How Institutions Think (Frank W. Abrams Lectures) Paperback – July 1, 1986

This book begins with a dilemma around a group of men, caught in a cave by falling rocks. Starved of food they have to decide whether cannibalism would be acceptable. Douglas argues that there is no universal solution to this dilemma and that the different possible outcomes depend on the instituional relationships among the individuals.
Mary Douglas admits in the early chapters of the text that this is the book she should have written first. Many of her earlier books, including Purity and Danger and Natural Symbols attracted critical acclaim but failed to make her theoretical orientations explicit. Her approach is in a minority position in contemporary social science and is informed by the sociology of Durkheim as practiced in anthropology in the first half of this century.
As with all of her books, HIT is a great pleasure to read and she illuminates critical academic concerns with in a voice that is engaging. The book was written as a response to Olson's 'Logic of Action' and so focuses on demonstrating the contrary position that collective action does have rational foundations.
This is a short text which serves as a useful introduction. It is impossible to understand the breadth of Douglas' profound insights without also reading 'Implicit Meanings', 'Purity and Danger' and 'Risk and Blame'.

Ever since the time of Descartes, and very probably since the time of the ancient Greeks, we have been deeply enamored with the idea that we - conscious, rational, decision-making beings - control the way that we think and act. While Mary Douglas certainly doesn't suggest that we are just mindless cogs in a machine, she does offer some interesting insights into how we think about institutions, categories, and rationality that have serious implications for the idea of wholly autonomous human intellectual agency. Douglas, one of the greatest social anthropologists to come out of England in the twentieth century, is known better for her "Purity and Danger," "Risk and Blame," and "Implicit Meanings." "How Institutions Think" is a series of Frank W Abrams Lectures that she delivered at Syracuse University in 1985.

Some scientific ideas enter the world, readily accepted and widely read by an eager scholarly community. Others languish - but not because they are of a lesser quality, and not even because they are incorrect. Ludwig Fleck's book on the discovery of syphilis titled "The Genesis and Development of a Scientific Fact" (Entstehung und Entwicklung einer wissenschaftlichen Tatsache), was one of the latter. Anticipating a sort of social constructivism, Fleck said that scientific ideas are accepted or rejected into a canon for reasons not because of their inherent worth, but because certain social conditions (Fleck called these "thought-conditions" or "thought collectives") allow or disallow their admittance. The Denkkollecktiv is a whole matrix of social circumstances, thoughts, and assumptions that envelop the scientific project. If a new idea substantively differs from one some aspect in the Denkkollecktiv, it will be ignored - not consciously by individual scientists, but by the scientific profession as an institution.

This is where Douglas picks up on her anthropological history of the "classificatory enterprise." How can the phenomenon that Fleck described really happen? What was it about science - or any institution - that shapes social cognition and categories so profoundly? How do these institutions develop, and why? Following Durkheim's lead, Douglas claims that autonomy (in the sense of radical social individualism and atomization) was in many respects an illusion, and that we are marked by a strong sense of social solidarity through shared "classification, logical operations, and guiding metaphors [that] are given to the individual by society."

One of Douglas' implicit arguments is that the difference between sociology (group action) and psychology (action of a single agent) is wrong-headed. Instead, she asserts that for a rule to turn into a legitimate social convention, it needs a parallel cognitive convention to sustain it (46). Social institutions encode information, and then use that information to minimize the entropy, or inherent disorder, in decision-making. The stabilizing principle of institutions - what keeps them from breaking down - is that they "naturalize social classifications." By naturalizing the social in reason, the institution automatically legitimizes it. After all, one of the first priorities of the institution should be legitimacy, or else it would incur so much doubt that it would eventually be destroyed. Douglas considers the common social analogon of likening the roles of men and women to that of the left and right hand; the logic of complementarity in nature legitimizes the social order, constructing a rationality which seems like it was there before time began. Even the institution of sameness (yes, even sameness - logical similarity - is an institution) is time- and culture-sensitive. Douglas cites the example in Leviticus of the camel, the hare, and the rock badger: they all chew cud, which would lead us to believe that they would all be classified as cud-chewing ungulates. However, since they don't have cloven hooves, they are excluded. The criterion of difference here, having a cloven hoof, is completely arbitrary - yet it is the sole category that bestows "sameness" on a group of individuals. Douglas makes it clear that categorizations like this are not cool, objective observations into the inner working of nature. They are very telling maps that "model the interactions of the members of the society."

Institutions do more than order categorical knowledge. They also filter information in such a way that they can be said to remember and forget. Fleck's book recounts a classic case of institutional forgetting. The failure of a legitimate scientific idea to be accepted into the prevailing canon of knowledge is usually the result of a lack of "formulaic interlocking with normal procedures of validation." For an idea to gain acceptance, it sometimes has to exploit "the major analogies on which the socio-cognitive system rests." Evans-Pritchard's study of the Nuer focuses on institution remembering, according to Douglas, discovering "an explicit demonstration of how institutions direct and control memory" (72). He showed how economic interactions, including weddings and cattle distribution, order the memory of ancestors.

The one thing I was curious about while reading the book was that Douglas never mentioned Berger's "Social Construction of Reality," which covers much of the same territory. Of course, not being an anthropologist, Berger has a somewhat different take on matters. "How Instutitions Think" should definitely be read alongside the Berger, and I think the reader will find that they shed light on one another. Highly recommended.

I was recommended this book by a social science colleague whose judgment I trust. I certainly would not have bothered on my own, after reader her book with Wildavsky, which I found to be shallow and misguided (see my review of this book).

Douglas's major thesis in this book is that the notion of rational action as the behavior of the socially isolated self-regarding actor is incorrect. Her alternative is that "mind is society writ small." That is, individual beliefs, desires, and hence actions, are the product of the individual's social connections and social experience, and have a strong other-regarding and ethical element of self-sacrifice on behalf of the group and its moral principles. She develops this theme with great eloquence, insight and erudition.

Nevertheless, her arguments could have been stronger. Even thirty years ago, she would have benefitted from the incredibly powerful works of the sociologists Talcott Parsons and Ralph Linton (Douglas only recognizes the late Parsons, whereas the early Parsons of 1937 is relevant for her thesis). Moreover, she does not understand the scientific notion of rationality, as developed by Von Neumann and Morgenstern as well as, even more critically, Leonard Savage (1954). In terms of modern decision theory, stemming from Savage, we would summarize her argument is that the subjective prior of rational choice theory must be replaced by a model in which beliefs are socially constructed. Most important, her critique of rational choice theory as necessarily self-regarding is completely incorrect.

Douglas' stress on the importance of Durkheim is well-founded, but she does not find a way to overcome the most common objection to Durkheim, which is that its portrayal of society is completely functionalist---society functions when people agree on basic beliefs and conventions. This is simply incorrect. Societies can function quite well with serious ideational conflict (although general adherence to certain core principles tends to facilitate constructive social cooperation). I have argued that individuals are embedded in social networks of minds, and individual beliefs and knowledge are distributed over these networks. We may call this socially distributed cognition.

Socially distributed cognition almost never takes the form of a single network of consonant information and values, and indeed, major forms of social change are the result of the clash of contrasting networks of social cognition. With this emendation, Douglas' brilliant argument makes much more sense.