Computers and the Humanities, Vol. 9, pp. 197-201. PERGAMON PRESS, 1975. Printed in the U.S.A.
Reviews The Metaphorical Brain: An Introduction to Cybernetics as Artificial Intelligence and Brain Theory, Michael A. Arbib. New York: Wiley-Interscience, a Division of John Wiley & Sons, Inc., 1972. Pp. 243. $14.95.
operates in metaphorical fashion: A is like B; therefore I can respond to B as I did to A. The suggestion is a marvelous one to consider in the study of literature, systems of learning, and mental health and development, in addition to the more specifically designated areas of neural circuitry and control circuitry for robots. The writing is clear, controlled, urbane, and only rarely irritating to the non-scientist in its use of initials and coded references to various sections of the book, but even this is readily forgiven as the scope of the book takes shape and becomes a working idea. The assumptions we have made about our brain and crucial functions we have not thought about at all are called into conscious immediacy; this alone is worth the price of the book, but there is much more there.
By Doris Adler Perhaps it is cheating, but I enjoyed this book so much more after reading the final chapter that I would recommend reading it first. The Metaphorical Brain is a fascinating, provocative study of "brain theory" written for both the layperson and the scholar. The stated claim that it is "accessible to anyone who reads Scientific American" is justified, and it certainly seems that the theory, the critical apparatus, and the impressive bibliography would "merit the attention . . . of experts and serious students of cybernetics, artificial intelligence, or neuropsychology." Beginning with the two metaphors, "Humans are machines" and "Humans are animals," the book compares the functions of the human brain to those of computers and animals and evolves an intriguing new set of principles for "a fruitful way to go about modelling brain function." Just this is the main value of the work: it does not propose to be the model, or the set of principles, but it opens new areas of consideration, new possibilities, new ways of viewing our mysterious interior control panel. Strong challenge is made to the theory of the constant-recording camera of our minds, and it is suggested instead that the human brain is a highly evolved animal brain that selects and records what is needed for response and builds interior models of the world based on an accumulation of previous input. Individual brains processing the stimuli of external reality by uniform mechanical means produce individual internal models of reality. To oversimplify almost beyond meaning, the brain Doris Adler is the chairperson o f humanities at Howard University, Washington, D.C.
Principles and Practice of Textual Analysis, Vinton A. Dearing. Berkeley, Los Angeles, London: University of California Press, 1974. Pp. xi + 243. $13.50. The Art of Literary Research, Richard D. Altick. Rev. ed. New York. W.W. Norton & Co., 1975. Pp. xv + 304. $7.50. By R.L. Widmann Dearing's volume is a clarifying and rethinking of many of the problems explored in his Manual of Textual Analysis (1956). The preface to Principles promises to establish the distinction "between the genealogy of the ideas or complexes of ideas that these physical objects transmit." Dearing also says that he "provides much more detailed instructions as to how to perform textual analysis with computers and without, and for the first time formulates the axioms of textual analysis and R.L. Widmann, a member o f the English department at the University o f Colorado (Boulder), is book review editor o f CHum. Her research includes textual analysis and bibliographical criticism in Renaissance literature.
