Quantifying errors of bias and discriminability in conditional-discrimination performance in children diagnosed with autism spectrum disorder
Hannula, Courtney, Jimenez-Gomez, Corina, Wu, Weizhi, Brewer, Adam T., Kodak, Tiffany, Gilroy, Shawn P., Hutsell, Blake A., Alsop, Brent, and Podlesnik, Christopher A. (2020)
Abstract:
Antecedent- and consequence-based procedures decrease errors during conditional discrimination training but are not typically guided by error patterns. A framework based in behavioral-choice and signal-detection theory can quantify error patterns due to (1) biases for certain stimuli or locations and (2) discriminability of stimuli within the conditional discrimination. We manipulated levels of disparity between sample (Experiment 1) and comparison (Experiment 2) stimuli by manipulating red saturation using an ABA design with children diagnosed with autism spectrum disorder (ASD). Lower disparities decreased discriminability and biases were observed for some participants during the low-disparity conditions. These findings demonstrate the use of these analyses to identify error patterns during conditional-discrimination performance in a clinically relevant population under laboratory conditions. Further development of this framework could result in the development of technologies for categorizing errors during clinically relevant conditional-discrimination performance with the goal of individualizing interventions to match learner-specific error patterns.
Citation: Hannula, Courtney, Jimenez-Gomez, Corina, Wu, Weizhi, Brewer, Adam T., Kodak, Tiffany, Gilroy, Shawn P., Hutsell, Blake A., Alsop, Brent, and Podlesnik, Christopher A. (2020). Quantifying errors of bias and discriminability in conditional-discrimination performance in children diagnosed with autism spectrum disorder. Learning and Motivation, 71. 101659. https://dx.doi.org/10.1016/j.lmot.2020.101659