Quantifying Errors of Bias and Discriminability Emitted by Children during a Matching-to-Sample Task
Date of Award
7-2018
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Behavioral Analysis
First Advisor
Christopher Podlesnik
Second Advisor
Corina Jimenez-Gomez
Third Advisor
Adam T. Brewer
Fourth Advisor
Darby Proctor
Abstract
Children diagnosed with Autism Spectrum Disorders (ASD) make errors during discrimination training regardless of antecedent or consequent procedures implemented to decrease errors. Further, these interventions are not guided by the source of errors. Two equations from Davison and Tustin’s (1978) framework can quantify errors due to bias and discriminability, known as log𝑏 and log𝑑, respectively. This framework categorized errors emitted by children diagnosed with ASD during a matching-to-sample task. The task was displayed on a touchscreen device in which touching a sample stimulus at the beginning of each trial resulted in the appearance of two comparison stimuli. Researchers delivered reinforcement for touching the matching comparison stimulus. More similar sample stimuli were introduced during Phase 2 while keeping the comparison stimuli the same which affected sample discriminability only with little effect on biases for two of three participants. This framework accurately categorized errors emitted by children with ASD when levels of difficulty between the sample stimuli were manipulated. Future research might be able to use these equations to better categorize errors children with ASD exhibit during conditional discriminations. Future research might also be able to improve teaching procedures by targeting interventions to mitigate or eliminate specific errors due to biases or reduced discriminability.
Recommended Citation
Hannula, Courtney, "Quantifying Errors of Bias and Discriminability Emitted by Children during a Matching-to-Sample Task" (2018). Theses and Dissertations. 123.
https://repository.fit.edu/etd/123