Document Type


Degree Name

Master of Arts (MA)




Faculty of Science

First Advisor

William Hockley

Advisor Role

Thesis Supervisor


The present study examined the effect of environmental context on the mirror effect in recognition memory. In seven experiments, participants studied either high (HF) and low-frequency (LF) or noun and nonnoun pairs of words followed in a old/new item recognition test by which proportion correct, response time, and confidence judgments were measured. Single item or word pair targets and distractors were presented in same- or different-context conditions. Context was defined as the unique combination of foreground and background colour and position on a computer screen for two experiments while position was removed as a context variable for five experiments due to influences on response time. The nature and types of the different-context conditions were manipulated in addition to the number of study context repetitions. Single item recognition tests produced statistically independent mirror effects (higher hit rates and lower false alarm rates for LF compared to HP words and nouns compared to nonnouns) and effects of context (higher hit rates and higher false alarm rates for old compared to new contexts). Results also indicated that words were not associated with specific contexts but rather the type of context as new combinations of old context elements and repeating old contexts had no effect on performance. Confidence level also had no effect on performance as these results were replicated when conditionalized on high-confidence responses. Tests of word pairs failed to show an effect of context while demonstrating an effect of word type for hits only. For the most part, results support the view that effects of environmental context are due to a response bias arising from changes in response criteria (upon which recognition decisions are based) across test context conditions. This leads to a greater likelihood of “new” responses when recognition tests are in a novel context and a greater likelihood of “old” responses when recognition tests are in a same/old context. Evidence of differences in discriminability between different word types supports the notion that the mirror effect is a memory-based phenomenon characterized by properties that make participants more sensitive to LF over HF words (and nouns over nonnouns).

Convocation Year


Convocation Season