A REVIEW OF SYSTEMS ENGINEERING MODELS OF HUMAN-MACHINE INTERACTION by William B. Rouse Publisher: North Holland (This book is Volume 6 of the "System Science and Engineering" series) Reviewed by: Kamesh Ramakrishna Address: Carnegie-Mellon University, Pittsburgh, Pa. 15213 Net-Address: Ramakrishna@ CMUA This review is in two parts: the first part more or less describes the author's goals and the organization of the book, while the second part presents the reviewer's opinions. The author had three goals for the book: "to emphasize the current state of the art", "to provide a treatment of a highly mathematical topic while avoiding calculus, differential equations, Laplace and Fourier transforms, and so on", and, "to include basic tutorials on the modeling methodologies of interest and thus avoid requiring the reader to consult other basic sources." The book is based on the lecture notes of a graduate course in Industrial Engineering offerred at the University of Illinois to mostly engineering and psychology students. The author feels that the material in the book is "nicely complemented by having students pursue a series of small design projects in which they have to choose among the various available models, resolve measurement problems, and so on." The book has seven chapters. The Introduction is a somewhat loose introduction to the purposes of modeling. The last chapter, entitled "Human-Machine Interaction in Retrospect" carries on where the first left off. Though there are references back into the body of the book, these two chapters are quite detached from the rest both in style and in content. The remaining chapters of the book are: "Estimation Theory" (Chapter 2), "Control Theory" (Chapter 3), "Queuing Theory" (Chapter 4), "Fuzzy Set Theory" (Chapter 5), and "Production Systems, Pattern Recognition, and Markov Chains" (Chapter 6). Each chapter contains some introductory text, some mathematical model (except for Chapter 6, which reads in part like a chapter on "other techniques"), an application of the model and a report on a study conducted in the style of the technique. The mathematics begins hard (in the estimation theory and control theory chapters) and eases off as one goes further into the book. This seems counterproductive for a book intended for students with a weak mathematical background -- it seems to me that it would have been better to begin easy and get .harder later. Also in this reviewer's experience, most graduate students in psychology are not prepared for the mind-numbing symbolism of the matrix algebra used in control and estimation theory (bold-face, upper case letters, functional notation, operator notation, carets, tildes, vertical bars, etc.). For that matter, the reviewer himself blanks out in the presence of more than two carets and a tilde, especially when it is not clear where the mathematics is headed. This happens occasionally in Chapter 2. In each chapter, some examples of applications and some studies are discussed. Work by people other than the author are dealt with somewhat perfunctorily, at best. The best-considered examples are of studies performed by the author and his collaborators. This is not entirely a bad thing, as it probably makes for clear exposition; however, it is definitely not in the spirit of the author's stated goals. As a result of this, the book neglects careful discussion of techniques that the author may not be familiar with. This is most apparent in the section on Production systems, in which the examples are somewhat trivial, pre-1970, and current state of the art techniques ignored. Since the reviewer is familiar with work in this area, he found it disappointing. Probably as a result of the above bias, there are lots of references to the author's own work -- 17 of 150 references list the author of the book as the primary author. The bibliography is reasonable; the bibliography covers a wide range of work of the last twenty years, though some classic works and results in the human factors field do not appear, such as by Welford, or Fitts, (and they are surely relevant, if only in the introduction or the conclusion, given the structure of the book). Some references appear that were not very relevant; though the reviewer could identify these where they occurred (the egregious ones being to Kuhn, Ziman, and Weizenbaum), the references and the local context appeared to have been created for the purposes of inserting the reference. The author does not discuss the relation between parameters to models developed in the book and concrete behaviours exhibited by subjects in the reported experiments sufficiently. This lack probably reflects on some differences in emphasis between the reviewer and the author on the uses of systems engineering models -- one of the reasons that the reviewer prefers production system models is the possibility of linking model parameters to observable components of the user's behaviour. For example, in the chapter on Fuzzy Sets, the author describes an experiment whose results could be modelled using a "psychological distance" measure. This is a useful trick measure; and the author informs us towards the close of the chapter that "... success in using models to provide succinct behavioural descriptions often depends upon the somewhat serendipitous discovery of useful intervening variables . . . . " Quite so. And one of the purposes of models is to clarify the relation between such intervening variables and the computational or other structures that the subject employs. Neither in this study (apparently), nor in this exposition of the study, was an attempt made made to identify such a relation. The publisher's blurb bills the book as a "reference book"; in the reviewer's opinion, the book fails short of satisfying his somewhat fuzzy criteria for a reference book. In summary, the author deals competently with three areas: control theory, estimation theory, and queueing theory, less well with fuzzy sets, and least effectively with production systems and related techniques. If the reviewer were teaching a course in the field, he would consider using the book for those areas. The examples in these three chapters are reasonable, even though they are not chosen from a large population. The introduction and concluding chapters are inadequate; they should probably have been a little longer, less philosophical, and more concrete. As explained by the author in the introduction, a course based on the book would have to be supplemented by design projects -- the examples in the book can be guides for designing such projects.

/lp/association-for-computing-machinery/review-of-systems-engineering-models-of-human-machine-interaction-by-A8LO3OtePs