Benny Hochner on octopus and motor control

  • cover play_arrow

    PLAY EPISODE


Season 2015
Season 2015
Description arrow_drop_down

Description

How does an animal with no skeleton, no somatotopic brain map, and eight arms containing more neurons than its central brain manage to produce precise, goal-directed movements? Neuroscientist Benny Hochner reveals how the octopus solves the seemingly impossible problem of controlling a soft body with infinite degrees of freedom. Subscribe for more from the Convergent Science Network podcast series. Benny Hochner joins Paul Verschure and Tony Prescott at the BCBT summer school to discuss his research on motor control and learning in the octopus , an animal he describes as the most intelligent invertebrate and a remarkable case study in convergent and divergent evolution. With half a billion neurons, most distributed across its eight arms rather than centralized in the brain, the octopus has evolved a radically different solution to motor control than vertebrates. For reaching movements, it reduces its theoretically infinite degrees of freedom to just three by propagating a wave of muscle stiffening along the arm, creating a simple but effective motor program that can be generated even in a completely severed arm. The discussion explores the hierarchical organization of the octopus nervous system, from autonomous arm reflexes to coordinated whole-body behavior. A severed arm can still grasp food and pass it along its suckers toward where the mouth would be. The central brain appears to encode motor programs rather than body maps , no somatotopic organization has been found for either motor commands or sensory processing. Remarkably, tactile discrimination learned with one arm generalizes to all others, confirming central involvement in learning but not in arm-specific representation. Hochner also describes convergent findings in learning and memory: the octopus vertical lobe resembles the mammalian hippocampus in structure and exhibits robust activity-dependent long-term potentiation, though mediated by molecular mechanisms modified from simpler molluscan ancestors. Key topics include why the octopus is scientifically important as an independently evolved intelligent invertebrate, how muscular hydrostats solve the degrees-of-freedom problem through embedded motor primitives, why no body map exists in the octopus central brain, how the fetching movement creates a temporary articulated structure from a boneless arm, what the vertical lobe reveals about convergent evolution of learning mechanisms, and how the octopus challenges conventional assumptions about the necessity of body representation for coordinated action. Part of the Convergent Science Network podcast series from the BCBT Summer School.

Tagged as:

About the author call_made

CSN Podcasts

Both the triumphs of humanity and its most evil deeds have resulted from collaboration. In a time where humanity is required to aspire to the former and minimize the latter, the question arises of how collaboration arises and why it fails. Surprisingly, this phenomenon, so central to who we are, is not well understood. Hence, a collaborative effort is required to understand collaboration in its full biological, psychological, sociological, cultural, and economic complexity and to translate this understanding into operational impact. This series of podcasts is one step toward achieving these complementary goals. The Collaboration Podcast presents interviews with people who are central orchestrators of collaboration in various domains including business, government, science, art, health, sustainability, and the military. The discussions were conducted by Prof. Dr. Paul F.M.J. Verschure and members of the Program Advisory Committee of the Ernst Strungmann Forum on Collaboration (https://www.esforum.de/forums/ESF32_Collaboration.html) during 2021 and had the goal to sketch a map of opportunities, challenges, and obstacles in human collaboration. The forum took place in May 2022, and now we would like to share this series of interviews with a broader audience. The full report of the Forum will be published in 2023 by MIT Press. The podcast was produced by the Convergent Science Network (https://www.convergentsciencenetwork.org/). Context: The stability of social systems depends critically on realizing sustainable methods of “collaboration,” yet how and by which means collaboration is achieved is not clearly understood; neither are the conditions or processes that lead to its breakdown or failure. Collaboration can be understood as cooperation between agents toward mutually constructed goals. Part of the reason for our lack of understanding is that the phenomenon of collaboration is, by nature, a highly multidisciplinary problem, and effective research into its complexities has been difficult to achieve across the broad range of scientific and technical disciplines involved. The need for a fundamental understanding of collaboration, however, has become increasingly important. Not only does humankind demand answers as it attempts to address critical challenges at multiple scales (e.g., climate change, migration, enhanced automation, social and economic inequality), but ever-increasing technological and economic means of interconnecting people and societies are disrupting long-established, familiar patterns of how we interact. Radical technological changes that are ongoing have the potential to reshape collaboration in ways that are currently hard to predict or influence (e.g., by altering configurations in interaction, information creation, and modes of communication). On one hand, such changes could disrupt hitherto stable forms of collaboration by affecting critical communication channels and traditional roles, as can be observed in the rapidly changing patterns in governance, commerce, and social interaction. Conversely, technology could lead to the emergence of novel, successful forms of collaboration that deviate from traditional “hierarchical” architectures. Evidence of this can be seen in areas as diverse as highly automated manufacturing plants, the open science movement, collaborative software repositories, user-centered services, and the sharing of economy-based modes of organization. Without a fundamental understanding of the mechanisms, processes, and boundary conditions of collaboration, it is not possible to evaluate or predict which of these possible scenarios are sustainable or even plausible. The Forum “How Collaboration Arises and Why it Fails” (May 8–13, 2022, Location: Frankfurt am Main, Germany) Chairs: Andreas Roepstorff and Paul Verschure Program Advisory Committee: Jenna Bednar, Julia R. Lupp, Bhavani R. Rao , Andreas Roepstorff, Ferdinand von Siemens, and Paul Verschure

