Features of Computerized Multimedia Environments that Support Vicarious Learning Processes - New Science of Learning - page 63

curriculum domain. It contrasted 8th, 9th, 10th, and 11th grade students presented
with interactive (AutoTutor) tutoring sessions with similar students presented vicar-
ious learning conditions involving either computer literacy or Newtonian physics.
The 8th and 10th graders were asked to learn computer literacy, while the 9th
and 11th graders were presented with Newtonian physics. Three conditions at each
grade level included interactive sessions with AutoTutor, monolog presentations of
sentences in the ideal answers and expectations, and dialog with deep questions pre-
ceding each sentence in the ideal answers and expectations, as described in Craig
et al., 2006; Gholson & Craig, 2006).
The analysis of most interest revealed significant differences between the exper-
imental conditions. Those in the dialog (with deep questions) condition showed
pretest to posttest learning gains nearly twice as large as those in the interactive and
monolog conditions. The interactive and monolog conditions did not differ from
each other. These data provide support for the “deep questions effect” (Craig et al.,
2006; Gholson & Craig, 2006) among 8th to 11th graders in the domains of com-
puter literacy and Newtonian physics. As noted earlier, the effect had previously
been shown only among college students in the domain of computer literacy (see
Craig et al., 2006; Driscoll et al., 2003; Gholson & Craig, 2006).
Next we consider the potential role of presenting vicarious explanations of course
content during learning. During the past two decades, researchers have shown that
when learners overtly explain aloud each concept or solution step as it is presented
to them, those self-explanations produce very large learning gains when compared
to various controls (Ainsworth & Burcham, 2007; Chi, 2000; Chi et al., 1989; Chi,
Leew, Chia, & LaVancher, 1994; Chi, Siler, Jeong, Yamauchi, & Hausmann, 2001;
Rummel & Spada, 2005). Self-explanations tie new content to previously presented
materials and/or embed into a rich web of prior knowledge. Chi (2000; Chi et al.,
2001) suggested that self-explanations increase knowledge acquisition by encourag-
ing cognitive activities that fill knowledge gaps, modify existing schemas, and link
new content to related material (Chi et al., 1989; McNamara et al., 2004).
This research led to the following question: What learning gains might be
achieved by vicarious learners under conditions in which they overhear a virtual
agent provide scripted “self-explanations” of course content? Would overhearing
dialog that included virtual agents presenting scripted self-explanations of course
content facilitate learning gains? Thus, as part of preliminary investigations Craig,
Brittingham, Williams, Cheney, & Gholson (2009) included conditions in which
vicarious statements analogous to a self-explanations (e.g., Chi et al., 2001) were
presented to each student by a virtual tutee. Following each statement containing
course content, spoken by a virtual tutor, the virtual tutee provided an explanation
that was analogous to the self-explanations provided by learners in earlier research
(e.g., Chi, 2000; Chi et al., 1989).
The first study was conducted in the laboratory and included only college stu-
dents as participants. Students with high or low knowledge (based on pretest scores)
concerning Newton's three laws of motion were randomly assigned to one of four
experimental conditions: In a vicarious question + explanation condition, the pre-
sentation of each concept involved a three-event sequence. First the virtual tutee