Notes from Donna Mauer’s talk at Oz-IA 2006.

Based on George Lakoff’s book about categorization and cognition. The way that we think about categories in our brains. It offers a challenge and alternatives to the classical theory of categorization. (Most of the good stuff is in the Preface and Chapters 1 and 2.) Understanding categorization is central to information architecture.

The reason it’s important is because the idea of category is central….most of our words and concepts designate categories. How we think is in categories. We don’t think in discreet elements.

This goes against the classical theory of a category, which is an abstract container with things inside or outside clear boundaries. Categories are defined by common properties of the members. No member of a category has a special status–all are equal. All levels of a hierarchy are important.

All of this isn’t true.

There are no clear boundaries. There aren’t common properties.

Basic-level categories.

Generalization occurs at the middle of a hierarchy. Learned earliest. Usually has a short name that is used frequently. A single mental image can reflect the entire category. No definitive basic level in all cultures. Dependent on who is thinking about it.

Prototype Effects.

Categories often have best or prototypical examples. (Robins are more “bird”-y than a pelican, for instance.)

Other challenges to classical categorization theory.

There are degrees of membership in any category and no clear boundaries. Not always clear what is inside or outside a category.

Idealized cognitive models: some of our models are complex structures that need a frame of reference (like what a weekend is or what a bachelor is or what a lie is).

Cluster models.

Mother–birth mother, genetic model, martial model, genealogical model.

Radial categories.

A central concept, with variants (like Mother).

Ideals.

Abstract ideal cases, which may not be typical of stereotypical (“the ideal husband”).

Why is this important to us? Our computer software is built around the classical idea of categorization. Especially in our file systems! We define categories to fit the computer.

How we can use basic-level categories in our work.

Analyze user research data to identify basic-level. Words that are used frequently, that children understand, and come out easily.

Use basic-level categories as trigger words. Easily recognized and have good scent.

Card sort with basic-level content items rather than more granular content elements.

In navigation, get people to basic-level as soon as possible.

Instead of top-down or bottom-up, working at the basic-level up and down.

Knowing all this will make you less stressed about why the categorization is not neat.

“Miscellaneous” categories are cognitively real, just not easy to use as navigations.

Being approximate is ok. We can put some things in two places.

Use prototypical items when communicating because they represent the category well. People connect strongly with them.

Is this why tagging is so popular?