Discrimination, response bias, and context: An explanation for elusive environmental context effects in recognition memory

Alicia Janise Knoedler, Purdue University

Abstract

Environmental context effects have not been consistently observed in studies of recognition memory. Several hypotheses have been generated to explain why this inconsistent pattern exists. Some hypotheses have focused on methodological issues, others concern how the environment is reinstated at test, and still others have focused on the nature of the contextual cues in relation to the cues associated with item information. A simpler hypothesis may explain this inconsistency and, in addition, suggest a novel approach with which to conceptualize context effects in recognition. This hypothesis focuses on the memory and decision components of recognition memory. Typically, context effects are found in hit and false-alarm rates and measures of response bias but not in measures of discrimination. From these findings, it appears as if context manipulations affect a person's decision strategy about the stimuli rather than their memory for the stimuli. Decision strategy is measured through shifts in response-bias. Measures of response bias, however, are not usually reported in studies of context effects in recognition. Typically, the proportion of hits, and sometimes the proportion of false alarms, are reported to reflect memory performance in a changing-context situation. Without measures of discrimination and response bias, a complete explanation of context effects is not possible. The discrimination and response bias explanation suggests that context effects in recognition memory should be measured with hit and false-alarm rates, measures of discrimination, and measures of response bias to give a complete explanation for the context effects. Four experiments were conducted to examine this explanation of context effects in recognition. Two experiments attempted to replicate previous studies that have identified reliable context effects in hit and false-alarm rates. Two additional experiments directly manipulated response bias to further explore how changes in response bias affect hit and false-alarm rates in addition to measures of discrimination and response bias. The findings from these experiments suggest that response bias may be the factor underlying context effects in recognition, but directly manipulating response bias only changes overall performance. It does not change performance differentially between the same and different contexts.

Degree

Ph.D.

Advisors

Neath, Purdue University.

Subject Area

Cognitive therapy|Psychology|Experiments

COinS