it out would have torn the brain up too badly.

    According to the surgeon, “whatever brain function he ever had is apparently undamaged”.

    Something to remember when playing “William Tell”. Shooting an arrow in is OK, but when you pull it back out you cross an important threshold where the disruption is just too much.

  17. says

    According to the surgeon, “whatever brain function he ever had is apparently undamaged”.

    Yeah, but what was his baseline, John? This is someone who was playing a drinking game that involved arrows, after all.

  18. says

    Loren Petrich : Bah! You’re obviously just part of the Scientific Establishment COnspiracy to deny the REALITY that A) Octobos are possible, and B) They would make good overlords for humanity. The fact that I can’t find any evidence whatsoever to support this assertion is _proof_ of a shadowy conspiracy, for who else could make such compelling evidence disappear? It is _vital_ that we create noble Octobos to rule us, or the Communists will create evil Squidobos instead! :-)

  19. says

    Well, PZ, I thank you for increasing my appreciation of a much maligned and misunderstood species: the comparative biologist.

    You now make more sense to me.

    And, BTW, the octopus and its most original solution to the quest for general cognition is absolutely fascinating. For humans, we have a raft of theories about what genetic or environmental watershed may account for the rapid [in evolutionary timescales] improvement in our brain power. Is there any corresponding timeline of IQ increases for the cephalopods?

  20. says

    I didn’t think so. Considering where they live and what they have in the way of skeletal remains, I don’t suppose there is much in the way of fossil record for octopi. no stone tools. no cave drawings, no ruins, aquaducts or temples. Come to think of it how DID they get so far ahead of us in the “tidy” department?

  21. Helen Woods says

    H, It s lkly ths ntry cld b ff tpc bt nywys, Hvng gn brwsng bt yr wbst nd t lks vry cl. It s bvs y knw th tpc nd y sm pssnt bt t. W r bldng frsh wblg pls I’m strgglng t mk t lk gd, pls prsnt th bst qlty mtrl. I’v dscvrd gd dl frm ths wbst nd ls I lk frwrd t frthr qlty nfrmtn nd wll b bck sn. Thnks y.