197
19 8
REVIEWS
demonstrates their inevitability" (pp. ix-x). The book itself, however, really concentrates primarily on defining textual analysis, as a field distinct from bibliography. Dearing's use of computers lies behind much of what is in this book, and he uses the results of many of his computer studies for the examples and illustrations in the body of the text. But the reader will not find the kind of guidance necessary in computer-assisted work beyond fairly rudimentary kinds of advice and comments. Because he does not seem to have defined his audience carefully, Dearing varies from very simplified to extremely complex analysis, particularly when he is discussing his theories of the calculus of variants. Dearing begins by likening textual analysis to information theory, but the analogy does not hold throughout the volume. He uses language and concepts from computers in outlining the theory of textual analysis, particularly in subsection 2 of chapter 2, in his comments on "Deciding on the variations." He mentions on page 67 that computers can be used to break rings (the establishment of which is illustrated on pp. 59ff); his example is that 10MSS. and ten printed editions of Dryden's Heroick Stanzas of 114 lines provide about 400 rings. He also points out the possible usefulness of computers in restoration of lacunae and cites Paul Tasman's work (see the footnote on page 83). In addition to such sporadic references to actual computer applications by other scholars, Dearing concentrates his discussion of computer use in pp. 152-54, where the commentary is closely linked to computer collation advice. An appendix B also describes the processes of a variety of programs which Dearing has deposited in the UCLA library and which are available upon request. The programs include ones for probabilistic studies (chapter four in this book) for determining the preliminary diagrams for textual and bibliographical trees, and for locating the archetype in a preliminary diagram. All three discussions would probably baffle the novice and be of somewhat limited interest to the person already making computer applications in his or her own research. This is not to say that Dearing's work is not valuable, for it certainly is. It would probably have been helpful to have more of the programs themselves printed and less of the somewhat prolix discussion. This book has not really established its audience. The clarity and sharpness of Dearing's article "Computer Aids to Editing the Text of Dryden," (pp. 254-78 in Art and Error: Modern Textual
Editing, ed. Ronald Gottesman and Scott Bennett, Bloomington: Indiana University Press, 1970) does not inform this book, Principles. It is a quirky book, needing careful reading and attention. This reviewer is bothered by the wild variations in style and content from the simple to the complex, paragraph by paragraph at times. Beginners are advised to start elsewhere. One such leaping-off place is in Altick's revised edition of The Art o f Literary Research. In a workable chapter five, "Finding Materials," pp. 142-46 briefly discuss and evaluate possible uses of computers in literary studies and guide the beginner to published works which have used computers and to possible areas for further research. The Altick volume, of course, is not attempting to lay the basis for a theory of textual analysis, but as a handbook, it is much more serviceable than Dearing's volume. Argentina and the Failure of Democracy: Conflict among Political Elites 1904-1955, Peter H. Smith. Madison: The University of Wisconsin Press, 1974, three appendixes, bibliography, and index. Pp. 215. $12.50. By Alberto Cisneros-Lavaller This book is the most important contribution to the study of behavior and electoral leadership in Argentina. Concerned with the formation, duration and decay of democratic-rule, and looking for knowledge about the sources of political weakness in the country, Argentina and the Failure of Democracy stresses a non-deterministic explanation of changes. Peter Smith has successfully detected and quantitatively analyzed behavioral trends in the Argentine Congress over time, as a "microcosmic representation of effectively competing forces in the national community, as an arena for the articulation of policy alternatives, and as a stepping-stone for men who went on to make executive decisions" (p. xix). Focusing on the sequences and rates of change, drawing from Deputies' social background and careers, discussion issues, sources of conflict, and types of alignment, he has handled role call data in an impeccabJe way. The same can be said about the analysis of social structure through which he has A l b e r t o Cisneros-Lavaller is a graduate s t u d e n t in the Institute o f Latin A m e r i c a n Studies a t the University o f N o r t h Carolina, Chapel Hill.
REVIEWS
199
identified "aristocratic elites" in the Congress. With Le Malpas Rockshelter: A Study o f Late Paleolithic the aid of factor analysis, he has also quantitatively Technology in its Environmental Setting, Anta measured changes in Argentina during the first half Montet-White, with chapters by Marie-Madeleine of the century. That fifty-year span permitted a Paquereau, Richard A. Rogers, and Jean-Pierre greater longitudinal analysis, enhancing the historical Texier. University of Kansas Publications in Anthropology, 4. Lawrence: University of Kansas Press, dimension as well. The description with which he introduces us to 1973. Pp. vii + 135, figures, graphs, map, tables, "Society and Politics" in Argentina, although fairly references cited. $3.50 (paper). good, requires some qualifications. Even though in practice the federal intervention in the provinces By William 11. Marquardt takes place, as Smith states, "whenever the federal government seemed necessary," constitutionally that This attractively printed and reasonably priced intervention requires certain specific conditions, e.g., monograph reports the results of archeological, a state of internal disorder. Furthermore, although sedimentological, and paleoecological investigations the Saenz Pefia Law explicitly limited political carried out from 1966 to 1970 at an Upper Paleorepresentation since by its statutes only two parties lithic site in the Couze Valley, Boumiquel, were represented in Congress, with two-thirds of the Department of Dordogne, France. Goals of the seats going to the winner and one-third to the first research were to reconstruct the sequence of late minority (that is why it is called "lista incom- Pleistocene microenvironments in the vicinity of the pleta"), that electoral law cannot be interpreted as site, to develop an informative general methodology not extending political rights to urban workers for the description, functional identification, and because it had explicitly extended the right o f comparison of lithic artifacts, and to use the voting to every citizen without any discrimination. environmental and artifactual information to conThe study of the dimensions of conflict (issues, struct a model of late Pleistocene human parties, etc.) as a product of pre-existent cleavages paleoecology. deserves our attention. Smith has made a detailed In addition to introductory information, the first analysis using a typology of Deputies that includes five chapters report the depositional sequence, the whole political spectrum from conservatives to excavations, sedimentological analysis (J.