Article – Suzan Mazur
Cyberneticist Slawomir Jaroslaw Nasuto has a radiance that can only be Polish. He was born in eastern Poland, in Lublin, during the Soviet years (Stalin was embraced by the post-WWII provisional Polish government headquartered there) and is now doing …
Darling, Which of Us Is the Robot?– Slawek Nasuto
by Suzan Mazur
December 2, 2013
SLAWOMIR JAROSLAW NASUTO
“Visual experience is just simulations of the retina, but we impose an extremely rigid interpretation of it. We see the world in terms of trees and dogs and rivers, etc. But often the question is, what are those concepts?” — Noam Chomsky
Cyberneticist Slawomir Jaroslaw Nasuto has a radiance that can only be Polish. He was born in eastern Poland, in Lublin, during the Soviet years (Stalin was embraced by the post-WWII provisional Polish government headquartered there) and is now doing ground-breaking research on the brain and embodiment at the University of Reading in the UK. Some of this work he shared weeks ago at a cognitive robotics conference in Bergamo, Italy — where we met — in particular his experiment with the “animat”: a CLOSED-LOOP system of fetal brain cells in a dish over electrodes + computer + robot.
Click for big version.
Slawek (pronounced Suave’) Nasuto was fascinated by math and science at an early age, as well as by Western literature smuggled into Poland. His father was a professor, now retired, of physical chemistry at Maria Sklodowska-Curie in Lublin, the university named for Madame Curie, the first woman to win a Nobel Prize.
Nasuto’s MS degree is in Pure Mathematics from Maria Sklodowska-Curie and his PhD in Cybernetics from the University of Reading. He is currently associate director at Reading’s Centre for Integrative Neuroscience and Neurodynamics and led the initiative to establish the new Brain and Embodiment Laboratory (BEL) there at the School of Systems Engineering, where he serves as a director; BEL will analyze data at multiple scales, including multi-channel electrophysiology and EEG in various closed-loop experimental paradigms.
Nasuto has authored and/or co-authored 120 papers in computational neuroanatomy and neuroscience, the role of synchronization in cognitive processing, EEG-based brain-computer interfaces, the relationship between structure and function in individual neurons and their networks.
Despite some of the challenging questions I put to Nasuto during our conversation, he remained upbeat about the benefits from robotics research. His enthusiasm and that of his colleagues at the University of Reading has also caught the attention of BBC.
I last saw Slawek Nasuto in September in the grip of one of the alpha males at the Bergamo event who was trying to extract from him just how Origin of Life might intersect with robotics — the answer so far remains elusive. . .
My interview with Slawek Nasuto follows.
Suzan Mazur: What was it like growing up in Poland during the Soviet years?
Slawek Nasuto: Not so remarkable. I was born in Lublin, a city in eastern Poland. I spent all my childhood there. I had a very influential math teacher in secondary school, a really strong character. Thanks to him the majority of our class — 17 – 20 students out of 30 — went on to study mathematics. Quite unusual. I was actually interested in science from a young age.
Suzan Mazur: Do you come from a science family?
Slawek Nasuto: Yes, my father, now retired, was professor of physical chemistry.
Suzan Mazur: Where was he a professor?
Slawek Nasuto: At the University of Maria Sklodowska-Curie in Lublin.
Suzan Mazur: Was your mother also a scientist?
Slawek Nasuto: She was not. She has always been an avid reader. I developed a passion for reading from my mother.
Suzan Mazur: What else stands out about those years?
Slawek Nasuto: Poland was relatively isolated from the Western world and I was not very much interested in politics. Kids tend to think of the environment they grow up in as normal. If a kid doesn’t have an example of an alternative, it’s difficult for them to compare.
Suzan Mazur: The Soviets were promoting science.
Slawek Nasuto: Right, yes. Science books in Poland were very cheap. Translations were actually mostly from the East Bloc. For anybody interested in culture and reading it was also possible to find books.
