Kanzi
Go here for my ideas on the origin of human language.
The Ape at the Brink of the Human Mind
by E. Sue Savage-Rumbaugh and Roger Lewin
What is it that makes a man different from a chimp?
This book explores the similarities between
the language learning abilities of common chimps, bonobos, and brain damaged humans.
Kanzi is a bonobo. Will we get a chance to know this human relative before we make it extinct?
Just how closely related are chimps and humans?
Bonobo
Get to know the closest living relative of the human species
(we've killed all the other vocal apes, will the poor bonobo be next?)
Bonobos are more vocal than common chimps.
Their social structure is more oriented around the use of
sexual behavior for maintaining day-to-day social
contacts than is the case for common chimps.
The
Symbolic Species
Can we map the genes which make the human brain unique?
The answer is yes and Deacon explains the first steps being taken to do so.
The middle part of this book, which deals with the first steps that are being taken towards a science of mapping how specific types of brain activity produce human language, contains the solid core of Deacon's work. If you read just the 150 pages of Part 2: The Brain, you will recognize the vast amount of experimental work that still needs to be done.
A short review by Greg Nixon.
Extensive and systematic review by William Gale.
A critical review by Yehouda Harpaz.
That context, of the primitive nature of our knowledge of how human brains produce human language, allows Deacon to ridicule the simplistic theories of language which have been so popular historically in diciplines like linguistics (See Part 1 of the book). It is strange then to watch Deacon turn in Part 3 of his book to the task of introducing his own simplistic theory of the origin of human language. Deacon suggests that human language is required to make possible human male/female cooperation for child rearing.
If you have the stomach to keep working through Part 3 to the bitter end, you get to see Deacon let his philosophical stream of consciousness run free. On page 437, after so many pages celebrating the capacity of the human brain to produce human language, Deacon observes that the same human brain functions that make language possible also produce "what is most pathological in human behaviors". His line of reasoning is as follows: humans have a natural fear of death which comes to be cloaked in the "twisted justifications" of mystical and religious belief systems. Since these arbitrary belief systems "are subject to being rendered meaningless by contradiction", the history of human existence is "is sadly written in the blood the irreconcilable symbol systems have spilt between them". Where does this argument leave us? Humans kill each other because they fear death? I wonder if Deacon's Judeo-Christian childhood is in control of his philosophy. Many human societies see death as a natural part of existence, not some hell-fire source of fear. If I try to put "pathological human behaviors" into a millions of years long history of human evolution I cannot escape the quiet voice of E. O. Wilson reminding us that we are a species built on a sociobiology which allowed our ancestors to compete (often with other humans) for limited resources. For me, it is only our higher brain functions and language abilities that allow us as a species to craft memetic systems that can unite disparate tribes within a global social system based on material wealth and the pursuit of human happiness. Is the glass half empty or half full?
Deacon has thus entered deeply into the subject of memetics and the new science that seeks to map the dynamics of cultural inheritance. Maybe this will be the topic of his next book?
On page 442 Deacon states the anti-dualism sentiments that almost anyone who actually learns some brain biology comes to share. The popular Western dicotomies of mind/brain, intentional/mechanical, and human/animal are clearly antiquated misconceptions. The philosopher Dan Dennett has spent years and written at length in a vain attempt to enlighten his fellow philosophers about this point. One must wonder who Deacon is trying to reach as he covers this same ground. People who know no biology will stop reading after being attacked in Part 1, before they even get to Part 2 which deals with the biology that they really should know about. Any philosopher who can slog through Part 2 will already be aware of the work of people like Dennett, and have little to gain from Deacon's coverage of issues in the philosophy of mind. I agree with 95% of what Deacon has to say, but I wonder about how effective his presentation can be for those who do not already agree with him. How to get philosphers of mind to move closer to the biology of brains is an issue in memetics that I would like to know the answer to, I just fear that Deacon has not found the answer.
Deacon is weakest in his coverage of the issue of man-made intelligence. In 1979 Douglas Hofstadter's book, Godel, Escher, Bach, presented a 700+ page account of how brains use symbols and how artificial intelligence researchers are attempting to learn how to construct artifacts that will do the same. As the title of Deacon's book implies, he centers his investigation of human language on the capacity of the human brain to deal with symbolic representation. It seems that Deacon has never read Hofstadter's book, although he is clearly working along similar lines of thought. Deacon has clearly been influenced by the results of attempts to teach non-human primates like Kanzi to use human language: he spends a large portion of Part 1 of his book describing how hard it is for primates to learn language unless they make the "cognitive leap" into the mental world of using symbolic categories. This is exactly the same issue that Hofstadter so brilliantly placed at the center of AI research back in 1979.
Deacon provides half of a useful refutation of John Searle's famous "Chinese Room" thought experiment. This thought experiment purports to show that human mental processes cannot be decomposed into mindless algorithmic processes in an attempt to prove Searle's contention that a human mind cannot be a form of computer program. The response of Dennett and Hofstadter to the "Chinese Room" thought experiment is given in their book The Mind's I. I have had a little fun with this conflict between Dennett and Searle here.
