The Introduction begins: concepts are the glue that holds our mental world together. And what Murphy has done is to survey the current (as at 2002) state of the art regarding what concepts are, how they arise and how people use them.
He's quite careful about his terms of reference. The introduction includes a section "How This Book Differs from the Book You Expected To Read", contrasting the organisation of The Big Book of Concepts with its predecessor as a survey of the field, Smith and Medins' CategoriesAndConcepts?; specifically, The Big Book of Concepts does not use the various theories of concept as its organising principle. Interestingly, this is because, in the intervening twenty years, the study of concepts has become more, not less, strongly riven by dissent and disagreement over the theoretical basis of concepts, while also making great strides in the detailed study of how people create and use them.
At the end of the book, too, is a an "Anti-Summary", reminding the reader of what has not been covered.
Well, I haven't read the whole thing, nor will I for a long time yet. Amongst other things, it's a long and complicated book about a difficult subject, but with no diagrams. I find diagrams helpful in understanding complex content, and miss them in this case. Some of the text is a little rambling, too, I find.
I have received value from what I have read, however, and this is it--It seems that there are several competing theories of how concepts work, which is very interesting all by itself. There is the "classical" view of categories, arising no later than the work of Aristotle, and based on definitions. Arising no later than Aristotle, and dominating linguistic, philosophical and psychological discourse until around 1970! In the classical view, concepts are represented in the mind by definitions, and definitions are built out of lists of necessary and jointly sufficient characteristics for an entity to conform to the definition (and therefore be part of the concept). An important feature of the classical view is to use the Law of Excluded Middle to claim that categories are disjoint.
More recent views of categories are based less on rules about membership, and more on comparing individual instances.
Beginning in 1973 Rosch publishes a number of alternatives to the classical view, based upon prototypes. in this view a category has some sort of best example, to which other things can be compared and membership of the category determined by comparing entities with it. Fans of this view vary as to whether the best example is an imaginary ideal instance, a particular real instance, a feature list, a feature list with weightings attached to each feature, or a feature list organised by some schemata. This is where I found my mental ears pricking up: a schema represents the properties of an item by slots, which hold values specified by filters. The filter on a slot says what set the values in the slot may be drawn from, and slots may be related to one another by constraints. Some of you are probably ahead of me at this point.
It's broadly this view of concepts, and its criticism of the classical theory, that lies at the heart of Lakoff's WomenFireAndDangerousThings.
In 1978 Medin and Shaeffer publish a paper in the Psychological Review proposing what's now known as the "exemplar" view of categories, in which your understanding of a category is your memory of all the things that you've seen before that you put in that category (also using some sort of similarity metric).
These theories (amongst which there is a good deal of variation between practitioners), are responses to the classical theory, and there has been a response to them, the "knowledge" approach. In this model (which is new and still developing) categories are things worked out from our general knowledge of the world, which is captured in terms of facts and theories. New knowledge gives rise to new categories which give rise to new knowledge. Some call this the "theory theory" of categories.
The interesting result is that there is no clear winner amongst these theories, each of them accounts for a bunch of psychological and cognitive phenomena that overlaps with that accounted for by the others. The bulk of the book, indeed, is concerned with these phenomena and how each theory deals with them. But, and it's a big but, there is a clear loser. That's the classical view, which is definitely wrong. What might be nice is a discussion of why the classical view is so attractive, while being so very untenable and unhelpful.
The failure of the classical view of concepts is a shame, since the currently dominant view of object-orientation (which in one sense is nothing but the business of taking a user's concepts and reifying them into code) is very firmly rooted in the classical view. Oh dear. Perhaps that's why so many people find classical object-orineted analysis so hard to do well, find it so difficult to understand and create those models, "obvious" as they idea is presented as being. And the news may be even worse for relational fans. And that makes The Big Book of Concepts urgent reading for anyone embarking on any sort of large scale information modelling exercise.--KeithBraithwaite
Rambling is a kind adjective to use in describing the book. I wouldn't rate it as worth buying, neither would I rate it as worth reading attentively, other than the chapters on word meaning which are more coherent. One of the assumptions carried throughout the text is that concepts and categories can be treated as synonymous, another is that there is only 'a way' in which we conceptualise. -- HuwLloyd
![[Home]](/wiki.gif)