When I was growing up the censorship wasn’t that strong, so many of the books, even Western classics were available. Some were not. As a teenager I remember reading George Orwell’s 1984. My best friend got the book somehow. It was an illegal copy smuggled to Poland from a publishing house in Paris and translated into Polish.
Suzan Mazur: So you benefited from an East and West perspective in science?
Slawek Nasuto: To some extent yes. In secondary school I was very, very interested in popular science.
Suzan Mazur: What are some of your interests outside science?
Slawek Nasuto: I tried a lot of things when I was younger. I was extremely skinny as a kid so I began working out at the gym. My parents were afraid that I had to put some weight on, to develop muscles or I’d have problems with posture. I’ve been going to the gym ever since. Also, I have a fascination with the Far East and have investigated different martial arts. I like sailing as well.
Suzan Mazur: Did you say your family name Nasuto may be Eastern?
Slawek Nasuto: We don’t really know where it comes from. It’s a very rare name in Poland.
Suzan Mazur: You led the University of Reading’s ground-breaking research on “animats.” In your recent papers, you describe the animat as a system with culture and robot coupled via a closed loop, the culture consisting of tens of thousands of dissociated cortical neurons and glia cells taken from a fetal rat and placed in a Petri-like dish with a grid of electrodes embedded at its bottom.
I understand the culture was then pulsed with electrical signals from a computer (also part of the closed loop) and within an hour of placement there was a reconnecting with other neurons and internal communication — chemically and electrically — even without the electrical pulsing.
Slawek Nasuto: Actually, all this happens relatively gradually with the first signs of cells reconnecting within hours from seeding and then activity coming on once the connection reaches a sufficient density. We have found considerable variation between cultures, as well. Because there are experimental factors we can’t precisely control, e.g., seeding density of cells, and also due to the intrinsic variability of cultures. So there are large ‘error bars’.
Suzan Mazur: The robot (part of the closed loop) was wirelessly controlled using bluetooth, a system also used to print photos from a mobile phone, for instance. Can you round this out for us? How soon after placement was the culture signaling to the robot?
Slawek Nasuto: The culture typically takes some time for the cells to settle. Once the cells settle, they start to form connections. As the culture becomes more dense, it begins to communicate — typically we see some level of connectivity and activation at 24 – 48 hours.
One goal of our research was to connect the robot to the culture as soon as possible and let it develop as it interacts with the environment. We could not achieve that in the original project for technical reasons. We are about to restart the wet lab in the university’s Brain and Embodiment Laboratory so this line of research is going back on the agenda.
Suzan Mazur: You say the culture matures at one month and lasts for three months. Meaning the animat “dies” after three months?
Slawek Nasuto: Right “dies” with quotes, the culture is not a sentient being. In some of my papers I argue this is actually not the case (e.g., the zombie mouse). The culture will survive as long as laboratory conditions enable it. It’s possible to keep cultures for even longer, given sufficiently stringent laboratory procedures, but it is tricky. What’s important is that we were able to keep the culture alive for up to three months.
Suzan Mazur: Do you toss or preserve the experiment when it dies?
Slawek Nasuto: We make records of the culture. When it dies, it is disposed of.
Suzan Mazur: Did you also experiment with adult brain cells, and, if so, did they exhibit a similar plasticity?
Slawek Nasuto: We used an immature culture and waited until it matured before we connected it with the robot. Again, one of the goals was to connect as early as possible. We did not start any of the experiments with adult brain cells. Fetal cells are used because they’re easier to maintain in a healthy and responsive mode. But, yes, even in adult cells there is some plasticity.
Suzan Mazur: You conclude in your paper “Controlling a mobile robot with a biological brain” that “a robot can have a biological brain to make decisions,” that there’s a need for further investigation. But do you envision those robotic decisions in the future to be meaningful decisions?
Slawek Nasuto: That’s a million dollar question. Let me answer indirectly. When we were building the animats, we didn’t think about whether the robotic decisions would be meaningful, in the intuitive understanding of the term. This was a ground-breaking experiment, our very first approach.