More posts

Timestamp

  • fast_forward00:00:03 - This is the Convergent Science Network podcast. Leading researchers in the domain
  • fast_forward00:00:10 - of neuroscience, brain theory and technology are interviewed by Paul Verschoor and Tony Prescott.
  • fast_forward00:00:17 - It's Paul Verschoor with the Convergent Science Network podcast together with
  • fast_forward00:00:21 - my colleague Tony Prescott.
  • fast_forward00:00:23 - And we're here at the BCBT Summer School 2015 in Barcelona. And we're talking
  • fast_forward00:00:28 - to Benny Hochner, who gave a talk this morning about his work on an amazing animal, the octopus.
  • fast_forward00:00:36 - So, Benny, of course, we're very familiar with the octopus in different restaurants
  • fast_forward00:00:43 - and so on, because of culinary experience.
  • fast_forward00:00:46 - But why is an octopus so interesting also from a scientific perspective?
  • fast_forward00:00:53 - Okay. Octopus is interesting from a scientific point of view for several reasons.
  • fast_forward00:01:01 - First, it's the best example
  • fast_forward00:01:04 - for invertebrate or animal that
  • fast_forward00:01:08 - uses a soft body for doing motor or action with a very sophisticated control
  • fast_forward00:01:15 - mechanism of this very complex computational task of controlling movement in flexible structure.
  • fast_forward00:01:25 - And the second reason to the interest in the octopus is because octopus is considered
  • fast_forward00:01:34 - to be the most intelligent invertebrate.
  • fast_forward00:01:40 - And therefore, it's very interesting from a comparative viewpoint to see what has evolved.
  • fast_forward00:01:48 - Independently in the octopus that enabled this high, what we can call cognitive
  • fast_forward00:01:55 - capabilities of this animal.
  • fast_forward00:01:58 - And by such a comparative approach,
  • fast_forward00:02:02 - we may find the universal common mechanism which are important to the evolution
  • fast_forward00:02:13 - of animal with cognitive capability and complex behavior.
  • fast_forward00:02:22 - Okay. So one thing you emphasize is complex behavior.
  • fast_forward00:02:26 - Yeah. So, what kind of behavior should we be thinking about for the octopus?
  • fast_forward00:02:31 - What's the most complex behavior an octopus is capable of?
  • fast_forward00:02:36 - This is a tough question for a human being.
  • fast_forward00:02:41 - I think you should ask the octopus what is very different. No,
  • fast_forward00:02:44 - but given that you are the ambassador of the octopus in this room, we're forced to ask you.
  • fast_forward00:02:51 - So...
  • fast_forward00:02:54 - If we speak about, for example, from a point of view of motor control in an
  • fast_forward00:03:02 - animal which doesn't have any hard skeletal structure,
  • fast_forward00:03:07 - that is able to do so much different forms of behavior, From swimming, to crawling,
  • fast_forward00:03:19 - to walking, to hiding, to go through very narrow holes, to extend an arm into a target.
  • fast_forward00:03:33 - This is a very, very, the repertoire of behavior is very amazing.
  • fast_forward00:03:40 - And this is a predator animal with a very good visual system,
  • fast_forward00:03:47 - and it's very sensitive to what's going on in the environment.
  • fast_forward00:03:54 - And then it can organize its attack behavior and all of it with this flexible body.
  • fast_forward00:04:09 - Another reason I think which I got from your talk as to why octopus is so interesting
  • fast_forward00:04:14 - scientifically is that from the point of view of mammals or vertebrates,
  • fast_forward00:04:19 - which is what we are, octopus is really quite an alien species.
  • fast_forward00:04:24 - So we only share a common ancestor if you go a very long way back.
  • fast_forward00:04:29 - So the interest in octopus is partly around how it has come up with sometimes
  • fast_forward00:04:36 - convergent solutions to the problems it faces, where it faces similar problems to vertebrates,
  • fast_forward00:04:42 - and in other ways, perhaps divergent solutions.
  • fast_forward00:04:45 - Would you agree with that? Yeah, I think this is exactly what we find in our research.
  • fast_forward00:04:53 - And interestingly, we study both motor control and learning memory in this animal.
  • fast_forward00:05:03 - So on one side, because of this flexibility of the body and soft body.
  • fast_forward00:05:09 - We find a very unique solution, evolution solution.
  • fast_forward00:05:14 - To this big problem of controlling the flexible body, while on the other side,
  • fast_forward00:05:21 - in learning memory, we find in the brain an area which is very similar to our hippocampus,
  • fast_forward00:05:27 - not only by structure, but also, for example, by its very robust long-term potentiation,
  • fast_forward00:05:36 - activity-dependent long-term potentiation,
  • fast_forward00:05:39 - which is probably a common mechanism to mediate changes in the nervous system,
  • fast_forward00:05:45 - which is the basis for long-term learning and memory.
  • fast_forward00:05:52 - So there may be some convergence in these more cognitive functions.
  • fast_forward00:05:55 - Yes. But the divergence that you're seeing in motor control,
  • fast_forward00:05:59 - some of that you then put down to the very different morphology of the octopus,
  • fast_forward00:06:05 - purpose, that it has no internal skeleton, it's just made of soft parts, and that then,
  • fast_forward00:06:12 - enforces a different way of solving the problem of controlling limbs?
  • fast_forward00:06:16 - I think yes, because it creates two main problems that I would define.
  • fast_forward00:06:21 - One is that a soft structure has infinitely large degrees of freedom.
  • fast_forward00:06:29 - That the animal has, the control system has to control while doing a certain task.
  • fast_forward00:06:36 - For example, in our arm, we have only seven degrees of freedom,
  • fast_forward00:06:42 - which is relatively easy to solve, and this is due to the skeletal structure of our arm.
  • fast_forward00:06:49 - When you have an arm without any rigid steltron, any joint, this can do any
  • fast_forward00:06:57 - movement into any direction, elongate, bend,
  • fast_forward00:07:01 - shortening, twisting, and the octopus uses this kind of movement.
  • fast_forward00:07:07 - So it has to evolve a completely different approach to solve this complexity and how to use it.
  • fast_forward00:07:20 - To produce a goal-directed, for example, a goal-directed movement.
  • fast_forward00:07:26 - Just to push you on that, I mean, although I have only seven joints in my arm,
  • fast_forward00:07:30 - I still have many ways of reaching to a given point in space.
  • fast_forward00:07:33 - So I have the same problem of redundancy of degrees of freedom,
  • fast_forward00:07:38 - not to the same extent as the octopus, but I still have to reduce the degrees
  • fast_forward00:07:43 - of freedom and I do that by exploiting synergies.
  • fast_forward00:07:46 - And isn't there some convergence, Maybe in the way that the octopus is also
  • fast_forward00:07:49 - finding things. I think this is right.
  • fast_forward00:07:51 - Also, what we find in the octopus is that in certain kinds of behavior,
  • fast_forward00:07:59 - and this is, I think, a great achievement of the octopus, it can use all its degrees of freedom.
  • fast_forward00:08:06 - For example, if the arm is as if it behaves by itself when it's searching the
  • fast_forward00:08:11 - surrounding and looking for and saving as a probe, then it can let the arm loose
  • fast_forward00:08:17 - and do whatever it likes according to the to some uh.
  • fast_forward00:08:23 - Motor primitives that are embedded in the arm itself, while when goal-directed
  • fast_forward00:08:28 - movement, the octopus has to use a way to reduce the number of degrees of freedom to only very few.
  • fast_forward00:08:37 - So for example, if the octopus reached to a target, use only three degrees of freedom.
  • fast_forward00:08:43 - One which took, sorry, which control the propagation of a bend in the arm toward
  • fast_forward00:08:51 - the target, and the two others just to control the direction of the arm in space.
  • fast_forward00:08:58 - Similarly, when the octopus gets food with one of its suckers along the arm,
  • fast_forward00:09:06 - and it wants to bring the food to the mouse, it creates an articulated structure from its arm.
  • fast_forward00:09:13 - He reshapes its arm into an articulated structure, and then it can bring the
  • fast_forward00:09:18 - food to its mouse very accurately, similarly in the ways that we are doing this behavior.
  • fast_forward00:09:26 - But in our case, it's embedded in our skeletal structure, while in the octopus,
  • fast_forward00:09:34 - it's basically embedded in the motor program.
  • fast_forward00:09:38 - He uses a certain way to reshape its arm into an articular structure,
  • fast_forward00:09:46 - structure, which is a dynamic,
  • fast_forward00:09:48 - can be dynamically adjusted to the site where the octopus got the target and
  • fast_forward00:09:53 - bring the target accurately to the mouse.
  • fast_forward00:09:57 - But you use the same mechanism, the same strategy of doing accurate point-to-point movement.
  • fast_forward00:10:03 - Now before we delve into your own research on the behavior and the control of
  • fast_forward00:10:09 - these arms by a neocupist.
  • fast_forward00:10:12 - So is it actually even useful to still talk about degrees of freedom in this case?
  • fast_forward00:10:18 - Should we just forget about that because it's practically an infinite set of
  • fast_forward00:10:24 - control points it can work with?
  • fast_forward00:10:27 - To what extent is it really helpful to talk about degrees of freedom from that
  • fast_forward00:10:31 - perspective? I think it's good from the point of view that you set to yourself