-P. Texier), socialists and the like. For the first time, patterns of palynological analysis (M.-M. Paquereau), and faunal party alignment as indicators of a high correlation analysis (R. Rogers with L. Martin). Drawing from between conflict and social strata are studied. the climatic and biological information, chapter 6 The explanatory power of his interpretation of provides a reconstruction of the late Wi~rm environpolitical change in Argentina is both convincing and ment obtained from computer simulations. Chapter accurate and perfectly demonstrated in the last part 7 contains analyses of the Solutrean industry from of the book. His explanation that a cyclical type of the western part of the site, chapter 8 describes the crisis causes sequential changes in participation, Perigordian industry from the eastern section, and legitimacy, and distribution allowed him to break chapter 9 furnishes a summary and some concluding down his study in three different periods: from comments. Doubtlessly reviewers of this work for 1904 till 1930 when crises of participation archeological and anthropological journals will critioccurred, the decade of the thirties when legitimacy cally examine its contribution to research in Upper was questioned, and finally the Peronist era when Paleolithic prehistory. Remarks here will be crises of distribution appeared. confined principally to the value and appropriateThis study has demonstrated that political crises ness of various computer-assisted analyses. appear as a result of rapid socioeconomic changes, Recording all field and laboratory data with that reactions to one crisis at one time can generate computerized data storage and retrieval in mind another crisis at another time, and that the outcome eliminated the intermediate step of special data of crisis can affect the system's ability to cope with coding, and assured the availability in the field of subsequent crises. Since this is a perfect way to adequate amounts of information. A strong point of understand Argentine politics, by all means this the research design is the technique of recording the book deserves our appreciative recognition and perimeter of lithic tools by means of a polar grid. should be highly recommended not only for "Argentinists," but also for the broad mass of social William H. Marquardt is an assistant professor of anthroscientists as well. pology at the University of Missouri-Columbia.
200
REVIEWS
Such a practice, in concert with more traditional observations, provides for attention to a wider range of morphological details and for more realistic definition of functional artifact types in terms of interacting attribute complexes. Shortcomings of the work can best be related in three aspects: context, quantitative analyses, and the reporting of results. These will be discussed in turn. First, it is clear from the technical (pp. 14-31) and illustrative information presented that most of the prehistoric remains at Le Malpas are not found in the rock shelter itself, but in a talus slope disturbed by numerous historic occupations. Not only is there evidence of internal movements of the upper (Solutrean) deposits caused by water drainage (p. 14), but it is clear that the lower (Perigordian) levels were deposited at a time when the environment of deposition is in doubt (pp. 30-31). A second weakness of the report lies in the insufficient justification given for certain methodological choices. For example, it is unclear why frequency counts are not displayed for the chisquare analyses (tables 21 and 22). Out of many alternatives, unweighted pair-group clustering was employed using Pearson's r as a similarity measure (pp. 82-87), but the reasons for these choices of technique and coefficient are not explained. Finally, grave doubts must be expressed concerning the mixing of nominal and ratio scale data in the burin assemblage correlation analysis (p. 98). The third and perhaps most serious flaw in the work is the inadequate and inconsistent presentation of results. The report could have benefited considerably from more thorough proofreading. Incomplete sentences (pp. 97,98), inaccurate reporting of data (e.g., table 28), insufficient labeling of drawings (e.g., pp. 88,96,105,117), incongruity between terms used in text and in diagrams (e.g., table 12 and fig. 18, p. 116 and fig. 51, p. 119 and fig. 53), and inconsistency in the use of scientific and common names (e.g., pp. 42-46) are examples of errors which make difficult the understanding of substantive discussions. Those readers interested in replicating or further experimenting with analyses will be particularly disappointed. For example, Pearson correlation matrices (e.g., table 10, table 12) are displayed as "+" for a significant positive correlation, " - " for a significant negative one, and "0" for a non-significant one. The level of significance is not stated, nor arc raw data supplied for the reader's inspection. In the one instance in which both the correlation matrix and the corresponding
principal component analytic results are displayed (see p. 122), I could not obtain a comparable result. Neither raw data nor correlation matrices are given for plant and for sediment observations (see pp. 42-52), nor is it clear on what basis principal components analysis was carried out on the sedimentological data. In summary, despite a well-conceived research design, this ambitious monograph falls short (1) because the design was applied to a site in which context is questionable at best, and (2) because of the inconsistent and inadequate reporting of research results. Innovative research designs must always be judged with reference to the appropriateness of their application to the object site(s). All of us who depart from traditional analytic techniques owe it to our audience to provide lucid, internally consistent, well-illustrated discussions of the logic of our methodological choices and of the suitability of particular descriptive and analytic techniques. Doing less risks the unwarranted rejection of the best conceived, most carefully executed research, and seriously delays the progress of scientific archeology.
