Agents are the lifelike characters in multimedia software and online applications that pop up on the screen to explain rules, provide hints, or prompt the user to use the program’s features. They can be human or nonhuman, animated or static. What is their role in learning? Can they assume the role of tutor or mentor, increasing the learning value of technology applications? How can you determine whether an agent supports learning or is merely a distraction?

Researchers have investigated the role agents’ play in complex multimedia environments, evaluating the extent to which they guide learners--offering coaching and support--versus distract them. It is not clear that all learners interact with and perceive agents in the same ways. In fact, it is highly likely that different learners would benefit in different ways from agents in a multimedia environment.

This Research in Brief article outlines the evidence regarding the ability of agents to support learning and offers guidance for using agents in multimedia programs in the classroom. Although much of the research on agents has been conducted with college-age students, there is a small body of research involving younger students, and there are some studies involving students with disabilities. The findings from these studies will be highlighted wherever possible. This article includes four main sections: an overview of the effectiveness of agents in supporting learning, a guide for choosing programs with agents, a list of resources, and a more detailed description of the research literature.

Using Agents to Support Learning

Agents increase interest and lessen content difficulty

Students respond positively to agents who speak, interact with the content of the multimedia environment, and demonstrate personality. In fact, students respond more positively to agents that are interactive and speak in a personalized tone (using “you” and “me/we”) than to agents that take the form of static graphics. Researchers and designers have found that when an agent is programmed to demonstrate emotion and to speak in a personalized tone, students’ learning rates and interaction is higher (see review in Moreno, 2005) and students also report that the content is more interesting and less difficult when an agent is available (Lester, Stone, & Stelling, 1999).

Agents can serve as an effective mentor or tutor

The role of tutor is an obvious one for animated agents in multimedia environments. Digital agent/tutors can provide students with supplemental instruction and guided practice on any number of academic skills. Several reading programs use agents to help students build skills and fluency or to prompt the application of comprehension strategies (see the topic paper on Multimedia Materials for Reading). Similarly, agents are embedded in simulation software teaching mathematics and science (see the topic paper on Games and Simulations). Multimedia environments that use agents can support students’ understanding of content area concepts and the relationships among ideas and concepts in a discipline. They can provide the opportunities to practice new skills and knowledge with support and feedback so crucial to learning. In fact, some research shows that programs incorporating agent tutors benefit students with less initial knowledge of the subject area more than they do students with more initial knowledge of the subject area (McNamara & Shaprio, 2005), making them an effective tutor for remediation or initial presentation of a new content area.

Teachable agents can simulate peer tutoring

Teaching others is a powerful way to learn. Peer-assisted learning is beneficial to both the tutor and the tutee (see the work of Fuchs and Fuchs). Reciprocal teaching (Palinscar and Brown, 1984) is an effective form of comprehension strategy instruction that puts students in the role of the teacher. Structuring learning to require self-explanation is also a powerful learning strategy (Graesser, McNamara, & VanLehn, 2005). See Research Support for a review of this literature.

Some designers and researchers are using artificial intelligence to develop agents that can learn from the user. The goal is for students to apply their knowledge about a discipline (science, math, or social studies) in teaching an agent how to solve a problem. If the agent has been taught sufficient skills and knowledge, it can solve the problem; if not, it requests more instruction. Coaching agents can be embedded together with these teachable agents such that students have support in organizing and structuring their teaching.

Teachable agents are the next wave of innovation in this area; currently there are no commercial programs with teachable agents. However, one study demonstrated the promise behind this line of innovation. Researchers found that learning was enhanced when students worked in pairs rather than alone with a program with an agent and had structured explanation opportunities--students were prompted to stop and discuss content with each other (Craig, Driscol, & Ghilson, 2004). In the absence of programs with teachable agents, teachers can realize the benefits of learning-by-teaching by pairing students or arranging small groups and eliciting self-explanation while students are working in multimedia learning environments.

Choosing a Program

Multimedia environments with well-designed agents can provide just-in-time strategy prompts that support students’ learning of content. Research shows that well-designed agents have a positive impact on learning. Most students enjoy agents and find their explanatory advice valuable. When evaluating a potential multimedia program for your classroom, it is important to determine the type of interaction that the agent has with the user as well as the quality of interaction it solicits from users. The best agents ask questions that promote higher-order thinking (i.e., why, what-if, and how questions) and do not unnecessarily define or limit the user’s thinking about a topic (i.e., beware of shallow definitions of key concepts such as “Culture is…” or “History is…”) but instead represent multiple perspectives and prompt reflection. Leverage the benefits of learning-by-teaching by encouraging dialogue and small-group or peer learning.

Summary

Certainly more research into the effectiveness of agents as learning tools is needed, but there are clear indications regarding the possible learning benefits of agents in multimedia environments:

  • Students report increased interest in and less difficulty with the content when using agents;
  • Students show increased learning with agents who provide mentoring on content; and
  • Agents that can be taught and controlled may engage students in higher-order learning.

