Surfaces and Essences: Analogy as the Fuel and Fire of Thinking
Though in the introduction the authors tell the reader, “[t]he main goal of this book . . . is simply to give analogy its due,” Surfaces and Essences ranges wider in scope, delving into the complex relationship between words and meanings.
The reader is provided numerous examples of idiom, double meanings, subtle meanings, and differences in translation that prove a word’s meaning is not always captured by its dictionary definition.
It should be no surprise that our ability to use language expands and becomes more discriminating with education and experience. And on this particular point Surfaces and Essences exposes its seams and falls apart.
The authors ask, “Is it really plausible that the very same mechanisms underlie the act whereby a two-year-old spots a Saint Bernard and exclaims ‘sheep!’ and the act whereby a physicist of great genius discovers a subtle and revelatory mapping between two highly abstract situations?” This reviewer responds, well, why not?
The authors dig in and repeat themselves to the point where a reasonable person would simply stop and toss Surfaces and Essences out a window or feed the book to a shredder or woodstove. But I’m not a reasonable person, and I had to read this book—so you do not have to.
After many pages of non-illuminating repetition the authors move on to categorization. All examples of categorization basically collapse to the consideration of “Is an a ? For example, is Pluto a planet? Solutions for this question are a process: define , define , define the rules you use for distinguishing similarity and dissimilarity and after taking into account context and extraneous issues, make a judgment. In the example of Pluto, categorization comes down to how you define a planet and how Pluto applies to that definition; the context might include the year you ask the question and your personal stake in the answer.
The process of categorization can be clumsy. Most of us instinctively take shortcuts, making assumptions and looking for social cues, or choose by gut. If humans didn’t take shortcuts, social conversation would be, well, unsocial.
Categorization, all categorization is based on human judgment, and judgment often changes with greater knowledge. Common sense isn’t all that common. These are this reviewer’s views. The authors do not provide a useful definition of categorization nor of judgment.
The authors do, however, recognize that the classical interpretation of categories for the notion that categories have sharp boundaries doesn’t always hold. New categories are constantly created, old categories broken or extended, and boundaries continually sharpened or fuzzed. For example, science at one point considered all living things to be either plants or animals. Today scientists acknowledge that there are single cell organisms having both plant and animal characteristics.
The authors also point out that when things that are complex, we might gain advantage by not focusing on differences but on similarities. As differences are expressed through categories, similarities are expressed through analogies. Analogies are the stepping-stones to idiom, figures of speech, and fables. For example, prior to the second Gulf war the question was, would the consequences of the U.S. invading Iraq be similar to the consequences of Vietnam?
Taking a step back from that example, even a simple sentence can have complexity. Surfaces and Essences offers four meanings of an offer to “have a coffee:”
1. Having coffee
2. Having a coffee variety, perhaps a caramel macchiato
3. Offered coffee, but taking tea
4. Offered coffee, but having any different drink
Note the meaning changes with the intent of the speaker or listener. In not accounting for intent, the authors miss the possibility of yet another choice that “having a coffee” could be code. In this instance an offer to “go for a coffee” has hidden intent. For example if the speaker was male and listener female, “going for coffee” could be interpreted as an offer to date without the social awkwardness of actually having to say, “go on a date.” So, too, consider the phrase “hiking the Appalachian Trail”—governor-speak for leaving town to visit one’s mistress—without actually having to say it.
Surfaces and Essences offers pages and pages of examples where a talker uses a word in an abstract sense and a listener interprets in an overly literal manner. For many if not most of these examples the speaker or listener is a child. The reader soon recognizes that an educated adult has a much greater grasp of abstraction than a child but whether the authors ever grasp this remains unknown.
The bigger picture, the consequence in the real world of an inability to make and understand abstraction is sorely lacking, in particular with regards to using language to influence others. Using language to influence does show up in Surfaces but only in a piecemeal fashion. There are many examples of exaggeration, caricature, political rhetoric and advertising but again the big picture, the nature of rhetoric does not.
Additional disappointments come from missed opportunities. One is the authors’ shallow analysis of Google’s online language translator, Translate. Translate uses statistical methods for word translation. The significant failure of Translate is that while individual words will be translated correctly, sentences can still fail to make sense. The missed opportunity is the authors’ failure to analyze the significance of this failure.
For example, comparing Translate with Searle’s “Chinese Room,” a seemingly perfect scenario of a computer-like translator being indistinguishable from a human or evaluating Translate in the context of the Turing test, which is another scenario of a computer program being indistinguishable from a human. Or comparing Translate against IBM’s Watson, a computer that won the TV game Jeopardy.
Why are humans are so good at language while computers do so poorly. The authors are so stuck in word definitions and idiom/language translations they never transition to a “big picture”, a consideration of mind. The flexibility of the human mind allows any speech act or sign to signify anything. Consider if any utterance or sign can mean anything, then how can one figure out (by algorithm) anything (anything correct) at all? Yet humans do quite well, starting from a very early age (apart from abstraction and experience—hmm, there’s something interesting here . . .).
The ramifications of how we arrive at meaning have led many philosophers to different theories of mind and theories of communication. Douglas Hofstadter or Emmanuel Sander produce no theories, only examples. The authors’ claim that our advantage is due to our ability to use analogy and categorization seems shallow—perhaps even a tautology.
The authors also contradict their claim, admitting that analogies can be complex, easy to misuse, and when used in argument may be used both for and against the very same thing.
Surfaces and Essences provides many lengthy digressions, including an attack on styles and methods for teaching mathematics. The authors’ argument for this seems to be a straw man. Although examples of poor teaching methods are given, there is no naming of the schools that use poor methods, or naming of schoolbooks that contain poor methods, or naming of education standards bodies that recommend poor methods. Exactly who are the authors attacking?
The next to last chapter provides a tour of Einstein’s theories through the analogies Einstein used. However, the authors never take the conceptual leap to tie analogies to model making, and model making to formal model making - that is, models that follow the rules of mathematics. And this: analogy -> model -> formal model is much more key to Einstein’s use of analogy than simply his use of analogy.
The last chapter is an “Epidialog” similar to dialogs from Gödel, Escher, Bach but in this instance feels inconsequential, being an argument over which, analogy or category is more important. The authors appearing to never having made the cognitive leap that both are inevitably and irretrievably linked like light and shadow.
This reviewer’s admiration for the old Douglas Hofstadter, the author of Gödel, Escher, Bach, makes it difficult to be so critical for Surfaces and Essences in this review. How could such brilliant minds have become stuck in such ruts?
Surfaces and Essences gives off a whiff of obsession, similar to Stephen Wolfram’s A New Kind of Science. For Wolfram all of life’s patterns were reduced to automata, and Wolfram devoted 1000+ pages to automata. For Douglas Hofstadter and Emmanuel Sander, all of thought may be reduced to category and analogy—but not deeper than category and analogy.
One feels pity for brilliant minds gone astray. The one advantage of Surfaces and Essences has over A New Kind of Science is that Surfaces weighs in at a relatively light 594 pages. When compared to Wolfram’s book, Surfaces and Essences has two authors and half the numbers of pages, let’s say Douglas Hofstadter and Emmanuel Sander exhibit one-quarter of Wolfram’s obsession, so this reviewer feels exactly one-quarter the pity.
In summary Surfaces and Essences offers a broad and rambling surface with very little essence, no sense of balance, misses the forest for the trees, and commits the greatest crime of all—becomes a slog to read. A reader would be better served by selecting a single topic from the table of contents or index and use what’s found within as a touchstone for his or her own thoughts on cognition.