Internal and External Context Effects Upon the Types of Information Encoded From Pictures

James G. Penland

This thesis has been withdrawn as it is a duplicate, the original can be found here:

https://commons.und.edu/theses/2501/

Abstract

Several investigators (e.g., Biederman, et al. 1974; Loftus & Bell 1975) have proposed that one's perception and subsequent recogni tion of pictured scenes results from the acquisition of two types of information. One is specific in nature and results from the direct inspection of object detail. The second type is more general in char acter and is thought to result from the processing of contextual in formation. Friedman (in press) has pointed out that context must be considered to encompass both internal (memorial) and external (physical) constraints, and as such to have a dual informational base. The present study was designed to assess the separate and combined effects of these internal and external sources of contextual information as they influ ence the amount of object detail later available to the observer.

Each of 72 (21 male) undergrads was presented with a written phrase prior to a 150 msec, exposure of a pictured scene, followed by a four alternative forced-choice recognition test. Subjects responded by selecting one of the four objects in the test as having been viewed previously, and then rated their confidence in that selection. The type of internal or memorial-based contextual information, prompted by a written phrase describing the theme of each stimulus picture, was varied within subjects such that each subject viewed one-third of the pictures preceded by a compatible theme, one-third preceded by a neutral phrase, and one-third preceded by an incompatible theme. External context and recognition test distractors were varied between subjects. External or physical context was either present or absent, and the distractor objects were objects from dissimilar scenes, objects differing from those viewed but from similar scenes, or objects possess ing the same generic name as the target but differing in some physical attribute.

Distractors from different scenes consistently resulted in the best recognition performance and the most confidence, with no differ ences in accuracy between the other two distractor types. Subject con fidence was a more sensitive measure of distractor effects, as ratings were significantly higher when distractors were varied from the target in object rather than attribute information. Recognition accuracy and confidence were also enhanced when the pre-stimulus prompt was compatible with the stimulus picture, but only when distractor objects were from dissimilar scenes. A trend was evidenced suggesting that recognition accuracy may have been enhanced by the presence of external context when test distractors were from dissimilar scenes, whereas the absence of context may have facilitated performance when distractors differed from the target object only in physical detail.

These findings were interpreted via a consideration of the influ ences of the two types of information available in real world scenes and of the demands imposed on the subject by the particular task employed to study their perception and recognizability.