  • fast_forward00:10:38 - the problem, what the problems the octopus has to deal with.
  • fast_forward00:10:43 - And what he has to deal with is to control many degrees of freedom.
  • fast_forward00:10:49 - I don't think the octopus knows about this many degrees of freedom because it's
  • fast_forward00:10:54 - embedded in his evolution and his structure and self-organization of his nervous
  • fast_forward00:10:58 - system to be able to use this.
  • fast_forward00:11:03 - Uh, many degrees of freedom and to collapse them in only few degrees of freedom
  • fast_forward00:11:10 - by doing stereotypical movement, which suit this kind of flexible structure.
  • fast_forward00:11:17 - And it's also, um, constrained by how, how complex the motor,
  • fast_forward00:11:25 - the motor command can be to the muscle.
  • fast_forward00:11:28 - For example, a simple motor command to a muscular hydrosat like the octopus
  • fast_forward00:11:34 - arm, where the arm is organizing longitudinal and transversal orientation of muscle fiber,
  • fast_forward00:11:42 - the activation of both muscles together, to the same extent, creates stiffening.
  • fast_forward00:11:49 - So, this is a very simple motor program. The brain has just to give the motor
  • fast_forward00:11:55 - neurons in the arm command to have the same output.
  • fast_forward00:11:59 - And this is enough to create the stiffening. And then you can use a rather simple,
  • fast_forward00:12:05 - I think, motor program to propagate this stiffening along the arm.
  • fast_forward00:12:11 - And this can create the reaching movement very simply,
  • fast_forward00:12:14 - not only by the degree of
  • fast_forward00:12:17 - freedom they need to be controlled in to produce the movement
  • fast_forward00:12:20 - but but also a very simple motor
  • fast_forward00:12:24 - program to activate the the the
  • fast_forward00:12:27 - reaching but before you look at the motor program
  • fast_forward00:12:31 - what is what is interesting about the octopus among many things it's like a
  • fast_forward00:12:35 - mollusk that that escaped from its shell i mean in phylogenetically it sits
  • fast_forward00:12:39 - in this family with a phylum of mollusks and right its Its closest neighbors
  • fast_forward00:12:44 - in that phylum are actually really simple animals that live in a shell. Right.
  • fast_forward00:12:49 - So, what are the...
  • fast_forward00:12:53 - What are the differences between the octopus and the other animals in that phylum?
  • fast_forward00:12:59 - Like if you look at the nervous system of a mollusk and an octopus,
  • fast_forward00:13:04 - what are the differences?
  • fast_forward00:13:05 - First of all, there is a big difference in the size of the nervous system.
  • fast_forward00:13:11 - The octopus nervous system contains half a billion neurons, while a garden snail
  • fast_forward00:13:20 - may have a few tens of thousands of neurons.
  • fast_forward00:13:25 - So there is a big difference in cell number.
  • fast_forward00:13:29 - And also what we see is that there is a huge difference in the way the nervous
  • fast_forward00:13:35 - system is organized in the body.
  • fast_forward00:13:40 - Both in the simple animal and in the octopus, the nervous system is organized
  • fast_forward00:13:45 - to maximize the efficiency of the nervous system controlling different parts of the body.
  • fast_forward00:13:51 - So for example, in the octopus, the majority of nerve cells are situated already
  • fast_forward00:13:57 - in the peripheral nervous system of the arm.
  • fast_forward00:14:01 - This fits what we are finding in the behavior in our physiological studies,
  • fast_forward00:14:06 - that many of of the motor program are generated at the level of the neuromuscular system of the arm.
  • fast_forward00:14:13 - So basically, the central brain, which contains only 50 million neurons,
  • fast_forward00:14:21 - actually need to process information and set command to the peripheral nervous
  • fast_forward00:14:27 - system how to produce stereotypical arm movement.
  • fast_forward00:14:33 - Now, the main difference really is that.
  • fast_forward00:14:38 - Lower invertebrate mollusks depend on their protection on their shell.
  • fast_forward00:14:46 - The shell is a very nice protection. I should mention that mollusks have another
  • fast_forward00:14:52 - way of protecting themselves, and they are a very rich source for all kinds of toxins.
  • fast_forward00:14:58 - So they are a very efficient, let's say,
  • fast_forward00:15:02 - factory of biochemicals or things that can help us in studying the nervous system
  • fast_forward00:15:12 - because they block ion channels and receptors and so forth.
  • fast_forward00:15:19 - So they used this kind of defense mechanism.
  • fast_forward00:15:26 - And the octopus lost his protecting shell and became to be a freely moving animal,
  • fast_forward00:15:33 - a predator, that depends very much on its visual system.
  • fast_forward00:15:40 - Them and losing the shell,
  • fast_forward00:15:45 - enable the animal, the heavy shell, enable the animal to become such a freely
  • fast_forward00:15:51 - moving and really during the
  • fast_forward00:15:53 - revolutions they actually compete with fishes at the time of evolution.
  • fast_forward00:16:00 - But is there anything conserved there? If we go from mollusks to snail to octopus,
  • fast_forward00:16:05 - is there anything conserved when we think go up into complexity?
  • fast_forward00:16:11 - I think this is a very good question. I used to work on.
  • fast_forward00:16:20 - Plesia californica, together with Professor Eric Kandel, on the basic mechanism
  • fast_forward00:16:27 - of learning memory in this animal.
  • fast_forward00:16:30 - So for me particularly, it's very interesting to see if the simple,
  • fast_forward00:16:36 - mechanisms that exist in the aplysia for
  • fast_forward00:16:40 - mediating simple form of learning and memory
  • fast_forward00:16:43 - in the reflex of the
  • fast_forward00:16:47 - defense reflex of this animal are conserved or completely different mechanisms
  • fast_forward00:16:55 - that is converging with our brain has been evolved in the octopus. and what we find that.
  • fast_forward00:17:07 - No. In the octopus, there are still mechanisms which are more likely to what
  • fast_forward00:17:14 - takes place in other mollusks.
  • fast_forward00:17:18 - But interestingly, they have been modified a lot in order to establish a new cellular mechanism.
  • fast_forward00:17:27 - So I mentioned before the mechanism of activity-dependent long-term potentiation,
  • fast_forward00:17:35 - which I think it's the universal cellular mechanism for mediating learning memory.
  • fast_forward00:17:41 - Memory, but the molecular mechanism that mediates this long-term potentiation
  • fast_forward00:17:47 - in the octopus brain is mediated by a mechanism which is conserved from other mollusks,
  • fast_forward00:18:01 - but it was
  • fast_forward00:18:02 - modified to do this kind of long-term potentiation as a cellular process.
  • fast_forward00:18:12 - But doesn't, in some sense, it makes the octopus a strange exception in its
  • fast_forward00:18:17 - phylum, but also a bit annoying because also as you described in your talk,
  • fast_forward00:18:22 - you looked at the evolution of the eye.
  • fast_forward00:18:26 - The octopus eye, eye which shares many properties with the human eye which
  • fast_forward00:18:30 - is is is sort of suggesting that the
  • fast_forward00:18:33 - common ancestor is is way beyond uh
  • fast_forward00:18:37 - a very very early in in phylogeny so um doesn't that really mean that that from
  • fast_forward00:18:45 - a genetic perspective uh also these very simple organisms are actually carrying
  • fast_forward00:18:50 - a lot of additional information that they share with us,
  • fast_forward00:18:54 - way beyond just the basics of cellular mechanisms and so on.
  • fast_forward00:18:59 - Yeah, I think this is completely true. And I think it fits.
  • fast_forward00:19:03 - First of all, with the idea that what we find is, you know, the number of genes
  • fast_forward00:19:09 - in our, or in octopus, is not that many.
  • fast_forward00:19:15 - Actually, probably there are rather basic building blocks of genes that the
  • fast_forward00:19:25 - first living creature already had when it developed.
  • fast_forward00:19:30 - So now for analyzing a light, you have molecules that can be shared by plants,
  • fast_forward00:19:41 - by animals, and by eyes and by other creatures that are using light to do all
  • fast_forward00:19:53 - kinds of camera photosynthesis and things like this,
  • fast_forward00:19:57 - and they are built to the world.
  • fast_forward00:20:05 - To collect light? To collect photons? No, to...
  • fast_forward00:20:13 - Transform? Transform.
  • fast_forward00:20:16 - Converge? Transform light into... Electrical signals. Yeah.
  • fast_forward00:20:22 - Transduce, maybe. Transduce. Transduce. Transform.
  • fast_forward00:20:27 - Anyway. Transduction. It's called the transduction mechanism.
  • fast_forward00:20:32 - Okay. okay so now we have a bit an idea about the octopus it's really a very
  • fast_forward00:20:40 - strange almost an alien species that sort of got plunked into a phylogeny but
  • fast_forward00:20:44 - still it shares properties with even with us now it controls this very strange body,
  • fast_forward00:20:51 - it has eight arms,
  • fast_forward00:20:53 - controlled by complex ganglia that again talk to a central brain what do we
  • fast_forward00:20:59 - know about this ganglia what do the ganglia do that are linked to a single arm?
  • fast_forward00:21:05 - The ganglia in the arm or the ganglia in the brain?
  • fast_forward00:21:09 - No, the ganglia in the arm. So I want to start at the periphery. Yeah.
  • fast_forward00:21:14 - Actually we don't know much about the ganglia in the arm, about the function of the arm.
  • fast_forward00:21:22 - We know that there are motor neurons which are innervating about 400,000 motor