Deacon's path towards refutation of the "Chinese Room" thought experiment is based on the distinction that can be drawn (and Deacon does so at length in Part 1 of his book) between "indexical relationships" and "symbolic relationships". [This is a good place to note that Deacon could not be bothered to provide a glossary of terms for his book.] This jargonistic distiction is concerned with the important issue of semantics, which Deacon has attempted to deal with in terms of his problem of "symbolic representation". In plain English, how do words have meaning for people?
The problem of "meaning" is at the core of the ancient philosophical discipline of epistemology, artificial intelligence research, and modern neuroscience. For biologists, the science of meaning is explored in terms of the on-going attempt to identify the mechanisms of memory that are used by biological brains to produce internal models of the external environment. In his book, Deacon has relied on the terms icon, index and symbol as used by philosophers like Charles Pierce in an attempt to categorize the various ways in which links are forged between lingustic tokens and the 'real world" (ignore the implicit dualism; many philosophers have imagined that mental representations are not real. Are they Ethereal?) physical objects they represent. To paraphrase Deacon (page 70), "icons have obvious similarity to the physical objects they represent." Think of the printer icon on your computer desk top. (If you do not have one, here is mine: 
The icon looks like a real printer. Some words sound like what they represent. Boom! Ha! An animal using an iconic token need only be able to recognize obvious structural similarities between the token and the "real world" object it represents.
When Deacon uses the term "index" we should think about animal conditioning experiments in which the animals are "mindlessly" trained to associate "real world" objects with arbitrary tokens of a rudimentary communication system. A mouse or chimp can learn to press a particular button on a computer input device when it sees a red light come on, but that button will not automatically function in the animal's mind as a meaningful symbolic representation of the light.
Some philosophers have thus wanted to use the term "symbol" only when a linguistic token is integrated seemlessly into a person's semantic understanding of the "real world" object which by convention we associate with the token. By this specialized definition of "symbol", animals can use tokens for communication without those tokens being associated with any human-like semantic content in their brains.
According to Deacon, in the "Chinese Room" thought experiment, a person who "mindlessly" produces Chinese by means of an algorithm could be forced (artificially) to accomplish the task by using indexical communication with no possibility of layering onto that meaningless manipulation of "indecies" true "symbols" ("symbol" is used here in the restricted sense defined above). In other words, Searle's thought experiment includes the claim that a mindless indexical algorithm could produce human language. This claim has been disputed by people like Dennett, but Deacon swollows it hook, line, and sinker. As Dennett has previously pointed out, once you accept this very questionable claim (that something as complex as human language could be produced by a mindless algorithm) then it is easy to fall into the trap of assuming that ANY algorithm for language would be mindless and unconscious, which is what Searle claims to have prooved with his "Chinese Room" thought experiment.
Deacon seems to agree with Searle, "No set of preprogrammed algorithms can smuggle symbolic reference in Searle's Chinese Room." Thus, Deacon seems to be in Searle's camp and he must imagine that there is something beyond algorithms that can produce language. Several pages later Deacon tells us what he has in mind (page 455). In talking about how to make a computer that would actually be able to use language tokens that would be linked to human brain-like semantic content Deacon says, "The key to this trick is....in the flow of patterns." This kind of jargon has become popular of late among philosophers like Searle who have divorced themselves from the idea that a 'mere' computer algorithm could produce mind. They are sure that something else is needed, some miraculous "pattern". They fail to understand that any pattern can be produced by computer algorithms.
Fortunately, Deacon actually has a bit more to say about how to make artificial intelligence than just the buzz word "pattern". He correclty points out that an intelligent machine should be a robot that can interact with and adapt to a complex environment. He then turns to the popular idea that a mind must be built from some type of "Darwin Machine" that can select particularly useful brain circuits out of a diverse pool of "bloomin', buzzing confusion" that contains many less useful circuits. Although such selectionistic processes are all the rage currently in mind/brain philosophy (See the work of Edelman, Harth, and Calvin for examples), there is also an important role for instructionistic mechanisms in the brain. As an example of how the behavior of neurons can be rapidly altered (instructed) in response to sensory input related to a complex environment, concider the hippocampal place cells. Within a short time of placing an animal in a new cage, these cells alter their patterns of activity to provide an internal representation of the animal's position in the new envoroment. This is not simply selecting among a pool of candidates for the best pre-existing cell for the job.
Presumably, if a "Chinese algorithm" were found (and make no mistake, this would be a huge task, one that has resisted the efforts of 50 years of AI research), it would be a relatively trivial task to layer on a an additional routine to get from the level of meaningless index manipulation to a related level of meaningful interpretation of the Chinese in terms of a true symbolic (semantic) representation. All we need to do is build our true artificial intelligences would be to include adaptive brains that can learn language from a complex social environment. Nothing to it! Deacon suggests (page 460) that a mindless trial and error research program in the field of artificial intelligence research should be able to replicate biological evolution's discovery of how to make a machine that can use human language. My bet is that the problem is complex enough that it will not be solved until we figure out more of the details of how biological brains do it and apply that knowledge to our robots. I'm not satisfied with neurophilosophers who bandy around buzz phrases like "Darwin machine" as if that solves the problem. There is a lot of hard work yet to be done within the new science of complex adaptive systems and also within the study of brain biology.
