[Online] Effects of Stimulus Type on Recognition Accuracy

Background

The classic Signal Detection Theory assumes equal variance between the target and lure distribution. However, several decades of research in recognition memory has almost never found equal variance results. What remains interesting in these literatures is that the reliable asymmetry in target vs. lure variability has been reported in two opposite directions - while numerous list memory experiments with word stimuli find evidence for greater target variability, a lot of eyewitness memory experiments are showing the opposite evidence for greater lure variability. Despite conflicting findings, it is possible that there are a number of procedure and stimulus differences underlying these experiments.

Research Questions / Hypotheses

The present study was designed to answer the question that, whether it is the stimulus type (words vs. faces) or procedure differences that influence on the findings of greater target or greater lure variability.

Participants

80 participants were recruited in this study. 14 participants were excluded either due to poor performance, a large proportion of fast responses (likely to be fast guesses), or insufficient data.

Methods

In this experiment participants completed several cycles of a computer-based recognition memory task. In each of the task, participants were presented with series of words or AI generated faces from four ethnicities (Asian, Black, Latino and White) and two genders (male and female). Participants were required to remember each item. They were then presented with a simple true/false maths task that acted as an interference task, with the aim of negating the effect of short term memory during the retention interval. At test, participants were presented with a selection of the original items (and some lure items) in the centre of the screen. The task was to identify whether they had seen each item previously, or whether it was a lure, and state their confidence in this recollection using a 6-point confidence rating scale.

Results

Both words and faces stimuli generated zROC slopes less than 1, meaning that both types of stimulus showed evidence for greater target variability.

Implications

The results suggest a possibility that it may be procedure rather than stimulus difference that produces different findings on target vs. lure variability in previous research.