  • fast_forward00:21:29 - neurons on in each arms that are innovating very every.
  • fast_forward00:21:36 - Every hundred micron, the muscles that run along the arm, and each muscle is
  • fast_forward00:21:45 - innervated by three types of motor neurons that have different properties.
  • fast_forward00:21:51 - But we don't know exactly what's going on in the arm in terms of the computational
  • fast_forward00:21:56 - properties of this structure, the reflexive behavior.
  • fast_forward00:22:01 - We know that there is some input from
  • fast_forward00:22:04 - proprioceptive into the central nervous
  • fast_forward00:22:08 - system of the arm but we don't know exactly what how
  • fast_forward00:22:11 - it's activated the for example the the reflex
  • fast_forward00:22:14 - of the suckers or the or the bending or any other function of the of the arm
  • fast_forward00:22:22 - we know better due to many not not our work but the the people in Wells and J.Z.
  • fast_forward00:22:33 - Young and other people that work mainly in Napoli,
  • fast_forward00:22:37 - they studied the structure of the central brain and they found very interestingly
  • fast_forward00:22:43 - that the brain is organized still,
  • fast_forward00:22:46 - it's very very centralized, but it's still composed of lobes,
  • fast_forward00:22:52 - of ganglia.
  • fast_forward00:22:54 - And by stimulating and by lesioning, they could assign a more or less specific
  • fast_forward00:23:02 - function for different lobes.
  • fast_forward00:23:06 - And this is what makes us interested in this central brain, because we think
  • fast_forward00:23:14 - that the separation between the periphery and the central processes,
  • fast_forward00:23:18 - the central brain seems to be dealing more with cognitive function.
  • fast_forward00:23:25 - We think it's an ideal preparation to study these processes because it's separated
  • fast_forward00:23:32 - from the input and the output structure.
  • fast_forward00:23:37 - And in vertebrate brains, you might look at sort of the spinal cord as being
  • fast_forward00:23:43 - doing a lot of this motor sensory stuff and processing things locally.
  • fast_forward00:23:49 - And then brainstem perhaps doing some more of that and then midbrain and forebrain
  • fast_forward00:23:56 - doing these more higher executive functions.
  • fast_forward00:23:58 - Yeah. Sort of deciding what are targets for movement and then executing movements.
  • fast_forward00:24:03 - I mean, does that kind of way of thinking about the decomposition of the nervous
  • fast_forward00:24:08 - system work at all for Octopus?
  • fast_forward00:24:10 - I think yes, but actually you can look at the nervous system of the arm,
  • fast_forward00:24:19 - the arm nerve cord, as a spinal cord.
  • fast_forward00:24:25 - But it's much more elaborated than, for example, our spinal cord,
  • fast_forward00:24:30 - because of the special separation of labor between the central nervous system and the peripheral one.
  • fast_forward00:24:39 - The peripheral one has a lot of function of its own, and therefore,
  • fast_forward00:24:44 - although we treat it as a peripheral nervous system, it's very elaborated in
  • fast_forward00:24:51 - its organization because its function, both in processing sensory information,
  • fast_forward00:24:57 - and activating motor program.
  • fast_forward00:25:01 - Is very rich.
  • fast_forward00:25:04 - But if you take the head off a chicken or you decorticate a cat,
  • fast_forward00:25:09 - they can still generate rhythmic motions, run on a treadmill,
  • fast_forward00:25:13 - these sorts of things. So is there a parallel there?
  • fast_forward00:25:16 - So I understand that the arm can do a lot of autonomous behavior.
  • fast_forward00:25:20 - Yeah, so for example, For example, the central brain is divided into two parts,
  • fast_forward00:25:28 - one which is above the oesophagus, which is called the supraesophageal part
  • fast_forward00:25:33 - of the brain, and there is a lower part which is the subesophageal part.
  • fast_forward00:25:38 - So the subesophageal part is really like the brainstem is more like dealing
  • fast_forward00:25:46 - with the vegetative function of the body,
  • fast_forward00:25:50 - like breathing and other very,
  • fast_forward00:25:54 - basic reflex.
  • fast_forward00:25:56 - And the upper part of the brain, the superesophageal part, is more...
  • fast_forward00:26:02 - Here is where the learning memory centers are organized, and the higher motor
  • fast_forward00:26:08 - control center is organized.
  • fast_forward00:26:10 - And so if you can create a decerebrated octopus by removing only this part of
  • fast_forward00:26:19 - the brain. And then you find that there is some.
  • fast_forward00:26:23 - Reflexes that are coordinated between the arms.
  • fast_forward00:26:28 - For example, if you take such a dis-erebrated octopus and you hold one of its
  • fast_forward00:26:34 - arms, you will find that the rest of the arm is coming to help the taken arm.
  • fast_forward00:26:42 - So there is a lot of coordination taken also at the lower level of the central nervous system.
  • fast_forward00:26:52 - And this area has this very basic surviving properties of the central nervous system.
  • fast_forward00:27:03 - So I think you're right that it's organized in more or less similar hierarchy
  • fast_forward00:27:10 - of responsibility, let's say, of the animal behavior.
  • fast_forward00:27:18 - So then so we could call these reflexes of the octopus this would be a typical
  • fast_forward00:27:23 - defensive reflex one arm gets stuck and you pull the other arms towards it right
  • fast_forward00:27:29 - but now if we go further down and let's say now we,
  • fast_forward00:27:35 - also we just sever an arm what can a single arm still do?
  • fast_forward00:27:42 - Arm can do a lot especially if you if you you.
  • fast_forward00:27:50 - Encourage it to do something like, like, you know, like, uh,
  • fast_forward00:27:53 - like bringing a piece of food into the sucker, uh, the arm can continue to survive
  • fast_forward00:27:59 - about half an hour in, uh, in a isolated, in, in, in normal seawater.
  • fast_forward00:28:05 - And the, and the, and the, um, and the arm will grab the food and will behave as,
  • fast_forward00:28:11 - as, as really it's bringing it, maybe even bringing it back to the place where the mouse is.
  • fast_forward00:28:26 - Actually, in 1971, there was a paper in Nature by Jennifer Altman,
  • fast_forward00:28:34 - who described that the octopus can bring food,
  • fast_forward00:28:38 - an isolated arm can bring food toward
  • fast_forward00:28:42 - the mouse by moving it along the
  • fast_forward00:28:45 - suckers and if you put an acid or
  • fast_forward00:28:47 - something on this food and put it on the arm again it will
  • fast_forward00:28:51 - move it outward so when we started we we already had some ideas that there is
  • fast_forward00:28:58 - a lot of autonomy in the behavior of the of the arm itself so but now you What you also showed is that,
  • fast_forward00:29:09 - so there is some sense of, let's say, hierarchical organization of this motor system.
  • fast_forward00:29:16 - But then if you go look for, let's say, the organization of,
  • fast_forward00:29:20 - let's say, the control signals of that in the central brain,
  • fast_forward00:29:23 - it looks surprisingly, let's say, disorganized.
  • fast_forward00:29:27 - So you don't find a somatotopic map, right?
  • fast_forward00:29:30 - You don't find that, let's say, arms are represented in some coherent,
  • fast_forward00:29:34 - at least so far, no one has found that. Yeah. So how do you interpret that?
  • fast_forward00:29:40 - I think this fits the idea that in order to control a flexible structure,
  • fast_forward00:29:49 - you cannot use what we use to define as a conventional mechanism by representing the body,
  • fast_forward00:30:00 - the body part in the brain, and use it as an input to a system that will,
  • fast_forward00:30:05 - together with the sensory information that is also represented in a somatotopic way,
  • fast_forward00:30:12 - will integrate into initiation of command.
  • fast_forward00:30:17 - And this is because it's very hard to envisage a way that such,
  • fast_forward00:30:23 - let's say, 300 suckers along the arm will be, and each arm, an eight arm,
  • fast_forward00:30:32 - will be represented in a specific location in the brain.
  • fast_forward00:30:37 - I think it's a lot of information. So, the octopus,
  • fast_forward00:30:41 - I think, solved this problem by that in the brain what is represented is what
  • fast_forward00:30:48 - kind of behavior the higher motor center should induce as a response to some
  • fast_forward00:30:55 - specific input, sensory input.
  • fast_forward00:30:58 - So the brain is organized in what we think, we have to test it yet,
  • fast_forward00:31:04 - in a more of motor or program organization.
  • fast_forward00:31:09 - But it is in some sense doing something a little bit like conventional sort
  • fast_forward00:31:14 - of inverse kinematics so that in order to solve this problem of reaching to a point in space,
  • fast_forward00:31:20 - it has to, as you explained, reduce the degrees of freedom in the arm and make
  • fast_forward00:31:25 - it to a number of relatively rigid parts with joints in between them.
  • fast_forward00:31:31 - And then you can imagine that the brain is doing some inverse kinematics in
  • fast_forward00:31:36 - order to work out how to control that reduced degree of freedom system to reach a point.
  • fast_forward00:31:42 - Yeah. So, yeah.
  • fast_forward00:31:45 - So, this is what is happening in the system.
  • fast_forward00:31:49 - Yeah. But in that case, wouldn't you expect something sort of convergent about
  • fast_forward00:31:54 - the control of these high-level structures compared to, for instance,
