Non-examples: teaching what a concept is not

Non-examples are near-miss cases that look like a concept but fail one defining feature — studying them fixes the boundary examples alone leave vague.

Why it works

Concepts have boundaries, and examples alone cannot define those boundaries because they only show where the concept applies. Non-examples — cases that are similar in surface features but violate the defining criterion — force the learner to process the concept’s necessary and sufficient conditions, which is exactly the knowledge needed for accurate application. The reason positive examples fail here is that they leave the learner free to latch onto whatever features happen to co-occur: shown only birds that fly, you cannot tell whether flight is part of what "bird" means or just a common accompaniment. A penguin settles it in one case. The pairing has to be close to work — a non-example that differs on many dimensions at once teaches nothing, because the learner cannot tell which difference was the disqualifying one, whereas a near-miss that differs on exactly the defining feature isolates that feature and makes it unmissable. This is the same logic as a controlled comparison: hold the irrelevant attributes constant, vary the critical one, and the boundary reveals itself.

How to do it

1. For any concept, create or find a plausible near-miss: something that looks like it fits but doesn’t.
2. Explicitly state why it fails: which defining feature is absent?
3. Alternate studying examples and non-examples, comparing each pair.
4. Test yourself: present a new case and decide whether it is an example or non-example before checking.

Evidence

Concept learning research consistently finds that contrast cases and non-examples help learners form more precise category boundaries than examples alone. This is especially important for concepts that are frequently over-generalized. Tennyson and Park’s review of the concept-teaching literature concludes that carefully matched examples and non-examples, varied along the concept’s defining and irrelevant attributes, are what let learners fix the boundary and avoid both over- and under-generalization. (observational)

Non-examples are most effective when paired with explanations of why they fail; unlabeled non-examples can reinforce misconceptions.

Sources

• Tennyson & Park (1980), the teaching of concepts, Review of Educational Research
• Tennyson, R. D., & Park, O.-C. (1980). The teaching of concepts: A review of the instructional design research literature. Review of Educational Research, 50(1), 55–70.

Common mistake

Using only positive examples, so the learner builds a concept that is too broad — applying it where it doesn’t belong because no boundary was ever defined.