We treated the culture as almost an input-output device. That by itself is not really taking too much of the culture’s biology into account. But when we talk about systems supporting cognitive processing, we are dealing with a system, the brain, structured by evolution and development. We have evolved very special signaling pathways and different centers in the brain thought to be involved in some cognitive processes.
If we take this biology seriously into account in construction of future animats, I’d say the jury is out as to whether robotic decisions can be meaningful decisions.
Suzan Mazur: What is the big lesson learned from this experiment?
Slawek Nasuto: In experiments in which one uses tissue from living beings, the objectives tend to be much more specific, questions focused on tangible (even if ambitious) aims that can more directly counterbalance the ethical issues inevitably associated with experimentation.
Big philosophical questions may or may not be answered but they by themselves cannot be the only reasons to conduct such experiments. One of our long-term goals was to create an alternative experimental platform for testing novel pharmacological agents to treat neurological disorders. For that we needed a system that offered us great access to information about how the animat works.
Its operation may be plausibly mapped onto information processing going on in the brain/nervous system when it engages in cognitive processing. At the same time the animat is not a sentient being, so that it does not suffer because of the experimentation.
I believe the animat will eventually be such a platform. In the future the creation of an animat capable of advanced information processing will contribute to a reduction of animal suffering in experimentation. This will also offer us unparalleled access to data as the animat processes information — data we cannot for technical and ethical reasons now collect so easily (or at all) from animal studies.
It was in our Templeton-funded project that we tackled the question as to whether such animat could be thought of as a sentient being. We concluded that it could not. However to answer this we needed to look at the meaning of computing, Turing or otherwise. We argued that arbitrary formal manipulation of nervous tissue does not lead to any cognitive capacity in it.
This is quite important because, if animat, which after all has a biological “brain” to speak of, can not be sentient, so can not robotic devices. No matter how sophisticated or human-like in appearance they might be.
Suzan Mazur: That leads to my next question. You cite in the above-mentioned paper, which appeared in Defence Science Journal, a publication of the Indian government’s defense department, the experiments of others who successfully sent control commands from a lamprey to a robot, and also from a mobile phone or PC to living animals, such as cockroaches and rats, with electrical implants and microprocessor backpacks. You question the ethics of such experiments. Here’s the quote:
“Regehr demonstrated that it was possible to use the brain of a lamprey to control the trajectory of a robot whilst others were successfully able to send control commands to the nervous system of cockroaches or rats as if they were robots. Although such studies can inform us about information processing and encoding in the brains of living animals, they do pose ethical questions. . . ” — Warwick, Xydas Nasuto, et al., “Controlling Mobile Robot with a Biological Brain,” Defence Science Journal
Would you comment further? What are the implications? Where could this lead us?
Slawek Nasuto: These studies are trying to understand how the nervous system works in order to understand when it goes wrong. So we can create better therapies and interventions, etc.
The problem in the case of the experiments you just cited, obviously, is that they are performed on sentient beings. The tests are carried out on a nervous system or on an entire animal. We can argue to what extent these animals have consciousness, and what that level of consciousness is, but they do have feeling. Hence the ethical questions become important when such experiments are performed.
Suzan Mazur: Well why is the Indian government, the Indian Ministry of Defense interested in this?
Slawek Nasuto: I think it was just an open journal. I don’t think there was any specific interest from the Indian military.
Suzan Mazur: Is there any understanding how far this can be taken at present in terms of controlling the nervous system of other animals and people? What’s doable?
Slawek Nasuto: At present, there are different kinds of research on hybrid systems, technologies interfacing with the nervous system. Ted Berger in the US, e.g., in experiments with rats, and more recently, monkeys, is attempting to build an electronics system able to restore memory formation. The system is based on a chip implant that collects data from one area of the hippocampus and applies an appropriate transfer function before passing the activation on to other areas. It was demonstrated to mimic the operation of a damaged natural information pathway. This work can be very promising if it can scale to humans and to more natural scenarios than lab-constrained experiments.