However, a cautionary note about animated agents is that they can circumvent a student’s active thinking or limit the ways in which a student thinks about a topic by offering a fixed set of choices (Clark & Felton, 2005). This limitation in design could undermine the learning benefits. As with all educational programs, teachers are advised to request demonstration or evaluation copies and review the program with peers and students to ensure that it is well designed.

Research Support

Expand this section to read more about the research behind this article. [Hide]

Agents increase interest and lessen content difficulty

A series of side by side comparison studies with youth and young adults have shown that students’ learning and interaction is enhanced when they work with an agent that is programmed to demonstrate emotions and act in some unpredictable ways and to speak in a personalized tone (using “you” and “me/we”) over other static graphic agents or voice narration only (Moreno, 2005). These findings are reviewed in Moreno (2005); here we provide some examples.

Atkinson (2002) evaluated student interactions with Peedy, a parrot with a personality, in a multimedia program designed to assist students with algebra word problems. Undergraduate students working with Peedy reported less difficulty and had higher post-test scores than students in a control condition working with the same narration but without the Peedy agent. Moreover, students who worked with a talking version of Peedy benefited on post-tests more than students who worked with a Peedy that presented written explanations in a thought bubble. Students in a study by Moundridou and Virvou (2002) also reported experiencing less difficulty and greater enjoyment with a multimedia program featuring an agent that helped students solve algebraic equations rather than the program without the agent.

Similar findings were reported in a study of another animated agent, Herman the Bug. College students reported greater interest and less difficulty with the material (environmental science) when they worked with a Herman who spoke and explained concepts versus a Herman who did not speak but merely showed explanatory text (Moreno, Mayer, Spires, & Lester, 2001). In another study, middle school students worked with multiple versions of Herman the Bug within a multimedia program about botany (Lester, Stone, & Stelling, 1999). Changes in test scores demonstrated that all students learned the material, but students who worked with a speaking Herman reported higher levels of interest and engagement than their counterparts who worked with less interactive versions of Herman.

Van Eck (2006) investigated the impact of a mathematics simulation game with pedagogical agents on middle school students’ attitude toward mathematics. In the game, students helped “aunt” and “uncle” agents fix up their house, applying skills such as calculating equivalent measurements and finding area and perimeter. One hundred twenty-three students were randomly assigned to various conditions differing in terms of the nature of the game (competitive or not) and the agents (available or not). Additionally, a group of students worked with a computer-based program that presented the same math problems without agents with and without competitive conditions. Students who played the game with the agents in a competitive condition demonstrated lower math anxiety levels, and all students working with the agents were able to apply their math skills studied previously in the classroom in the simulated environment. This finding surprised even the researchers, who note that the mere addition of pedagogical agents lowered anxiety levels for these students. They suggest more research to be done in this important area of the confluence of multimedia program design as it may influence mathematics anxiety.

Agents can support learning as mentors or tutors

Several lines of research show that pedagogical, interactive agents that can serve as a tutor can foster learning and inquiry among users. We describe some of this research here.

Researchers at the University of Memphis are designing agents that may increase reading comprehension by prompting students to self-explain their learning – asking themselves why, what-if, and how questions and engaging in an interactive dialogue that reinforces reading strategies. AutoTutor and iSTART are two web-based prototypes that incorporate such agents. Both have been found effective at increasing comprehension of science content text for youth and young adults (Graesser et al, 2003; Graesser, Lu, et al. 2004). AutoTutor uses a human-like head to provide explanation, while iSTART uses a collection of three-dimensional agents, each performing a different function in the training module. Interactive dialogues are incorporated directly into the program through the agent, or developed through peer interaction within student pairs. Peer dialogue around a multimedia learning experience has elsewhere been shown to improve learning for young adults (Craig, Driscoll, & Gholson, 2004; see also the (see the topic paper on Using Multimedia Tools to Help Students Learn Science).

An animated agent is an integral part of the commercial program Thinking Reader® (Tom Snyder Productions, Scholastic). The program embeds strategy instruction into award-winning novels for intermediate and middle school students and is based on research conducted with struggling adolescent readers (Dalton, Pisha, Eagleton, Coyne, & Deysher, 2001). The books are digitized and embedded with multiple supports including human voice narration, text-to-speech, a multimedia glossary, background knowledge links, strategy instruction, and a worklog. Agents prompt the students to “stop and think” (apply reading strategies), and they provide corrective feedback on their performance. The use of these books has been shown to significantly improve reading comprehension of struggling readers compared to traditional reciprocal teaching instruction (Dalton, Pisha, Eagleton, Coyne, & Deysher, 2001).

Bosseler and Massaro (2003) developed a multimedia training environment called the Language Wizard/Player that includes an agent, Baldi, who serves as a speech-language tutor. This agent provides specific feedback on students’ vocabulary and speech production. Baldi’s skin can be made transparent to show the articulatory movements in the mouth and throat. Young children with autism who worked with Baldi increased their vocabulary and generalized their new words to natural settings.

Teachable agents

Teachable agents, as described above, are being studied as the next step in instructional use of agents (Biswas, Leelawong, Schwartz, & Vye, 2005). Watch the Teachable Agent Group at Vanderbilt University for updates and news.