Illustrative Computer Programming for Libraries: Selected Examples for Information Specialists, Charles H. Davis. Westport, CT: Greenwood Press, 1974. Pp. xvi + 112. $7.95,
By Gerald Lundeen In what is essentially a workbook for applications programming in PL/I for a first course for library school students, the basic element is a series of ten practice problems illustrative of computer applications of interest to library operations and text processing situations. These are graded from easy (starting with merely reading in data and printing it with no further processing) to more advanced problems such as keyword indexing and weighted term search logic. Interspersed with the practice problems is a clear presentation of elements of PL/I and of programming concepts to enable the student to cope with the problems. Following each problem statement, solutions (for some, two or three variant solutions) are provided. These are then followed by an analysis of the solutions, pointing out the important elements and clarifying the logic Gerald Lundeen is an assistant professor o f library studies at the University o f Hawaii at Manoa, Honolulu, Hawaii.
REVIEWS
behind them. These analyses also give Davis a chance to make some points about programming efficiency. The book is divided into two main parts: "introductory examples" and "more advanced applications." Part one includes discussions of c o n t r o l cards, program documentation, the DECLARE statement, compilation vs. execution, input and output, some important built-in functions, PL/I logical operators, simple search, number and cost of acquisitions, and circulation tally. Part two includes SDI using Dewey and fiction codes, simple sorting, conventional keyword indexing, the binary search, keyword indexing using FLAGS, document retrieval by weighted-term searching, on-line interactive programming, and elements of auxiliary storage. The tone throughout is practical. Because information not essential to the problem at hand is kept out, the student is not burdened with an excess of material. The problems have been tried and tested in Professor Davis' programming course at the University of Michigan School of Library Science. The choice of these problems is excellent. They are simple enough for the beginner to understand without difficulty, when taken in sequence, and yet are meaningful in terms of the kind of applications the student will encounter in real life. At the same time the problems serve to introduce the major features of PL/I. For the most part the problems are not especially library oriented, but should be meaningful to anyone wanting to learn the essentials of PL/I programming for text processing, information retrieval, and similar applications, thus justify-
201
ing the subtitle. Those examples which do involve library situations specifically, not requiring any special knowledge of library science beyond that which most library users would have, should not be a barrier to wider use. Well illustrated with flowcharts, sample cards, and sample output, the text and the programs are in neat, easily read type, and appear to be free of errors or misprints. It might be useful, however, to point out one possible source of difficulty: the program provided as a solution for practice problem 10 (weighted term searching) was written for use with positive threshold values, so the approach to exclusive OR logic as discussed on page 85 will not work with it; if it is attempted, documents with weight of zero will be printed along with those wanted so the output will be the complement of the set of documents defined by logical AND (this difficulty can readily be fixed by changing the eleventh statement on page 89 to include a check that the threshold is not equal to zero). The major part of the text assumes batch operation, but following the last problem a brief discussion of one-line interactive programming is provided. Two brief bibliographies follow the text; the first, a list of references cited in the text; the second, a selected list of books on PL/I. The index is sufficiently detailed. This small book provides a clear, practical, easily assimilable introduction to the essentials of PL/I for those interested in text processing or other library and information science applications. It fills a real need which is not well met by the other PL/I texts available and would supplement these texts nicely.