Exploring Strategies for Melanoma Detection Utilizing Discrimination Training

Abstract

Melanoma is the deadliest form of skin cancer. Although melanoma is amenable to visual identification by those who might suffer from the disease, no consensus exists on a single strategy for promoting early detection. To date, the application of behavioral principles has been all but absent from the literature attempting to address this issue. The conceptually systematic knowledgebase on learning and behavior offered by behavior analysis has the potential to contribute substantially toward developing strategies for early detection of melanoma. In particular, generalization is a ubiquitous behavioral process with an extensive literature from which to draw. As such, the purpose of the current series of studies is to employ a use-inspired translational approach to explore strategies for promoting early detection by capitalizing on behavior analytic research regarding the processes of generalization and discrimination (i.e., peak shift). The purpose of the experiments was to (1) use discrimination training to establish generalization and postdiscrimination gradients with moles as stimuli, (2) determine the effects of parametric manipulations of training on postdiscrimination gradients, and (3) evaluate training with multiple discriminative stimuli. Results from Study 1 indicated that discrimination training produced gradient shifts as compared to a control group trained only with the S+. Results from Study 2 indicated that training with an S- more distinct from the S+ produced gradient shifts, but that S- stimuli more similar to the S+ did not. Results from Study 3 indicated that training with two S- stimuli from one extreme of the stimulus array produced relatively weak shifts in postdiscrimination gradients, but that training with an S- at both extremes of the array was effective in producing highly consistent response patterns. Theoretical implications and future directions toward more clinically relevant studies are discussed.