[quoting Bergeron, 19 May 99] These simulations are certainly more challenging to design than old-fashioned Skinnerian stimulus/response models...
I know this is contrary to popular misperception, but anyone who takes the trouble to read what Skinner actually wrote about teaching machines and study the examples of programmed instruction that he employed, will discover that Skinneian machine-based learning was not old-fashioned, and not based on notions of stimulus and response, but Socratic-style instruction-pupil interaction. In the Skinnerian model, students "discover" answers to queries with which they then CONSTRUCT the target knowledge.