  • fast_forward00:31:59 - the mammalian motor cortex?
  • fast_forward00:32:01 - Isn't the control of my arm solving something of a similar problem if I'm reaching
  • fast_forward00:32:06 - to a point in space guided by a vision?
  • fast_forward00:32:12 - Reaching is different than what we are doing. Right.
  • fast_forward00:32:20 - So the motor program, the behavior is completely different than what we are
  • fast_forward00:32:25 - doing, and this is taking advantage of the flexibility of the arm.
  • fast_forward00:32:30 - Because reaching to a target by propagating a wave wave of activation and bends
  • fast_forward00:32:38 - that propagate through the arms, it's something that we cannot do.
  • fast_forward00:32:41 - Right. But the command to induce this kind of behavior might be very simple.
  • fast_forward00:32:47 - It just has to activate this wave of stiffening muscles that will push the arm toward the target.
  • fast_forward00:33:01 - But if it would be so easy, I'm sure I would have been able to fully understand
  • fast_forward00:33:06 - it. And there's still something missing in that picture, right?
  • fast_forward00:33:08 - So, for instance, what is then also strange, if there's no smetotopy to the
  • fast_forward00:33:14 - organization of the motor neurons,
  • fast_forward00:33:17 - you could still maybe expect that in terms of the haptic information that the
  • fast_forward00:33:23 - central brain receives from the different arms, that since this would express
  • fast_forward00:33:27 - some topological organization,
  • fast_forward00:33:30 - might be reflected in a topological map of the body.
  • fast_forward00:33:33 - But since you haven't found it, does that maybe imply that the sensory information
  • fast_forward00:33:38 - never reaches the central brain?
  • fast_forward00:33:40 - Is that possible? That the central brain only knows something about what let's
  • fast_forward00:33:45 - say the ganglia tell it at sort of a very abstract level about the arm,
  • fast_forward00:33:50 - like, oh, the arm is still there.
  • fast_forward00:33:51 - Yeah. But it doesn't receive more detailed information from all the suckers, for instance.
  • fast_forward00:33:55 - So I think to answer your question, I think there are some behavioral experiments
  • fast_forward00:34:04 - that have been done by Wells in Napoli,
  • fast_forward00:34:11 - and they show that, for example, octopus, you can train octopus very nicely
  • fast_forward00:34:15 - to do tactile discrimination.
  • fast_forward00:34:18 - But you cannot train an octopus to do with one arm one task and with the other arm, different task.
  • fast_forward00:34:28 - The training, the learning is general to the entire arm.
  • fast_forward00:34:33 - To all arm? To all arm. So if I learn a tactile discrimination with arm number
  • fast_forward00:34:39 - one, I can also do it with arm number three? It will be generalized to all other
  • fast_forward00:34:42 - arms. So that would mean central brain is involved.
  • fast_forward00:34:45 - Yes, of course, it's involved in the learning, but it's not involved in the specific arm.
  • fast_forward00:34:53 - But still, no one has found some sort of somatotopic map of that in the central brain? No.
  • fast_forward00:35:02 - Not yet. And similarly, for example, is with motor command.
  • fast_forward00:35:09 - When octopus extend arm or several arm together, it never extend one arm in
  • fast_forward00:35:16 - a higher speed and the other one with a smaller one, with a slower speed.
  • fast_forward00:35:22 - It always will have the same speed as if a real, a single program has been evolved.
  • fast_forward00:35:31 - And maybe there is a gating mechanism that determine at the lower level which
  • fast_forward00:35:36 - arm will be activated by this motor program.
  • fast_forward00:35:42 - But now, about the sensory representation of let's say touch,
  • fast_forward00:35:50 - haptic information that comes from these arms,
  • fast_forward00:35:52 - people have looked for that and also found the same kind of broad distribution
  • fast_forward00:35:58 - in the central brain, or people have just not really looked for it yet in detail?
  • fast_forward00:36:05 - What I have showed today is that when we record from the central brain,
  • fast_forward00:36:12 - from the higher motor center,
  • fast_forward00:36:15 - we find that there is no specific representation of specific area in the body in the center brain.
  • fast_forward00:36:26 - But wait, I thought that was with respect to motor action.
  • fast_forward00:36:29 - What you showed us is for different movements, right?
  • fast_forward00:36:32 - No, but what we have shown also is that if you record from this side in the,
  • fast_forward00:36:45 - in the higher motor center in the brain, and you stimulate tactically different
  • fast_forward00:36:50 - parts of the body, you can get.
  • fast_forward00:36:55 - Response from different area in the body.
  • fast_forward00:37:00 - So if you get the response from one arm, you will get from the other arm and
  • fast_forward00:37:04 - even from the mantle or from the sacral.
  • fast_forward00:37:08 - So there is no, it seems that the conventional representation doesn't exist in between.
  • fast_forward00:37:16 - It still might be that this information is presented in different units that
  • fast_forward00:37:23 - are activated by by uh but you you would at this point in time given what you know say,
  • fast_forward00:37:28 - both the direct to the somatosensory processing and
  • fast_forward00:37:31 - motor control there is no no somatotopic exactly yeah organization yeah all
  • fast_forward00:37:38 - right so and that can also lead you to this point that that you pushed rather
  • fast_forward00:37:42 - strongly in your presentation that the octopus is able to generate these complex
  • fast_forward00:37:48 - complex behaviors without a map of its body.
  • fast_forward00:37:53 - So now I could argue with that by saying, well, one, you just haven't found it yet.
  • fast_forward00:37:58 - And the organization is such that it is not easy to interpret.
  • fast_forward00:38:02 - So there's still a body representation.
  • fast_forward00:38:04 - Yeah. Because in some sense, you do see that the central brain is involved in controlling the body.
  • fast_forward00:38:09 - We have this generalization that you also just described, right?
  • fast_forward00:38:13 - I can learn from one arm and execute with the other arm.
  • fast_forward00:38:15 - Yeah. Wouldn't that actually imply that there must be something like a body
  • fast_forward00:38:20 - representation somewhere and you just haven't found it yet?
  • fast_forward00:38:22 - Yeah, it very well can be.
  • fast_forward00:38:26 - But what I showed that at least two movements that I analyzed,
  • fast_forward00:38:34 - which is the reaching movement,
  • fast_forward00:38:37 - and the fetching movement, we can explain the motor program that generate this movement,
  • fast_forward00:38:46 - on the action of the peripheral nervous system.
  • fast_forward00:38:52 - Neuromuscular system itself. So, for example, in reaching movement,
  • fast_forward00:38:56 - we have shown that you can generate a perfectly well extension movement in amputated
  • fast_forward00:39:03 - arm by stimulating the arm nerve cord.
  • fast_forward00:39:08 - So it means that at least irrespective of central representation,
  • fast_forward00:39:13 - representation it is possible to generate such movement
  • fast_forward00:39:16 - in amputated arm it means that the program are embedded
  • fast_forward00:39:20 - in the well not necessarily
  • fast_forward00:39:23 - right because as we discussed earlier there might be hierarchical structuring
  • fast_forward00:39:27 - right so if you decorticate me i might still be able to make let's say walking
  • fast_forward00:39:31 - like rhythmic movements with my legs and that does not imply that the bit you
  • fast_forward00:39:37 - just removed from my brain is not representing my body or controlling it or
  • fast_forward00:39:41 - whatever just means i have
  • fast_forward00:39:42 - lower level reflexes that give structure to the behaviors I generate.
  • fast_forward00:39:46 - So, so isn't that then the same case here? Because we can look at,
  • fast_forward00:39:50 - let's say, let's look at the reaching actions.
  • fast_forward00:39:53 - So my argument would be, well, maybe the reaching is really like a very simple reflex.
  • fast_forward00:39:58 - It's a very important reflex for this animal that is basically used as,
  • fast_forward00:40:03 - as a movement primitive by higher level systems. Right.
  • fast_forward00:40:07 - So what you really show, and this is true, it is, it is definitely a reflex
  • fast_forward00:40:10 - that sits it's very close to the organization of a single arm because a single
  • fast_forward00:40:16 - arm can almost generate a reaching movement.
  • fast_forward00:40:20 - So how do reaching movements exactly come about? How does this work in the octopus arm?
  • fast_forward00:40:26 - So without the central brain, right? Yes. So what we think happens is that by
  • fast_forward00:40:31 - stimulating locally at the base of the arm,
  • fast_forward00:40:34 - we generate motor programs that propagate along the arm and activate both the
  • fast_forward00:40:42 - transversal and longitudinal muscle and create a stiffening wave,
  • fast_forward00:40:47 - which pushes actually passively at the bend forward.
  • fast_forward00:40:51 - And this has similar kinematics to what we see in a freely behaving animal.
  • fast_forward00:40:57 - So what we think is that maybe this movement or the command for this movement
  • fast_forward00:41:03 - are stored in the central brain.
  • fast_forward00:41:06 - So the way to activate this movement is stored in the higher motor center.
  • fast_forward00:41:11 - And it's based, and therefore we say that maybe in the central nervous system
