Response Latency During Acquisition: A Comparison of Stimulus Presentation Algorithms in a Match-to-Sample Task

Abstract

Four studies were developed to investigate response latency in a computerized fluency task with college students. The first study was exploratory, and it tracked the patterns of response latency throughout the frequency-building process. Results indicated that response latencies became more accurate, stable, and uniform as response frequencies increased and this stability occurred alongside traditional fluency outcomes. Study 2 used similar frequency building task as study 1, and it compared two stimulus presentation algorithms in an alternating treatments design. Each algorithm utilized its own respective stimulus set with of seven stimulus pairs. While both algorithms presented all the stimulus pairs in their respective stimulus sets, the correct-weighted algorithm presented stimuli that occasioned correct responses at a higher rate. The error-weighted algorithm presented stimuli that occasioned incorrect responses at a higher rate. All participants in study 2 failed to reach the mastery criterion likely because the task involved learning too many stimulus pairs at once. Study 3 was identical to study 2 with the exception of a reduced number of stimuli in each set. All participants met the mastery criterion under both algorithms. The correct-weighted algorithm produced mastery the quickest for 3 of the 4 participants. Study 4 was similar to study 3 with a procedural variation and a within-participants replication. For 4 of the 5 participants, the correct-weighted algorithm produced mastery the quickest. In sum, the present research presents novel methods to assess skill level, a method for identifying effective treatments, and an automated system for shaping behavior. The implications and future fluency research, the education industry, and basic experimental research are discussed.