How the Body Shapes the Way We Think: A New View of Intelligence (A Bradford Book) (英語) ハードカバー – 2006/10/27
Kindle 端末は必要ありません。無料 Kindle アプリのいずれかをダウンロードすると、スマートフォン、タブレットPCで Kindle 本をお読みいただけます。
An exploration of embodied intelligence and its implications points toward a theory of intelligence in general; with case studies of intelligent systems in ubiquitous computing, business and management, human memory, and robotics.How could the body influence our thinking when it seems obvious that the brain controls the body? In How the Body Shapes the Way We Think, Rolf Pfeifer and Josh Bongard demonstrate that thought is not independent of the body but is tightly constrained, and at the same time enabled, by it. They argue that the kinds of thoughts we are capable of have their foundation in our embodiment―in our morphology and the material properties of our bodies.This crucial notion of embodiment underlies fundamental changes in the field of artificial intelligence over the past two decades, and Pfeifer and Bongard use the basic methodology of artificial intelligence―"understanding by building"―to describe their insights. If we understand how to design and build intelligent systems, they reason, we will better understand intelligence in general. In accessible, nontechnical language, and using many examples, they introduce the basic concepts by building on recent developments in robotics, biology, neuroscience, and psychology to outline a possible theory of intelligence. They illustrate applications of such a theory in ubiquitous computing, business and management, and the psychology of human memory. Embodied intelligence, as described by Pfeifer and Bongard, has important implications for our understanding of both natural and artificial intelligence.
" In this thoroughly engaging and unusually wide-ranging book, Pfeifer and Bongard make the case for the central role of embodiment in understanding natural intelligence and building artificial intelligence. The body and nervous system are inseparable interacting constituents of an organism, and it is a mistake to think of the former passively obeying the commands of the latter: they operate in complex and subtle harmony. With great clarity and authority, the authors – both leading researchers in the area – map out the intellectual landscape, from biological intelligence to robotics to intelligent companies. This is an outstanding and very accessible book: without being overburdened with technical detail, the reader is taken deep into this fascinating and important subject." --Phil Husbands, Professor of Artificial Intelligence, University of Sussex
& quot; In this thoroughly engaging and unusually wide-ranging book, Pfeifer and Bongard make the case for the central role of embodiment in understanding natural intelligence and building artificial intelligence. The body and nervous system are inseparable interacting constituents of an organism, and it is a mistake to think of the former passively obeying the commands of the latter: they operate in complex and subtle harmony. With great clarity and authority, the authors & ndash; both leading researchers in the area & ndash; map out the intellectual landscape, from biological intelligence to robotics to intelligent companies. This is an outstanding and very accessible book: without being overburdened with technical detail, the reader is taken deep into this fascinating and important subject.& quot; -- Phil Husbands, Professor of Artificial Intelligence, University of Sussex
"In this thoroughly engaging and unusually wide-ranging book, Pfeifer and Bongard make the case for the central role of embodiment in understanding natural intelligence and building artificial intelligence. The body and nervous system are inseparable interacting constituents of an organism, and it is a mistake to think of the former passively obeying the commands of the latter: they operate in complex and subtle harmony. With great clarity and authority, the authors - both leading researchers in the area - map out the intellectual landscape, from biological intelligence to robotics to intelligent companies. This is an outstanding and very accessible book: without being overburdened with technical detail, the reader is taken deep into this fascinating and important subject."--Phil Husbands, Professor of Artificial Intelligence, University of Sussex
After reading the book I think I accomplished my intent and more. I think this book is an incredible learning resource for readers who want to learn more about artificial intelligence in general. The book simplifies many complex, fragmented advancements in the artificial intelligence community into simple categories and schemas - through a `theory of intelligence' which contains components like redundancy, sensory-motor coordination, ecological balance, parallel/loosely coupled processes, value, etc. As you read this book, Pfiefer and Bongard repeat these conceptual themes through solid examples in robotics (e.g. M-tran, Slimebot, passive dynamic walker, Denise, Eyebot, Stumpy, etc.), which gave me a profound respect for robotics researchers and the challenges that they face in seemingly very simple design problems (e.g. like walking or running). After reading the book, I looked at the world and my own nervous system differently; the theory is defined in the scope of not only robotics (Parts 1 & 2) but also through case studies and examples in the real world in some way (e.g. ubiquitous computing, design principles, cyborgs, entrepreneurship, and human memory). So in a way Pfeifer and Bongard did deliver on their promise "we will deliver on our promise to explore how the theory of embodiment...changes the way we view ourselves and the world around us." (Page 245).
The book begins with a foreword by Rodney Brooks stating "this book is a gentle assault on some of the collateral tenets of modern rationalism; not an assault on rationalism itself, but an assault on many of the things that are commonly assumed by rationalists." Brooks goes on to describe that the embodied approach laid out by Pfeifer and Bongard may help to "provide an alternative framework within which further research can be carried out, and not incidentally, within which practical robotic artifacts can be built." This perspective is similarly echoed in the preface when Pfeifer notes that the book is about the synthetic methodology: "By learning artificial systems we can learn about biology, but also about intelligence in general..." Therefore, the introduction and preface clearly indicate that the purpose of the book is to inspire others to design intelligent systems from the theory of "embodied intelligence" laid out in the book, with intelligence being defined in a purposefully general manner.
