Teaching Yourself by Teaching Others
Becoming Sir Allan Sugar’s protégé has become its own programme
“While we teach, we learn” – Seneca (Roman Philosopher)
Contents
1: What is the protégé effect?
2a: Main Paper
2b: Experiment 1
2c: Experiment 2
3: Issues
4: Future Research and Conclusions
5: References
What is the protégé effect?
Have you any younger siblings? If you do, then congratulations. Kristensen and Bjerkedal (2007) and Bjerkedal, Kristensen, Skjeret and Brevik (2007) in their studies found that younger siblings have a lower IQ than older siblings. Siblings have the same blood, the same genes and the only discernible difference is their ages. What these studies suggest then is that the older siblings have been teaching their younger counterparts about the world, about life, about the do’s and don’ts of society and the older brothers and sisters have actually become smarter for it.
This effect of teaching others, specifically others younger in age, has been labelled the “the protégé effect”. Put simply, when tasked with the education of others, individuals strive to understand the material to the best of their ability, so that they are able to pass on their knowledge effectively to others. As we will see, these teachers achieve better marks than those who do not need to educate others.
Main Paper
Teachable Agents and the Protégé Effect: Increasing the Effort Towards Learning
(Chase, Chin, Oppezzo & Schwartz 2009)
The two experiments contained within this paper both show that when learning for others, individual education is enhanced when simply learning for yourself. With technology increasingly being introduced into the classroom, software had been produced to act as the learners. These “teachable agents” (TA) come in all shapes and sizes (Schwartz et al 2007), but the specific TA that Chase et al (2009) use is called Betty’s Brain.
Betty’s Brain is taught information by the student inputting a pathway to certain facts which originate from a basic concept. For example; dieting decreases fat stores which decreases body weight, while bodybuilding increases muscle mass which increases body weight. These interlinked pathways become the basis for the TA’s mind and logical thinking. The TA will use the pathways given to it to answer any questions the student asks, hence answering correctly only if the student has input the accurate information.
Above is an example of the knowledge structure that the student creates for the TA in order to answer questions. The two experiments in this study utilise the Betty’s Brain TA to show that the protégé effect, (even with technology acting as the learner) can occur.
Experiment 1
Investigating the levels effort put forth by the students to learn material was the key question in experiment. While one subject group learned for their own benefit, the other group learned in order to educate their TA. Sixty-two 8th graders (13-15 years old) were used as participants for the study that was spread out over three days.
The first day involved the subjects becoming accustomed to using the Betty’s Brain software and administering the source material the experimenter would use to test the subjects’ knowledge. The subjects in both conditions (teaching TA and teaching themselves) both used the software to create the knowledge structures to enhance their learning, the only difference being that in the TA condition, the learner would eventually answer questions based on the pathways the teachers had made.
The final act on day 1 was to get the participants acclimatized to the test program of the study. Using a “game show” set up, the subjects could ask and answer questions on the material given to them against another subjects, all the while gambling points on their ability to get the right answer. Both the TA and self conditions completed a round of this “game show” using knowledge they had procured themselves (i.e. the TA did not answer any questions on day 1).
On day 2, subjects completed another round of the “game show”, however this time, in the TA condition, the answers to any questions were given by the TA itself, not the subject that had created the knowledge structure. Afterwards the subjects were given lessons in creating better knowledge structures and time, before their next round of playing the “game show”, which could be used to study or amend the pathways.
The next round was once again played against another human opponent with either subject answering the questions (self condition) or the TA giving answers (TA condition). Next the participants were given more time to either revise or play.
Day 3 consisted of a simple test that each subject completed regardless of condition. This test differed from the “game show” environment as it involved a simple written exam on the material intrinsic to the study.
The results of the experiment supported the protégé effect. Subjects that had to teach the content to the TA worked harder and longer (across both days 1 and 2) on updating the knowledge structures and pathways, while also reading the source material for longer periods than the participants in the self condition. These results then translated into academic success for the TA subjects as they scored better than those in the self condition.
The results also illustrated remarkable improvements in subjects who were not considered top students in their classes. These “poor marks” students recorded similar results in the day 3 written test than the students in the self condition who were considered “top marks” students.
Experiment 2
While experiment 1 looked at the protégé effect itself, experiment two was designed to look at why the effect occurred in the first place. Chase et al (2009) specifically looked at how the subjects viewed the TA’s in terms of responsibility and how they would adapt to failure.
This time, twenty-four 5th grade pupils (10-11 years old) participated in experiment 2, which was designed very similarly to experiment 1. Again there were two conditions (TA and self), each participating in the “game show” with the TA (in the TA condition) and the subject (in the self condition) answering the questions set to them. Differences included the “game show” being completed alone, i.e. not in competition with other subjects.