Deacon introduces an interesting analogy between the age-old idea of an immortal soul and our modern view of the brain as a mechanism for constructing a "virtual reality" model of the world inside our brains. Deacon rather poetically talks about the propagation of human "selfs" by human social processes that allow our "virtual natures" to be skimmed out of one brain and transmitted to other brains. Deacon does this in order to show how to replace mistaken human intuitions about the apparently dualistic nature of mind vs. brain with a rational understanding of how information flows between individual human brains during social behaviors.
Why human societies universally contain the idea of a soul is an important issue that is just a little bit larger than Deacon paints it. The fact that humans evolved under circumstances that allowed them to be subjectively aware of their own conscious thoughts without having any awareness of the possibility that such wonderous things can be created by material brain processes, allowed people to create and use the memetic construct "soul". People also used to say "sunrise" and actually think that the Sun was moving, not the Earth. It will be interesting to see how quickly people can shift from belief in immortal souls to an understanding of minds as precious possessions that only last as long as our material brains.
An interesting issue that Deacon raises is the idea that the evolution of human-style intelligence has not been just a matter of figuring out how to make a human-style brain, but possibly also a matter of over-coming certain active defense barriers to our human style of existence. I agree that it is dangerous to give large brained animals too much memetic control over their behavior. The tried and true method of closely controling behavior by genes with little room for learning by individuals or social groups is a well worked-out survival strategy. It remains to be seen if the human strategy of giving control over from genes to memes will work in the long run. As to the question of why only humans have this strategy, I suspect that an important part of the answer is that humans have systematically killed off all other meme-using primates.
Another philosophical issue raised by Deacon in the closing pages of his book is Free Will. Deacon seems to accept the conventional platitude that since we cannot predict human brain activity and human behavior, we have Free Will. A more complete coverage of this issue can be found in Dennett's book, Elbow Room.
I enjoyed Deacon's comments on the possibility of an objective "test" for consciousness (page 462). When I was growing up, I often wondered how the sensors of the Star Ship Enterprise could detect "life" and "intelligent life". I now know that life and mind are due to complex molecular events. A "scan" for the right type of complex processes could be used to detect living cells and active brains. What about a consciousness detector? Conscious brain processes drift on an ocean of unconscious brain activity. It would be tricky to spot just those molecular and cellular processes that correspond to consciousness. This is the difficult task that people like Crick have taken on, finding a neural correlate of consciousness. I am sure that we will find the neural mechanisms of consciousness that are used by human brains, but they will be rather subtle little tricks of memory that are used by certain cells in the brain along with other the other tricks that produce our unconscious thought processes. I doubt if there will be a single rule for easily detecting any possible mechanism that results in consciousness in Romulans or robots. The conscious machines we construct in the future will use different mechanical memory mechanisms than do human brains. The brain scanner we might one day use to check on human consciousness will fail to work on our robots.
Finally, Deacon has a few comments on the ethical implications of the growing human understanding of minds. Many philosophers panic at the idea that we are "mere machines" or that computers might be made to behave with human-like intelligence. Deacon points out that we have a choice: we can either begin (or in many cases, continue) to treat people like "mere machines" or we can look forward to one day treating conscious computers like people.
I agree that we are better off trying to learn how to control the mechanisms of our minds than remaining in our current state of ignorance about how our brains make minds. I want to know how to make intelligent robots. I want a future for people in which intelligent machines can work for us just as stupid machines do now. We must even contemplate our possible transhumanistic futures. My only reservation is that I fear that the social reactions against people like Galileo and Darwin will look like friendly banter compared to the reactions that will come against manipulation of human minds and civil rights movements for conscious robots. There is a lot of memetic engineering that will have to be done in order to integrate the science of mind into human society.
As a biologist who came to the study of the human brain from the perspective of wanting to know how to make intelligent robots that could talk, Deacon's book is a real pleasure to read. Unfortunately, it only reinforces my conviction that a frontal assault on a topic as complex as how the human brain makes human language possible is still premature. We have a few decades of grunt work left to perform in the neurobiology lab in order to reveal the molecular and cellular basis of learning and memory. We will then be positioned to apply that fundamental knowledge to specific issues in human behavior such as language use. Eventually, our understanding of how biological brains do it will allow us to construct artifacts that will make good dinner conversation.
Go to John's Book Page.
Go to John's
Home Page.
Magazine Description Natural History magazine translates the natural sciences into fascinating reading and spectacular color photography. From anthropology to zoology, the inner man to outer space, Natural History's articles are engrossing and easy to understand. The magazine is a unique combination of visual beauty and scientific accuracy that explores, defines, and depicts. Each issue is an adventure you won't want to miss.