Resources

Betty and Billy

Betty and BillyBetty and Billy, teachable agents created at the Teachable Agents Group at Vanderbilt University. These agents interact with users by asking to be taught concepts necessary to solve problems in the program.

AutoTutor

autotutorThe AutoTutor program and research on the use of teachable agents, created at the Institute for Intelligent Systems at the University of Memphis. The animated agent dialogs with the user as a tutor, eliciting deep reasoning and explanations.

iSTART

iStartThe iSTART (Interactive Strategy Training for Active Reading and Thinking) offers online modules based on research on reading comprehension and self-explanatory teaching. Pedagogical agents provide students with interactive and adaptive training to use active reading strategies.

Peedy the Parrot

Peedy the ParrotPeedy the Parrot is a Microsoft agent that can be incorporated into multimedia programs. Check out how this teachable parrot is being used by a Stanford University group and their explanation of pedagogical agents.

BookBuilder

bookbuilderBookBuilder is an online tool to develop e-books with three different embedded agents that offer strategic comprehension prompts. There is even an agent to help authors create the books. This library will grow as users post the books they have created.

References

Atkinson, R. K. (2002). Optimizing learning from examples using pedagogical agents. Journal of Educational Psychology, 94(2), 416-427.

Biswas, G., Leelawong, K, Schwartz, D., Vye, N., and the Teachable Agents Group at Vanderbilt (2005). Learning by teaching: A new agent paradigm for educational software. Applied Artificial Intelligence, 19, 363-392.

Bosseler, A., & Massaro, D. (2003). Development and evaluation of a computer-animated tutor for vocabulary and language learning in children with autism. Journal of Autism and Developmental Disorders, 33(6), 653-672.

Chi, M. T. H. (200). Self-explaining: The dual processes of generating inference and repairing mental models. In R. Glaser (Ed.), Advances in instructional psychology: Vol 5. Educational design and cognitive science 161-238. Mahwah, NJ: Erlbaum.

Clark, R. E., & Felton, D. F. (2005). Five common but questionable principles of multimedia learning. In R. E. Mayer (Ed.), The Cambridge handbook of multimedia learning, pp. 97-116. New York: Cambridge University Press.

Craig, S. D., Driscoll, D. M. & Gholson, B. (2004). Constructing knowledge from dialog in an intelligent tutoring system: Interactive learning, vicarious learning, and pedagogical agents. Journal of Educational Multimedia and Hypermedia, 13(2), 163-183.

Dalton, B. Pisha, B., Eagleton, M., Coyne, P. & Deysher, S. (2001). Engaging the text: Reciprocal teaching and questioning strategies in a scaffolded learning environment. Final report to the U.S. Department of Education. Peabody, MA: CAST.

Graesser, A. C., McNamara, D.S., & Van Lehn, K. (2005). Scaffolding deep comprehension strategies through Point&Query, AutoTutor, and iSTART. Educational Psychologist, 40(4), 225-234.

Graesser, A. C., Lu, S., Jackson, G. T., Mitchell, H., Ventura, M., Olney, A., et al. (2004). AutoTutor: A tutor with dialogue in natural language. Behavioral Research Methods, Instruments, and Computers, 36, 180-192.

Lester, J. C., Stone, B., & Stelling, G. (1999). Lifelike pedagogical agents for mixed-initiative problem solving in constructivist learning environments. User Modeling and User-Adapted Interaction, 9, 1 – 44.

McNamara, D. S., & Shapiro, A. M. (2005). Multimedia and hypermedia solutions for promoting metacognitive engagement, coherence, and learning. Journal of Educational Computing Research, 33(1): 1-29.

Moreno, R., Mayer, R. E., Spires, A. H., & Lester, J. C. (2001). The case for social agency in computer-based teaching: Do students learn more deeply when they interact with animated pedagogical agents? Cognition and Instruction, 19(2), 177-213.

Moreno, R. (2005). Multimedia learning with animated pedagogical agents. In R. E. Mayer (Ed.), The Cambridge handbook of multimedia learning, 507-523. New York: Cambridge University Press.

Moundridou, M. & Virvou, M. (2002). Evaluating the personal effect of an interface agent in a tutoring system. Journal of Computer Assisted Learning, 18, 253-261.

Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities. Cognition & Instruction, 1(2), 117.

Pedersen, S. & Liu, M. (2002). The effects of modeling expert cognitive strategies during problem-based learning. Paper presented at the Annual Meeting of the American Educational Research Association. Seattle, WA. ERIC Document TM 032 801.

Strangman, N. & Dalton, B. (2005). Improving struggling readers’ comprehension through scaffolded hypertexts and other computer-based literacy programs. In M. C. McKenna, L. D. Labbo, R. D. Kieffer, & D. Reinking (Eds.), International handbook of literacy and technology, Volume II, 75-92. Mahwah, NJ: Lawrence Erlbaum Associates.

Van Eck, R. (2006). The effect of contextual pedagogical advisement and competition on middle-school students’ attitude toward mathematics and mathematics instruction using a computer-based simulation game. Journal of Computers in Mathematics and Science Teaching, 25(2), 165-195.


Site developed by Anilogic - Powered by Dot Org Publisher