טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentSoria Adi
SubjectPeer Performance as Harmful Information for Meta-
Reasoning
DepartmentDepartment of Industrial Engineering and Management
Supervisor Professor Rakefet Ackerman
Full Thesis textFull thesis text - English Version


Abstract

Metacognitive research deals with factors that affect people’s judgments regarding their own performance while performing cognitive tasks, such as learning and problem solving. Decision making literature suggests that people are often affected by previously considered numerical information, provided externally, that serves as an anchor when making numeric judgments. An anchor for one's own judgments could be information regarding peers’ performance which may serve as a hint for the difficulty of the task. Yet, such social information has been rarely studied from a metacognitive perspective.

Existing research has led researchers to expect accurate information regarding peers' success to improve the accuracy of metacognitive judgments. However, this claim has not been examined empirically. Moreover, only a few studies have examined the influence of framing information in terms of success or failure on metacognitive judgments. In particular, no previous research investigated the effects of either anchoring or framing on metacognitive processes involved in problem solving.

Meta-Reasoning is the research domain which deals with metacognitive aspects of problem solving and reasoning. The aim of the present study was to examine the effects of information regarding peers’ performance on monitoring and control processes in a problem-solving task. The research was guided by three hypotheses. First, I hypothesized that when providing participants with a congruent anchor?reflecting actual success of peers?judgements regarding solving would be more accurate, compared to no anchor. Second, I hypothesized that a failure-framing would improve monitoring accuracy compared to success-framing, due to increased attention to mistakes. Third, I anticipated these effects to depend on the congruency and framing of the information presented, so that information congruent with actual peer performance would result in overconfidence bias, while incongruent information, about lower performance among peers, would attenuate this bias. Moreover, particularly discouraging information regarding 65% of failure (incongruent information) would result in the smallest bias.

To examine these hypotheses, two experiments were conducted by using verbal-analogy problems in large difficulty variability. Participants rated confidence in each solution. In Experiment 1 I examined the effects of congruent anchor with failure-framing (45% failure of peers) on confidence accuracy. To motivate participants, an incentive structure was administered for correct and incorrect solutions. Surprisingly, the congruent and failure-framed anchoring was found to harm confidence accuracy. In Experiment 2 I further examined the interactive effects of anchoring (congruent versus incongruent information) and framing (failure versus success). Experiment 2 replicated the results of Experiment 1. As hypothesized, monitoring accuracy was enhanced for failure-framing compared to success-framing. Additionally, two opposite biases were found: an 'optimistic' bias, which reflected overconfidence bias together with low performance, and a 'pessimistic' bias, which reflected under-confidence together with high performance. This study provides a first insight into the effects social cues on Meta-Reasoning processes. It appears that anchoring effects are not necessarily productive, and that framing is an important factor to consider when relying on information provided for problem solving tasks. Teachers should be aware of social information which presented to their students, since it may have negative impact on students' performance and self-evaluations.