Back to Instructor's Page

In addition, these data are combined into the discrimination index referred to as A’ (A-prime) where

Rates are computed as the proportion of judgments that fall into each of the response categories. The resulting number A’ ranges from .50, mere chance discrimination, to 1.00, perfect discrimination. (For information on A’ and other indices of discrimination, see a standard sensation and perception text such as Coren, Porac, & Ward.). In summary, then, five pieces of data are stored for each participant who completes the Day 1 portion of the experiment: Number of Hits, Misses, False Alarms, Correct Rejections, and then the derived measure A’.

When participants return to the Facial Recognition Experiment on Day 2, the program retrieves the list variables that specify which faces were assigned to which list (Most Wanted, Foils List1, and Foils List 2). Based on these, a new Mug Book identification task is constructed. As before, there are 20 serially presented faces. Ten are the Most Wanted faces and the other 10 either Foils List 1 or Foils List 2. The choice is made randomly based on whether the minutes on the system clock is odd or even.

Using two different Mug Books on Day 2 creates a between subject variable for analysis. Research participants who get a Day 2 Mug Book that includes the original Most Wanted faces and 10 previously unseen foils are control subjects. Their performance decline from Day 1 to Day 2 will reflect the effect of time alone on the memory for the Most Wanted faces. But the memories of subjects who get a Day 2 Mug Book that includes the same 20 faces as the original Day 1 Mug Book will be affected by more than just time. Their memories will be affected also by the intrusion of false recognitions that result from seeing a "familiar" face. For that reason, this study of facial recognition is really a study in false recognition: To what extent does seeing a distractor face in the same temporal context as a to-be-remembered face increase the probability that the distractor face will later be falsely identified as one of the to-be-remembered faces?
« Return to Top»

One way to address this question statistically is to compare the discrimination index scores for participants in the Control and False Recognition groups. Discrimination index scores are good because they combine information from the four specific measures–Hits, Misses, False Alarms, and Correct Rejections. But it would also be of interest to examine false alarm rates in the two groups because this is a direct measure of the effect that repeating foils should have.

The data to the left show the results of a false recognition study conducted at UM in Fall 1998. The dependent variable in this case is the discrimination index A’. The raw data are given in the table below along with summary statistics.