A New Flavor in Training: Learner Control over Intelligent Agents

Topic: Training
SIOP Presentation: Trainee-trainer similarity in e-learning: Effects with computerized trainers
Presenters: T.S. Behrend and L.F. Thompson
Reviewed By: Benjamin Granger

Despite its disadvantages, e-learning is becoming more and more popular in organizational and educational settings and thus the task is for researchers to explore ways that can help trainees benefit from all of e-learning’s many advantages.

In a paper presented by Behrend and Thompson, one avenue for helping trainees get the most out of e-learning is the use of intelligent agents which act as virtual tutors to trainees (think of the Microsoft paper clip except human!).  When intelligent agents are created to posses human attributes, they are known as animated pedagogical agents (APAs).

Although some research supports the use of APAs, Behrend and Thompson were interested in how learner choices about the physical appearance of, personality of and feedback received from their own APAs affects training outcomes. In their study, the intelligent agent was named PJ and while some trainees were given control over what PJ looked liked and acted like, others were not. Indeed, Behrend and Thompson found that trainees do benefit from making choices about what PJ looks like!

That is, control over appearance positively impacted trainee self-efficacy.  BUT, trainees who were given control over what kind of feedback PJ gave to them, had lower post-training self-efficacy than trainees who were given no control over this aspect.  Finally, control over PJ‘s personality (humorous vs. unemotional) had no effect on training outcomes.

Interestingly, although Behrend and Thompson expected that trainees would report being more satisfied with the training course when given more control over the characteristics of PJ, no such effects were found. Moreover, control v. no control had no effect on how engaged trainees were during training.

Finally (and perhaps most importantly), participants who were given control over multiple characteristics of PJ (appearance, personality and feedback type) tended to learn more from the training course than learners with fewer choices (no choice or choice over only one characteristics of PJ).

So I wonder what setting PJ‘s appearance to 36-24-36 would do for my self-efficacy?

Behrend, T.S., & Thompson, L.F. (2010). Trainee-trainer similarity in e-learning: Effects with computerized trainers. Paper presented at the 25th annual meeting of the Society for Industrial Organizational Psychology.