My colleague from the University of Reading, Evangelos Delivoupoulos, has also been involved in development of flexible electrodes. Electrodes are typically placed either in the sensory cortices and help decode neural signals, or in the motor cortex in order to learn how the brain controls movement.
The flexible electrodes Delivoupoulos has been developing can be chronically implanted in nerves controlling the bladder — technology that offers promise in increasing quality of life, autonomy and dignity to people with lower spinal cord damage.
Suzan Mazur: What are your thoughts about the possibility of a non-algorithmic Trans-Turing System — “a new class of information processing system” — and the statement that one approach would be to “Simulate the TTS on a digital computer and evolve a population of TTS” in light of what Oxford philosopher Vincent Müller recently told me, which is:
“My view is anything to do with our computer systems is fundamentally algorithmic. If the system that we’re trying to generate on that is not fundamentally algorithmic, then we won’t be able to generate it.”
Slawek Nasuto: The answer is not straightforward. Personally, I agree with Vincent Müller’s statement, although the problem is that during the course of research in this area I came to appreciate as well that even in computer science there is no universal agreement on what computers can and cannot do. I’m trying to provide an answer that does not cut corners in my reply because this is one of the cases where I believe simplification leads to further confusion.
The answer is not simple, in part because we use computers everyday in science and engineering to approximate functions that are inherently non-symbolic. This is the pragmatic part of the answer. It does not address in a fundamental way away the question of Turing computing limitations of computers but it blunts the strength of the fundamental objections to their computing power.
The more subtle answer is that we have also begun to use computers in ways that go beyond the reach of formal algorithmic descriptions. Operating systems, word processors, Internet. These are just a few examples of such uses.
One of the characteristics of the above examples is the inherent interactions with the outside, the environment. The purpose is to engage in a successful interaction rather than compute specific function.
Such systems go beyond the computational capacity of a classical Turing machine, but that does not mean there is anything mysterious about them. For sure, they are built of algorithms, so each well-defined fragment is subject to the formal Turing-style description. But they are put together in a way that escapes such a formalism.
To give an analogy, it is a bit like a link and a chain. A link made of metal is rigid, cannot bend. Yet a chain made simply of such links sequentially interlocked can be as flexible as a rope. The system acquired a characteristic that is not present in the individual element.
What I am saying is that even with the Turing machine we can do things that go beyond Turing computability limitations. Because of computers we put a man on the moon. Tremendous advancements in science have been made thanks to computers that can approximate.
But the notion of computing is subject to active research, and extensions of classical computing have been proposed. In our group at the University of Reading we are trying to look at the possibilities of clarifying these issues. What computing is. Because there is a lingering belief in some areas of cognitive science that the brain can be thought of as performing formal manipulations on symbolic representations of reality akin to a Turing machine. And the Turing machine is a very specific model of an algorithmic process.
Our Reading group, in its investigation of what is computing, started from the premise that the notion of computing — in spite of its usefulness — cannot fundamentally describe mind or cognition. Computing thought is an inappropriate statement.
Suzan Mazur: The Turing Machine was never actually built, I understand.
Slawek Nasuto: The Turing Machine is an abstract concept [never actually built], a formalization of a mechanical process. But the process was well defined by Turing. Again, computers that have since been constructed we are using in ways that cannot merely be described as algorithms because the computer systems are engaging, interconnecting.
Suzan Mazur: Will you be returning to Poland anytime soon?
Slawek Nasuto: Never say never. But the Brain and Embodiment Laboratory here at the University of Reading is almost fully functional. It took two or three years to convince the university to give us the lab and I’m very much looking forward to some fascinating long-term projects.
Suzan Mazur is the author of The Altenberg 16: An Exposé of the Evolution Industry.Her reports have appeared in the Financial Times, The Economist, Forbes, Newsday, Philadelphia Inquirer, Archaeology, Connoisseur, Omni and others, as well as on PBS, CBC and MBC. She has been a guest on McLaughlin, Charlie Rose and various Fox Television News programs. Email: email@example.com