  • fast_forward00:41:17 - what is decoded is the motor program rather than information based on the different part of the muscle.
  • fast_forward00:41:26 - So what you're saying is, look, I have the longitudinal muscle,
  • fast_forward00:41:29 - I have the transversal muscle, and then we have, again, a muscle sheet around that at the outside.
  • fast_forward00:41:35 - If I drive those from inside, so from proximal to distal in sort of,
  • fast_forward00:41:44 - it doesn't need to be a wave actually, it's a single wave front that moves to the edge.
  • fast_forward00:41:48 - And by activating all these muscles, I create stiffness of this hydrostat, this arm.
  • fast_forward00:41:54 - And I just have that instruction and just has to move distally through my arm.
  • fast_forward00:42:01 - This is roughly the idea. What the brain has to do is to scale this movement,
  • fast_forward00:42:09 - because we know that octopus can extend its arm in different speed.
  • fast_forward00:42:13 - And this probably is done via a central command.
  • fast_forward00:42:19 - But the propagating signal then passes through a number of neural stages.
  • fast_forward00:42:25 - So how many neurons, so how many synapses do I cross to go from proximal to
  • fast_forward00:42:32 - distal? Many, many, many.
  • fast_forward00:42:34 - There are 400,000 motor neurons in each arm.
  • fast_forward00:42:43 - And they are organized very closely one to each other.
  • fast_forward00:42:47 - And the nerve roots that go from the nervous system to the arm leave the root,
  • fast_forward00:42:55 - leave the central nervous system of the arm every 100 micrometer.
  • fast_forward00:43:00 - So there is a very, very refined innervation of the muscle structure by the neural system.
  • fast_forward00:43:12 - But what's also interesting is when we study the properties of the neuromuscular connection,
  • fast_forward00:43:20 - we find that this connection is unlikely in all other invertebrates.
  • fast_forward00:43:28 - It doesn't have any short-term plasticity.
  • fast_forward00:43:30 - So it seems more that the neural activity actually determines I mean,
  • fast_forward00:43:36 - in a more or less linear way, what will be the motor action?
  • fast_forward00:43:40 - And this fit, what we find is that octopus probably use a feed-forward program
  • fast_forward00:43:46 - to activate this kind of, for example, the reaching movement.
  • fast_forward00:43:50 - Okay, but then in the case of reaching, do these motor neurons have any kind
  • fast_forward00:43:55 - of spontaneous activity?
  • fast_forward00:43:56 - Or they must be driven by, let's say, a command neuron that sits at the base of the arm?
  • fast_forward00:44:03 - I think when, probably these motor neurons also are active during local movement of the arm.
  • fast_forward00:44:11 - There is searching, there is bending, there is a lot of movement that goes on in the amputated arm.
  • fast_forward00:44:19 - So this kind of movements also are controlled by the motor neuron of the arm.
  • fast_forward00:44:25 - But then I assume assumes that this is coordinated by a local circuit,
  • fast_forward00:44:30 - which is embedded in the local ganglia.
  • fast_forward00:44:35 - Each ganglia is located next to one of the 300 circles that are running along the arm.
  • fast_forward00:44:44 - When a central command is coming through the axonal track, which runs above the ganglia structure,
  • fast_forward00:44:51 - it's probably activate all of the motor neurons which are needed to produce
  • fast_forward00:45:01 - this brain directed movement.
  • fast_forward00:45:04 - So the central brain accents cruise all the way through the arm to the end? Yeah.
  • fast_forward00:45:08 - Okay. So there is a good radiation, but I think there is a lot of… for example, one of the possibility,
  • fast_forward00:45:19 - and I think one student asked this question during my talk, is how the octopus
  • fast_forward00:45:26 - generates this band at different locations along the arm.
  • fast_forward00:45:30 - So we had the idea that what might exist in this system is a labeled line,
  • fast_forward00:45:35 - where neurons from the brain are going to a specific site.
  • fast_forward00:45:40 - Along the arm, and the octopus determines where, not only to activate the movement, but also where.
  • fast_forward00:45:46 - But we didn't find yet any indication for such label line theory.
  • fast_forward00:45:53 - But that's for the fetching behavior, right?
  • fast_forward00:45:55 - So now, if we just look at the reaching… No, this is for the reaching, actually. Okay.
  • fast_forward00:46:02 - What I'm speaking about. Okay. In the fetching movement, we are less…,
  • fast_forward00:46:10 - Let's say our ideas are not that concrete because we don't know exactly what's
  • fast_forward00:46:15 - going on. So let's wait for, let's first figure out the reaching.
  • fast_forward00:46:20 - But now a typical thing, at least as a non-expert, you see in these octopus
  • fast_forward00:46:26 - movies, also the ones you showed, is sort of this wavy wiggling of the arms.
  • fast_forward00:46:31 - So that would suggest that there is also some sort of pattern generator behind that.
  • fast_forward00:46:37 - So do these motor neurons in the arm have pattern generator-like properties? Yes.
  • fast_forward00:46:45 - I think pattern generator is a valid term to use in this case,
  • fast_forward00:46:55 - but I'm not sure, for example, I don't think that it's based on rhythmical pattern generating.
  • fast_forward00:47:03 - So you can generate patterns that will induce, for example, arm reaching.
  • fast_forward00:47:08 - So this is pattern, generating patterns which is producing the behavior.
  • fast_forward00:47:13 - But it's not in the general sense when we speak on central pattern generator,
  • fast_forward00:47:20 - where we create a rhythmical movement mainly to control locomotion and coordination between arms.
  • fast_forward00:47:30 - This, I think it's less likely, but… So it's more like a variable,
  • fast_forward00:47:36 - a more variable pattern generator.
  • fast_forward00:47:38 - But are you saying it's maybe not so dependent on single cells and more dependent
  • fast_forward00:47:42 - on networks of cells? Is that really the implication?
  • fast_forward00:47:45 - Yeah, I think it's very dependent on network connection in the arm.
  • fast_forward00:47:51 - But now, what you talked to earlier, which is important here,
  • fast_forward00:47:56 - is that, okay, when we talk about reaching, we sort of have a wave of stiffening
  • fast_forward00:48:00 - running through this arm.
  • fast_forward00:48:01 - Arm, if we now go to a next level of behavior, which is more complex,
  • fast_forward00:48:06 - which is fetching, what you showed is if an object touches the arm at some point,
  • fast_forward00:48:13 - this leads to, and it wants to now fetch this object to bring it to the mouth,
  • fast_forward00:48:18 - or however you call that in the Octopus.
  • fast_forward00:48:22 - Now, you see a very specific reconfiguration of the arm. Right.
  • fast_forward00:48:28 - That it creates even virtual joints in the arm.
  • fast_forward00:48:31 - So, how does that work? So, we correlate this behavior with a recording, EMG recording.
  • fast_forward00:48:40 - It's a very complex behavior, but we managed to do it.
  • fast_forward00:48:44 - And what we discovered that probably the touching of the food in one of the
  • fast_forward00:48:52 - sacchar activate a two-wave of propagation of muscle contraction,
  • fast_forward00:48:58 - One which is coming from the base of the arm and one which goes from the more
  • fast_forward00:49:06 - or less place where the arm touches the food and they are propagating in the
  • fast_forward00:49:13 - same more or less the same speed one toward the other.
  • fast_forward00:49:18 - And in such a form that where the point where they are colliding,
  • fast_forward00:49:24 - this will set the site where the virtual elbow, if you like, would be created.
  • fast_forward00:49:31 - And then in a kind of motor programs that we don't understand yet,
  • fast_forward00:49:39 - there is a rotation of these elbows that brings the food to the mouse exactly
  • fast_forward00:49:46 - as we do when we take something with our hand and bring it to the mouse.
  • fast_forward00:49:51 - But this configuration, reconfiguration looks very complex, right?
  • fast_forward00:49:54 - Because it means I have to stiffen the part that becomes a segment.
  • fast_forward00:49:58 - Yeah. I have to then release a small part that becomes my virtual elbow.
  • fast_forward00:50:05 - Yeah. However, I also have to give that a directionality. Yeah.
  • fast_forward00:50:09 - It cannot be, let's say, flexible in all directions. Yeah. It's a constrained flexibility.
  • fast_forward00:50:14 - Yeah. So how is that done? No, the arm has some structure that goes from the
  • fast_forward00:50:19 - dorsal to the ventral part.
  • fast_forward00:50:20 - So, you can imagine that all movements are done in the same planarity,
  • fast_forward00:50:26 - where the soccer is pointing downward.
  • fast_forward00:50:32 - So maybe, you know, that such a bend in the arm,
  • fast_forward00:50:36 - which serves as the elbow, and we don't know yet, and this is one of the experiments
  • fast_forward00:50:41 - that we would like to do, if simply two-way, because of the structure,
  • fast_forward00:50:46 - of the muscular structure,
  • fast_forward00:50:47 - it will be enough that two waves of activation will collide with each other
  • fast_forward00:50:52 - due to the structure of the muscular system,
  • fast_forward00:50:56 - a band, a rotation will be created at this point.
  • fast_forward00:51:00 - Otherwise, it may need a more specific activation of dorsal versus ventral part of the musculature.
  • fast_forward00:51:11 - But why would the collision of these two waves give me a joint?
  • fast_forward00:51:14 - I mean, then you would expect that dependent on the width of your wave.
