Abstract
Subjects performed a task which allowed measurement of their speed of comprehension of real-world scenes: They attempted to detect the incongruity in the relationship between an object and its context. In 100 msec, presentations of line drawings of real-world scenes, objects could be inappropriately positioned (e.g., a fire hydrant on top of a mailbox), sized (e.g., the hydrant looking larger than a truck) or floating in air. Detection times were not significantly affected by priming the subject with a verbal label of the scene prior to its viewing (e.g., telling him he was going to see a “kitchen”) but were significantly affected by the prior presentation of the scene upwards of 30 trials earlier. The results document a remarkably efficient, content addressable, accessing mechanism for visual recognition in humans.
Get full access to this article
View all access options for this article.