Book title: Understanding Computers and Cognition

Addison Wesley, 1988, ISBN 0-201-11297-3

Andrew Dillon

This item is not the definitive copy. Please use the following citation when referencing this material: Dillon, A. (1990) A Review of Winograd and Flores - Understanding Computers and Cognition, Hypermedia, 2(1), 71-74

Initial glances at a book with such a title might lead one to imagine that this is just another text on human-computer interaction (HCI) or a basic reader in cognitive science but both inferences would be false. Quotes on the cover from such notable reviewers as Joseph Weizenbaum and Howard Gardner testify to the book's "ground-breaking", and "deeply thought-out" qualities which render it a "very good and important work". The publisher's blurb makes numerous references to its "controversial" "new", and "broad-ranging" contents before summing it up with the statement that the book enables us to better understand "what it means to be human"! Who wouldn't be intrigued?

This book is sub-titled "A new foundation for design" and herein lie the clues to its contents and purpose. Winograd and Flores are interested in computers as life enrichers and liberators of humanity and encourage designers to "let our awareness of the potentials for transformation guide our actions in creating and applying technology" p179. However, such admirable goals are, according to the authors, unattainable within the design philosophy underlying contemporary technology which is presented here as narrow and rationalistic. What the authors attempt to provide is a clarification of the background of understanding in which the discourse about computers and technology takers place. In so doing they lead the reader on a mind expanding trip through hermeneutics, the biology of cognition and speech act theory (all this in the first sixty pages) before what we might euphemistically describe as "highlighting the shortcomings" of current work in AI and intelligent systems from where they put forward their alternative. As you will no doubt have noticed by now, this is no ordinary book.

The basic tenet of their argument is that computer science and psychology have got it wrong! The success of the hard sciences in identifying cause-effect relationships and building a body of knowledge about how the world works has led inappropriately to an acceptance of empiricism as the only way to further our knowledge of ourselves and our psychological world. Such an approach demands a view of humans as existing in, but distinct from, environments filled with objects on which we perform operations. In the opinion of the authors, this has resulted in generations of researchers and system developers taking for granted a false view of how we exist in and interact with our surroundings, with the end result that our computerised tools are a poor match for our needs. According to the dominant view, the human cognitive system is to be unquestioningly accepted as an information processing system capable of representing the world symbolically and intelligent behaviour is said to result from the correct manipulation of these representations in the mind. Not so, claim our intrepid explorers of uncharted intellectual territory, this is "the fallacy of instructive interaction" and in so saying dismiss thirty years of work in cognitive psychology.

Their alternative is based largely on the work of Heidegger in that it claims objective perception and detached analysis of the world are fundamentally impossible. There is no knowledge without involvement, "being-in-the-world" (Dasein, in Heidegger-speak) and "throwness" which they define as "the condition of understanding in which our actions find some resonance or effectiveness in the world" p33. In support they call on the work of Maturana and Searle, the former, heretofore unknown to me, being concerned with the biology of cognition and argues strongly in favour of "structural coupling" between organism and environment and concepts such as "autopoiesis" and "consensual domains", the explanations of which fall short of providing me with enlightenment. Searle is invoked in order to highlight the limitations of formal analyses of language that fail to consider the role of background information and the necessity of commitment to certain views in understanding the meaning of a statement.

The upshot of all this is the initially surprising proposal that "nothing exists except through language" and therefore:

"the domain in which people need to understand the operation of computers goes beyond the physical composition of their parts, into areas of structure and behaviour for which naive views of objects and properties are clearly inadequate. The 'things' that make up 'software', 'interfaces' and 'user interactions' are clear examples of entities whose existence and properties are generated in the language and commitment of those who build and discuss them" p69.

Attempts at building expert systems or natural language processors based on the dominant rationalistic model are doomed to failure according to the authors and they expend several chapters of effort describing and destroying some of the ludicrous claims made for computers and applications like ELIZA and SHRDLU. So far so good.

The final section of the book attempts to lay the foundations for a new design philosophy based on the earlier discussions. Unhappy with the perspective that views users of technology as rational decision makers searching a problem space considering alternatives, they present the user as "thrown" into a problem situation, possessing a set of "background assumptions" about the situation he is in and having "commitment" to the attainment of certain objectives. From this viewpoint, organisations are seen as "networks of commitments and recurrent conversations" in which "breakdowns" occur to which members of an organisation respond. The role of the tool is to support communication between members, it is a means not an end, and to be successful tools must be transparent.

These ideas are not difficult to grasp and indeed, some are not particularly original. What is required at this point is a description of a system embodying such ideas. What we get instead are references to, but few details of, something called "The Coordinator Workgroup Productivity System", a claim that the Apple Macintosh supports the "readiness-to-hand", "ontological simplicity" of their approach and some vague guidelines grouped misleadingly under the title "A Design Example". If you want to know how to do it, you won't find the answer here.

In fairness to Winograd and Flores they admit that they "offer no magic solutions, but an orientation that leads to asking significant questions" p.174. In this sense, the book is a runaway success. The authors adopt a position that some might see as scientific heresy and interestingly (and, to my mind, successfully) defend it against the imagined attacks of "rationalists". This is done in a manner and form that few will find too difficult to follow, it would have been all too easy to hide behind the often impenetrable prose of hermeneutics and linguistics but the authors, to their credit, have selected most of their terms carefully and argued clearly.

My criticisms are that they occasionally set the "rationalist" position up as a straw man, caricaturing the dominant view in a form to which few would subscribe (e.g., the concept of theory-free observation has fewer supporters in cognitive science than Winograd and Flores might suppose), and fail to provide a genuine example of how a system designed according to their approach might look. To those of us concerned with designing systems for real users this is a shortcoming that no amount of intellectual posturing can redress. For example, on the subject of electronic libraries, few of us interested in hypermedia will take much inspiration for design from the conclusion that "in providing a tool (for browsing and searching material) we will change the nature of how people use the library and the materials within it" p167.

Notwithstanding this, it is an interesting, occasionally controversial and in some senses, ground-breaking book (who am I to argue with Weizenbaum?). Readers with a background in linguistics and/or hermeneutics might wonder what the fuss is all about, they have known all this for years! Others who have been dismayed by the manner in which technology and design are implemented might be relieved to see written expression of their thoughts reach such exalted heights and intrigued that there is an established philosophical thesis they can adopt to lend credence to their position. Some people might even be surprised. Thus it is the sort of book many will find either stimulating or totally irrelevant. The ideas here are not new per se, but they are unorthodox in the domain of HCI and if for no other reason than this, it must be recommended.