  • fast_forward00:51:19 - Your joint would be a variable size. size, and also that you won't have such
  • fast_forward00:51:25 - a very discrete transition.
  • fast_forward00:51:27 - Yeah. Well, if you look at the behavior, it seems very well circumscribed where
  • fast_forward00:51:30 - you insert that joint, no? Yeah, I think you're right.
  • fast_forward00:51:34 - And therefore, I think the only way is to test it by an experiment to see if
  • fast_forward00:51:40 - first, maybe it's created by two waves of muscle activation.
  • fast_forward00:51:45 - So what we have to do is to stimulate the arm from two sides and see if we can
  • fast_forward00:51:51 - control controlled the point where the bend will be initiated, or it won't happen.
  • fast_forward00:52:00 - And then we may look for a mechanism whereby more specific activation of the muscle,
  • fast_forward00:52:08 - meaning that the dorsal muscle will stiffen while the ventral part will shorten
  • fast_forward00:52:14 - in order to create this bend.
  • fast_forward00:52:17 - But what you also showed is that dependent on the distance from the item,
  • fast_forward00:52:24 - from the stimulus that touches the arm to the base of the arm,
  • fast_forward00:52:29 - dependent on the distance,
  • fast_forward00:52:30 - the length of the segments will vary. This is one thing you showed.
  • fast_forward00:52:35 - But then is the number of virtual joints I generate always the same? Like I think it's two.
  • fast_forward00:52:41 - In your examples, it was always two. Two, yes, with one distal one,
  • fast_forward00:52:47 - which serves as a hand, but the two are exactly the same.
  • fast_forward00:52:52 - And this virtual segment are created in accordance with the site.
  • fast_forward00:53:01 - Where the foot is was taken so this is a dynamical uh uh way of constructing an an elbow,
  • fast_forward00:53:09 - in accordance with with what uh with what uh with where the food is uh has been
  • fast_forward00:53:16 - uh uh catched and and this is of course a dynamical structure completely different,
  • fast_forward00:53:23 - from that's already important no because it means it will never generate three
  • fast_forward00:53:28 - joints or one joint No, it's dictated by a very specific motor program, this structure.
  • fast_forward00:53:36 - But does it also already tell you some critical properties of the wave-like
  • fast_forward00:53:42 - response that you depend on?
  • fast_forward00:53:45 - Because that means this wave has a certain length, a certain boundary on its length.
  • fast_forward00:53:49 - It cannot extend beyond, let's say, what the segment length dictates,
  • fast_forward00:53:55 - and it can never get shorter than that the interference pattern gives rise to more than two joints.
  • fast_forward00:54:03 - So doesn't this wave-like response that you built your argument on,
  • fast_forward00:54:07 - doesn't the number of segments we get and the intersegment distance tell you
  • fast_forward00:54:12 - something about the phase of that or the period of that wave?
  • fast_forward00:54:22 - Because you talk about two colliding waves that give you joints.
  • fast_forward00:54:25 - Yes. So if you already know that these joints must be at some minimum distance,
  • fast_forward00:54:29 - it tells you something about the period and the phase of these two waves.
  • fast_forward00:54:32 - Yeah, but the time and the period, it depends on what determines them is the
  • fast_forward00:54:38 - site where the food was grabbed.
  • fast_forward00:54:40 - So the time and the distance is actually the variable which determines the size
  • fast_forward00:54:48 - of the the articulated structure.
  • fast_forward00:54:50 - And this depended on where the octopus got the target.
  • fast_forward00:54:55 - Right. So now we see that actually locally, such an arm already has a lot of capabilities, right?
  • fast_forward00:55:02 - In terms of its articulation, controlling these unspecified numbers of degrees of freedom.
  • fast_forward00:55:07 - And actually when it has to act, it freezes the degrees of freedom and reduces them into few. Right.
  • fast_forward00:55:14 - Maybe, I guess, four, right? Because of the base of the arm.
  • fast_forward00:55:18 - I think it's three. Well, I was thinking maybe the wiggly tip might count. I don't know.
  • fast_forward00:55:23 - Okay. Yeah, but three, let's say.
  • fast_forward00:55:26 - So, this is interesting. Basically, what Octopus is doing with its hydrostat
  • fast_forward00:55:31 - is given the contact, it just freezes its degrees of freedom. Right. Right?
  • fast_forward00:55:36 - But the point is that with that, it can build multiple configurations,
  • fast_forward00:55:39 - which then can solve the task. This is the beauty, because this is also what
  • fast_forward00:55:43 - people want to simulate in a flexible robot. Sure. This...
  • fast_forward00:55:51 - New special space of controlling the
  • fast_forward00:55:54 - structure controlling the the the shape
  • fast_forward00:55:57 - shape of the of the robot right reshaping it's
  • fast_forward00:56:01 - a very robust mechanism to you to build a flexible robot right yeah i want i
  • fast_forward00:56:06 - want to get to that but first i want to understand something else better now
  • fast_forward00:56:10 - we have an idea about what arms can do we have an idea about how reflex systems
  • fast_forward00:56:14 - can drive this arm for especially reaching.
  • fast_forward00:56:19 - And now we talk about those fetching as a reflex-driven response.
  • fast_forward00:56:23 - But we also have here this octopus with these huge eyes that are really highly
  • fast_forward00:56:27 - developed. The eyes are linked to the central brain.
  • fast_forward00:56:30 - And with the eyes, it will detect prey. And then given the prey detection,
  • fast_forward00:56:34 - the central brain will decide to reach for that prey and catch it, right?
  • fast_forward00:56:40 - So now, how do you see then the central brain interface Interface to the arms
  • fast_forward00:56:46 - and the arm control systems.
  • fast_forward00:56:48 - You mean in goal direction? In reaching movement. To make it goal directed. Exactly right.
  • fast_forward00:56:54 - We didn't study it, but I think what important part of it would be the base of the arm.
  • fast_forward00:57:00 - We don't know even the muscular organization, if it has some specific structure or not.
  • fast_forward00:57:06 - We know that along the arm, the muscular organization is exactly the same.
  • fast_forward00:57:12 - It's very similar, despite this tapering toward the end.
  • fast_forward00:57:16 - So, we assume that there is some system which is directing the base of the arm
  • fast_forward00:57:24 - in the direction of the target.
  • fast_forward00:57:26 - So, when the arm is stretched, it reaches the target.
  • fast_forward00:57:30 - It might be also that the octopus are using feedback control to correct this movement.
  • fast_forward00:57:37 - But in some experiments that we did where we trained the octopus to reach to
  • fast_forward00:57:44 - a target, and after the arms start to move, we move the target.
  • fast_forward00:57:50 - We see that the movement continues as a ballistic behavior.
  • fast_forward00:57:54 - This means that it's really a feed-forward kind of movement that the octopus
  • fast_forward00:57:59 - cannot readjust it after it has been initiated.
  • fast_forward00:58:04 - But maybe also because it moves in a medium, it would be really difficult to
  • fast_forward00:58:08 - control it differently, right?
  • fast_forward00:58:10 - Yeah. No, but what is nice, because the octopus is so fast learning,
  • fast_forward00:58:15 - that it's very difficult to do this experiment because it very fastly understands
  • fast_forward00:58:19 - that the movement is going to be moved.
  • fast_forward00:58:23 - So it takes it into account when you start the movement and can extend its arm
  • fast_forward00:58:28 - to where it knows or assumes that the target will be ended.
  • fast_forward00:58:35 - So how accurate is it then in these goal-oriented reaching movements?
  • fast_forward00:58:38 - It's not very accurate in the reaching. But you have to remember that the arm is very flexible.
  • fast_forward00:58:44 - So even if it reaches the target close enough, we can do then a wiggle or something
  • fast_forward00:58:53 - and to get the target by moving it a little bit.
  • fast_forward00:58:58 - Okay, so the central brain just has to do a pretty rough target setting.
  • fast_forward00:59:03 - Yeah. It doesn't need to be very precise.
  • fast_forward00:59:05 - Yes. So that would be with an accuracy of centimeters?
  • fast_forward00:59:08 - Yes, I think a few centimeters. Okay.
  • fast_forward00:59:11 - So you mentioned also when we looked at octopus motor control,
  • fast_forward00:59:19 - that it is also an example of what people call morphological computation,
  • fast_forward00:59:23 - for how properties of the body itself contribute to solving the control problem.
  • fast_forward00:59:29 - So how do you see that link exactly?
  • fast_forward00:59:32 - I think the examples that we see for this kind of computational control is really
  • fast_forward00:59:36 - what we just spoke about, and this is the fetching movement. Thank you for watching!
  • fast_forward00:59:42 - So I think that the idea is that the site of the elbow is controlled by propagating
  • fast_forward00:59:49 - wave on the structure itself.
  • fast_forward00:59:51 - And this is what the colliding point is determined the site of the elbow.
  • fast_forward00:59:57 - It's a sort of computational, which depends on the morphology of the structure.
  • fast_forward01:00:04 - But wait, is that fair to say? Because you could also argue it's essentially
  • fast_forward01:00:08 - a neural process that gives you your wave.
  • fast_forward01:00:10 - And it's it's it's interaction between the
  • fast_forward01:00:13 - neural dynamics right let's say target induced
