ITI: Simulated learning companions

Here's a long post about a session I saw yesterday. Sorry everyone! Skip it if you are not interested in simulations and/or computer-assisted interactions.

Session: Impact of the affect and gender of a learning companion on learning outcomes
Yanghee Kim, USU

Yanghee had some interesting ideas about simulated learning companions, which are basically computer-generated “peers” that e-learning students can relate to during a computer-mediated learning activity. She was reporting on her dissertation research about how the affect (how happy/grumpy) and gender of a learning companion could impact the learning. Her prerequisite assumption was that the more a learning companion showed emotion, the more believable the simulated peer would be and the more effective it would be in helping to mentor the students.

Her logic went something like this:
People learn better through interaction – they prefer interaction with peers – peer interaction is absent in strict Computer-Assisted Learning (we won’t talk about blended environments here) – would a simulated peer be able to fulfill this role?

She also did a second experiment investigating responsiveness of the learning companions. After completing each part of the CAI activity, there were emoticons, and they were supposed to pick one. If they said they were happy or doing well, the LC said “great! Glad you’re understanding it.” If the student was struggling, the LC expressed understanding and sympathy and gave reinforcement.

Her specific hypotheses were something along these lines:
A happy, positive learning companion creates more positive student attitudes
A responsive learning companion will positively impact self-efficacy and student affect.

The results? Of course her hypotheses were supported. She also found that male LCs were more effective at mentoring than female ones. I think this must be because the class was mostly female, and not because a male LC is better.

(As aside about positivistic research. Yanghee had to report her finding that, in her study, a male LC is a better mentor, even though she isn’t convinced. So we’re not convinced, but we report it as a “finding” anyway!)

Here are my thoughts that stemmed from her presentation:

• Could we create simulated learning companions that can take a variety of roles (mentor, coach, teacher, student, etc., whatever is needed for the learning situation)?
• Could students pick an LC that would most uniquely match their own attributes? I think Bandura said something about how we learn best from those with similar attributes to our own.
• Could we allow students to pick a new LC any day or for any part of the activity? For example, in Mario Bros. games, you can pick Mario, Luigie, Donkey Kong, etc to be your opponent. Could we have those options in an LC? Let the students pick a hunk to be their learning companion if they want.

• An exciting implication, I think, is to allow students to learn by teaching (she alluded to this but didn’t elaborate it). Maybe we could allow the live student to learn by teaching the simulated LC?