Introducing the Female Cambridge Face Memory Test – Long Form (F-CFMT+)

F-CFMT+ is designed to match all the parameters of the M-CFMT+ (Russell et al., 2009) at both the block (see Task Figure) and trial levels. The stimuli include high-resolution images of individual White women in the age range between 20-30 years. They were selected from the Radboud (Langner et al., 2010) and Karolinska (Lundqvist et al., 1998) faces databases. We selected images of multiple face identities from different viewpoints (frontal, 45, 135) to match the training trials in the M-CFMT+, and with different expressions to match the manipulations in the fourth block of the M-CFMT+. The images were converted to grayscale and cropped to exclude the background. The hair and all skin blemishes and discoloration (e.g., freckles) were removed from the faces. The 6 target white female faces were selected to exhibit a similar range of perceptual variability as exhibited among the 6 target white male faces in the M-CFMT+. As in the original CFMT and the M-CFMT+, the target identities appear on 1/6 trials throughout all four blocks of the task. Each target appears approximately 17 times throughout the task (range 16-18); the variability in rate of presentation is a function of the relative perceptual similarity between the target and the distractor identities. To create the task, each female target face (e.g., target face 1, target face 3) was substituted for the matching target male face (e.g., target face 1, target face 3) at the trial level throughout each block of the task. To match the M-CFMT+ task, 46 distractor identities were required with frontal and profile images. However, because of limitations with the stimuli databases that were available at the time of task development, only 38 distractor identities met these criteria. On average, each face appeared 5.37 times throughout the task. Therefore, the frequency of the distractor faces in the F-CFMT+ was a bit higher (1/19 trials) than in the M-CFMT+ (1/23 trials) in the first three blocks. To make the noise-distorted images for Blocks 3 and 4, we applied a 30% level of Gaussian noise to the images, as described in the creation of the M-CFMT (Duchaine & Nakayama, 2006). However, given the high resolution of the female face images, we had to apply multiple iterations of this Gaussian noise to degrade the images to a similar level as in Blocks 3 and 4 of the M-CFMT+. In the final block 4, as in the M-CFMT+, we selected 14 of most frequently occurring distractor faces to repeat more frequently and increase the difficulty of this block by reducing “the difference in familiarity between target and distractor faces” (Russell et al., 2009). Pilot testing on emerging adults at Penn State University was used to adjust the difficulty level of the task via selection of target and distractor items. Preliminary data with an earlier version of the task were published previously (Scherf et al., 2017). Results from this study were used to generate the power analysis and to refine the task for this round of testing and release. This final version of the F-CFMT+ uses the same java code in which the M-CFMT+ was originally programmed. The instructions, task structure, and task timing are the exact same between tasks. Psychometric properties of the task are reported in Arrington et al (under review).