Gosei when given a task analysis of no more than 4 steps, given full to partial physical prompts, verbal prompts, and gestures/modeling remain on task with 60% accuracy.