Reading focus
Syllabus: How Do We Organize Knowledge?
Ask what prototypes and exemplars leave open about why particular features matter. Examine the premises of the knowledge-based argument and its obligation to make constrained predictions.
A question to consider. Why might changing what you believe about an object's mechanism alter its category without changing its appearance, and what would a similarity account need to specify to explain that change?
Why this reading
Why do some objects belong together in our thinking while an arbitrary collection of similar features does not form a useful concept? Medin examines the limits of definitions and similarity, then asks how background knowledge gives a category coherence. The argument changes what an explanation of categorization needs to explain.
Before you read
A necessary property must be present for membership; sufficient properties guarantee it. A typical property can be common without being necessary. Conceptual coherence concerns why a category's features and members hang together. Here a theory can be an informal network of beliefs about causes and relations. Psychological essentialism is the belief that a hidden nature makes something the kind of thing it is, not evidence that such an essence exists.
Read for
- Identify the different objections to defining concepts by strict rules and by lists of characteristic features.
- Ask how choosing features and weighting similarity already depends on knowledge. Track the proposed integration of similarity with explanation rather than treating them as mutually exclusive.
- Test whether appealing to a theory explains coherence or simply moves the unanswered question to another level.
Consider afterwards
- Imagine two unfamiliar tools with similar shapes but different mechanisms. Which features would matter for classifying them by function, and what does your answer assume about how they work?
Editorial context
This 1989 synthesis develops an argument about the explanatory limits of similarity; it is not a single experiment or a complete computational theory. Its treatment of psychological essentialism concerns people's beliefs, which may be mistaken. The proposal that knowledge constrains categorization must still specify what knowledge is represented, how it is used, and which observations could count against the account. Murphy and Medin's 1985 article supplies the more extended original argument.