The main body of the experiment was comprised of four sections. First, the participants were given the same revision material as experiment 1 and instructed to use the Betty’s Brain software to construct a similar knowledge structure as before. The TA condition had to use the pathways to educate the TA, while the self condition used the pathways to teach themselves. The time it took to make the structures was recorded.
The subjects then took part in the “game show” while verbally communicating their thought processes during the questions. Then the participants were given time to go over the material and pathways, knowing that the next section of the experiment involved more difficult questions than previously asked. The subjects themselves decided when they were ready to complete these next set of questions.
The final section did include a test, but the experimenters intentionally did not administer the second round of the “game show”, instead opting to ask the subjects questions on the source material verbally, with the participants also responding verbally.
The results of experiment 2 mirrored those of experiment 1, with those who were teaching the TA taking more time to revise the source material and update the knowledge structure. However while the protégé effect was again supported by the results, evidence from the behaviors of the subjects in the TA condition suggested possible causes for the effect.
Chase et al (2009) note that one such subject actually named the TA, while others divided the responsibility of giving correct answers between themselves and the TA. This behavior suggests the subjects connected with the TA, truly viewing the software as a protégé. In terms of failure recognition, the TA subjects felt remorse at any mistakes made by their TA and were enthusiastic to correct any such mistakes. This kind of behavior was not shown as frequently in the self condition.
Issues
The protégé effect is a relatively novel way of education, and because of this, there is a severe gap in the literature in this field of research. Chase et al (2009) have shown that a protégé effect can occur, but a limitation of the research (if you want to view it as a limitation), is the lack of human protégés. While the TAs make fine substitutions, the effect might be more pronounced if the teacher is assigned to educate a protégé with a real face and real emotions.
Conversely however, in the age of economic setbacks, Chase et al (2009) have demonstrated that software can replace any live human interaction, allowing the students to teach a very simple computer program. This means that any class size can teach their own software and have their own learning enhanced as a result. The other potential negative with teaching a human protégé is the education of the protégé themselves. While the teacher may be working hard to understand the data, the protégé’s understanding may or may not increase, just because their teacher has put in the work.
One of the main points to take away from this study however is the fact that Chase et al (2009) have shown that students who have not achieved the highest grades have responded extremely positively to this type of learning. Introducing the protégé effect into schools with low exam result statistics may increase their overall pass rates and provide a brighter future to those who have trouble motivating themselves.
Future Research and Conclusions
The protégé effect, through the Chase et al (2009) study, has been highlighted as a legitimate learning technique, one that that can increase academic ability in pupils, even those who have failed to achieve high marks. Through the use of technology (which is increasingly present in schools and is in abundance at university and college levels), motivation to learn through teaching others has a very high ceiling in terms of overall potential to increase grades. It could inspire a whole new generation into teaching as a profession.
The most obvious area to venture in with future research must be the protégé effect’s applicability to human learners rather than computer based TA’s. However as Paul (2011) describes, in her article in TIME magazine, American universities have already started this initiative, with the University of Pennsylvania introducing a scheme that will send undergraduates to different high schools and teach the pupils there computer science. Then in turn, these pupils will be tasked with teaching middle high school pupils what they themselves have already been taught.
Example of a this type of system used at the University of Southern California
This “cascading mentoring” system takes the protégé effect and expands it to a massive scale. Chase et al (2009) have through their findings lent support to this type of learning mechanism, albeit with software. If programs like the one UPenn are running take off on a national/global scale, the protégé effect will encourage youngsters to take a motivated interest in their learning and help them to achieve better marks. If this occurs, who is to say the future will not look very bright indeed.
References
Bjerkedal, T., Kristensen, P., Skjeret, G.A., & Brevik, J.I. (2007). Intelligence test scores and birth order among young Norwegian men (conscripts) analyzed within and between families. Intelligence, 35, 503-514.
Chase, C.C., Chin, D.B., Oppezzo, M.A., & Schwartz, D.L. (2009). Teachable agents and the protégé effect: increasing the effort towards learning. The Journal of Science Education and Technology, 18, 334-352.
Kristensen, P., & Bjerkedal, T. (2007). Explaining the relation between birth order and intelligence. Science, 22, 1711-1712.
Paul, A.M. (2011). The protégé effect. TIME. Accessed via the World Wide Web on 21st March 2012 at: http://ideas.time.com/2011/11/30/the-protege-effect/
Schwartz, D.L., Blair, K.P., Biswas, G., Leelawong, K., & Davis, J. (2007). Animations of thought: interactivity in the teachable agents paradigm. In: R. Lowe, & W. Schnotz (Eds.), Learning with animation: research and implications for design (pp114-140). Cambridge: Cambridge University Press.