  • fast_forward01:00:16 - waves and base induced waves that sets up the muscular response right and that
  • fast_forward01:00:22 - then gives you the this is completely completely true but taking the advantage
  • fast_forward01:00:27 - that the nervous system is in parallel with the physical structure you can use
  • fast_forward01:00:33 - the propagation of the activity in the nervous system,
  • fast_forward01:00:36 - which is parallel to the physical structure of the arm, as a way to compute
  • fast_forward01:00:45 - where to build the structure.
  • fast_forward01:00:47 - So the structure of the arm is an important component.
  • fast_forward01:00:52 - The morphology of the arm itself, right. But it would place you a little bit
  • fast_forward01:00:56 - in a different camp than the more extreme view that some people have expressed,
  • fast_forward01:01:01 - that in some sense it can be pure morphology, or only the body that gives you
  • fast_forward01:01:06 - an adaptive behavior in your proposed it would still depend on nervous system intervention.
  • fast_forward01:01:12 - Yeah. Right? So in that sense it's not like an extreme... Yeah, but...
  • fast_forward01:01:17 - So for example, if you want maybe the sucker of the arm is built...
  • fast_forward01:01:23 - It can work without a sensory signal. It can, you know, attach to a surface by a simple physical...
  • fast_forward01:01:36 - Vacuum mechanism which doesn't need any.
  • fast_forward01:01:43 - Control. And actually what seems that the octopus have a special mechanism to
  • fast_forward01:01:52 - release the sucker instead of attaching.
  • fast_forward01:01:55 - Maybe attaching is occurring spontaneously but releasing the structure is more...
  • fast_forward01:02:03 - Okay, yeah, that's a good example, really, where morphology is directly implementing a function.
  • fast_forward01:02:08 - I would agree with you. But you were also mentioning, for instance,
  • fast_forward01:02:11 - Rod Brooks saying that the best robot has no control. Yeah.
  • fast_forward01:02:15 - I'm not sure the octopus would then be such a great robot if this is our definition.
  • fast_forward01:02:23 - There's still control. Yeah. Yeah, but I think that the flexible arm and that
  • fast_forward01:02:34 - is equipped, it's very important.
  • fast_forward01:02:38 - The sacchar is a very important part of this structure.
  • fast_forward01:02:43 - If there were no sacchar distributed along the arm, the behavior would have
  • fast_forward01:02:49 - been completely different. So, because you can imagine that now the sacral fingers
  • fast_forward01:02:54 - distributed all along the arm.
  • fast_forward01:02:57 - And this allow using such flexible motor program to generate the behavior.
  • fast_forward01:03:09 - I think in the Woodley-Brooke idea, building this kind of arm,
  • fast_forward01:03:14 - which is flexible, but including many types of passive soccer,
  • fast_forward01:03:19 - is a good idea to start in building a flexible robot.
  • fast_forward01:03:24 - Right. And put the control on these physical capabilities of the arm.
  • fast_forward01:03:32 - Right. But it's interesting that often people don't put their money where their mouth is, right?
  • fast_forward01:03:36 - Because with Rod Brooks, for instance, he is actually running a company selling
  • fast_forward01:03:41 - robots now, Baxter, a configurable industrial robot for fine manipulation.
  • fast_forward01:03:46 - And it definitely has control structures in it. And it actually uses very little
  • fast_forward01:03:51 - of morphological computation to be a successful product.
  • fast_forward01:03:54 - So it would be nice when people live up to their own sermons in that respect,
  • fast_forward01:04:00 - right? So, Benny... But I think it's a good constraint in simplifying the control.
  • fast_forward01:04:07 - I don't think it can replace control, but it may make control much simpler.
  • fast_forward01:04:16 - Absolutely. I agree with you. This is a good point, yes, absolutely.
  • fast_forward01:04:21 - So, Benny, you've been studying the octopus now for how long?
  • fast_forward01:04:25 - How many years? 20 years.
  • fast_forward01:04:27 - Okay, so So, and you learned a lot about behavior, the nervous system of the
  • fast_forward01:04:32 - octopus, so many things we have to discover.
  • fast_forward01:04:35 - So, if we want to now take your experience on board in how we started the brain,
  • fast_forward01:04:40 - what would be Benny's law?
  • fast_forward01:04:44 - I think that….
  • fast_forward01:04:49 - And it's very dangerous to try to understand the brain without the body.
  • fast_forward01:04:58 - And when people want to understand how behavior is generated and controlled,
  • fast_forward01:05:05 - they can study the brain in isolation just in order to find the properties of
  • fast_forward01:05:12 - the component of the central brain.
  • fast_forward01:05:18 - But to understand how these properties are organized,
  • fast_forward01:05:23 - maybe self-organized, in order to control the behavior,
  • fast_forward01:05:28 - you must, part of your study, do when the animal is activating the body and
  • fast_forward01:05:38 - better even in behavior.
  • fast_forward01:05:40 - And I think the idea of in vivo recording and doing experiments that involve
  • fast_forward01:05:45 - monitoring the action of the brain while doing actual behavior is more and more
  • fast_forward01:05:52 - common now in our days. Right. Yeah.
  • fast_forward01:05:59 - So in some sense, from this embodiment perspective, you also have been looking
  • fast_forward01:06:04 - at this emerging field of soft robotics.
  • fast_forward01:06:06 - And do you think soft robotics has really taken on board enough of the insights
  • fast_forward01:06:13 - and lessons from the study of Octopus yes.
  • fast_forward01:06:18 - I think the main problem in soft robotics is the material.
  • fast_forward01:06:25 - There is no material like musculoskeletal system.
  • fast_forward01:06:28 - So effective, so robust, so high strain.
  • fast_forward01:06:35 - And this is what limits the soft robotics at this stage, I think.
  • fast_forward01:06:42 - But I think there is a lot of progression in this area.
  • fast_forward01:06:45 - At this point, what people do is that they combine different ideas taken from
  • fast_forward01:06:54 - the octopus, for example.
  • fast_forward01:06:57 - So it's not only constant volume constraint, but it can be the controlling of
  • fast_forward01:07:05 - stiffness, it can be particle jamming, other ideas about that.
  • fast_forward01:07:12 - If you think about it, it may come from biology and be implemented in the system.
  • fast_forward01:07:21 - I think also from the control point of view,
  • fast_forward01:07:25 - the idea of distributing the control between the central and peripheral and
  • fast_forward01:07:30 - to leave much of the control and even more complex part of it,
  • fast_forward01:07:34 - like even learning a memory, learning at the level of the arm,
  • fast_forward01:07:40 - are ideas that comes from such studies that we are doing in the octopus.
  • fast_forward01:07:47 - Right. So then, last question. Five years from now, we're going to come visit
  • fast_forward01:07:51 - you in Jerusalem, and we're going to check whether a prediction you make today
  • fast_forward01:07:57 - has actually been verified or not.
  • fast_forward01:07:59 - So, what's the most important prediction or hypothesis you would like to see
  • fast_forward01:08:04 - verified in that timeframe of five years?
  • fast_forward01:08:09 - So, as I mentioned, we have two projects.
  • fast_forward01:08:15 - Now we are more into the learning memory mechanism.
  • fast_forward01:08:20 - And we are studying what is the mechanism of learning memory in the octopus.
  • fast_forward01:08:25 - Octopus, and we may discover a new solution to the problem of how the nervous
  • fast_forward01:08:35 - system stores information.
  • fast_forward01:08:40 - And I believe that there are many ways, and the octopus is a very good example,
  • fast_forward01:08:45 - there are many independent ways to build a complex brain. train.
  • fast_forward01:08:49 - But at the end, they will point to what is the universal importance in networks
  • fast_forward01:08:57 - that can mediate learning and memory.
  • fast_forward01:09:01 - And I hope that we will learn enough on the mechanism of motor control of the
  • fast_forward01:09:07 - octopus that it will be really more feasible to implement in the soft robotic.
  • fast_forward01:09:16 - Hopefully, while we are doing our biological research, the search into attenuator
  • fast_forward01:09:24 - material will progress as well. Right.
  • fast_forward01:09:29 - Okay. Benny Hochman, thank you very much for this conversation.
  • fast_forward01:09:32 - Thank you. Hochner. Hochner, sorry.
  • fast_forward01:09:37 - The CSN podcast was produced by the Convergent Science Network of Biometrics
  • fast_forward01:09:42 - and Biohybrid Systems, a project funded by the European 7th Research Framework Program.
  • fast_forward01:09:50 - For more interviews, recorded lectures, or upcoming conferences in the field
  • fast_forward01:09:56 - of biometrics and biohybrid systems, go to csnnetwork.eu.
  • fast_forward01:10:03 - Music.

Be the first to leave a comment

Leave a comment

Your email address will not be published. Required fields are marked *

Exploring the convergence of neuroscience, robotics, and AI through conversations with leading researchers since 2010.

A project of the Convergent Science Network Foundation.

© CSN Podcasts. Developed by IMCreativeWEBC

0%

Login to enjoy full advantages

Please login or subscribe to continue.

Go Premium!

Enjoy the full advantage of the premium access.

Stop following

Unfollow Cancel

Cancel subscription

Are you sure you want to cancel your subscription? You will lose your Premium access and stored playlists.

Go back Confirm cancellation