After the preface, Part 1 (Chapters 1 and 2) introduces the philosophical problems in defining intelligence through providing definitions of terms related to intelligence (specifically thinking, agent, robot, artificial intelligence, models, and embodiment). The book then frames these terms within the scope of different perspectives of modern robotics research (classical, embodied, neuroscientific, diversified, biorobotics, developmental robotics, ubiquitous computing, multiagent systems, and evolutionary robotics perspectives). In these chapters, Pfeifer and Bongard argue that it is best to view intelligence in lieu of its historical development within the artificial intelligence community and with an open mind through the lens of many different research perspectives. It was this part of the book that captured my attention, making me realize the significance of robotics research.
Part 2 (Chapters 3-7) then builds a conceptual outline for an embodied theory of intelligence. Chapter 3 focuses on the components that are necessary to build a theory and provides examples in robotics to define each component (compactness, ability to use theory for design, diversity compliance, stability-flexibility, exploration-exploitation, time perspectives, and emergence). Chapter 4 focuses on how to design intelligent systems through defining properties of complete agents (three constituents principle, complete agent principle, cheap design principle, redundancy principle, sensory-motor coordination principle, ecological balance principle, parallel loosely coupled processes principle, and value principle). Chapters 5, 6, and 7 introduce design principles for developmental robotics, evolutionary robotics, and collective systems respectively. Together, this part of the book places the advances in robotics into conceptual categories which could possibly be used to build embodied, (single or multi-agent) intelligent robots. This part was a bit dry and hard to understand, so I would skip reading this section if you are not as interested in the theoretical part of embodied intelligence and instead look at all the interesting figures and robotics examples. Despite being dry, Pfeifer and Bongard provide many rich examples to keep your attention (though I felt at times too many examples were used).
Part 3 (Chapters 8-11) subsequently frames the theory of intelligence outlined in Part 2 into real-life examples. Chapter 8 focuses on the topic of ubiquitous computing, or the idea where the primary goal is to empower users of technology by embedding sensors into the environment (scaffolding) or by allowing sensors that can both sense and act (embedded systems). Chapter 9 focuses using intelligent design principles to build and manage start-up companies - Pfeifer and Bongard show that much of the uncertainty in creating a company can be reduced in understanding basic design principles. Chapter 10 focuses on how design principles can help clarify the definition of human memory, which was a little bit hard for me to understand. And chapter 11 focuses on robotic technology in everyday life, citing examples like vacuum cleaners (e.g. Roomba), toys (e.g. Furby), medical robots (e.g. physiological data), and humanoid companion robots (e.g. WF-4). Through these chapters I came to appreciate the breadth of application that could result from a theory of intelligence. After reading this part I literally viewed the world differently. I saw how the theory of intelligence outlined in the book could apply in my everyday life because I was already involved in a start-up company, unconsciously using the design principles outlined in the book.
When viewed in total, the book is written in a style not too technical but technical enough to gain a conceptual understanding of the various perspectives relating to "intelligence." I can say having an engineering background helped me to understand many of the complex ideas in the book (e.g. dynamical systems). There are many figures and examples to guide you through complex sections, which made reading the book more interactive. And the first 3 sections (simply put: history, theory, applications) seemed to flow well together when put into a conclusion (Part IV), which summarized the previous sections.
Beyond reading, I found that attending robotics lectures while reading the book aided in me understanding the topics discussed in the book. Go out to a university and just hear a visiting scholar talk about robotics research and you will get more out of the book (I heard an applied physiology professor speak about walking and running and designing robotic systems around walking/running). I also was simultaneously enrolled in the Biologically-inspired design course taught by Dr. Yen/Dr. Khan at Georgia Tech, so I found that being in a class related to the book helped me in understanding the book's concepts.
The book also has a website to connect the themes of the book to other books and forms of media (www.ifi.unizh.ch/groups/ailab/HowTheBody). Particularly, I found the lectures on this site as a form of simulated virtual discussions to answer questions in the back of my mind (particularly the ShanghaiAI lectures), the online figures seemed to simplify complex parts of the book, and the youtube videos provided concrete examples beyond just figures - it helped me to take philosophical concepts of `intelligence' outlined in the book in terms of robotics and see how they are applied in the real world of research (e.g. a dead trout can move down a streamline because of the way it is structured, bicycle stability video, passive robot walking video, insect locomotion video, etc.).
In short, this book teaches you how to take philosophical concepts of intelligence and apply them in the real world. This book inspired me to read more about embodied intelligence, attend talks by robotics researchers, take a class in bio-inspired design, and look at the world differently. It re-scoped my view of intelligence from the brain as a central controller of the body to the body also influencing the way the brain is structured. I'd recommend this book to anyone who would like to learn more about robotics (or for teachers who teach robotics courses to use this in the classroom).
Perhaps the demand for real time operation and the simultaneous need to control computational complexity result in the need for highly parallel inputs and outputs. These many input and output devices, however they might be configured, would then constitute a "body." (They could be distributed across space in a way the human body can not be. This would constitute a superiority for AIs.)