Productive Failure
Productive failure (Kapur, 2008; Kapur & Bielaczyk, 2012) is a concept that comes out of research involving students working in groups on ill- or openly-structured problems prior to receiving targeted instruction. According to this research, although students typically generate poor quality solutions when they have little guidance or support to utilize, the collaborative processes involved in generating and evaluating each other’s ideas, and analyzing and critiquing potential solutions, yields greater gains for flexible long-term learning than when groups of students work on highly-structured, or well-defined problems. Indeed, research has shown that when students who work on ill-structured problems prior to instruction out-perform students who receive guided or direct instruction first on subsequent assessments that are either highly-structured or open-ended.
The idea of productive failure is thus to delay providing structure or feedback to students during problem solving activities, so as to allow learners the chance to generate their own understandings, representations, and explanations – even incomplete or inaccurate ones. In so doing, participation in the problem solving activity prepares students to encode key conceptual features during subsequent instruction (say, in the form of a short lecture or video).
Cognitive- and socio-cognitive theories of learning support productive failure activities as an effective means of developing robust knowledge structures, but from an instructional standpoint, these activities do require some careful attention to context and student participation. Nevertheless, research on productive failure demonstrates the benefits of this somewhat counter-intuitive process.
Theoretical Background
Problems that are highly-structured, or provide explicit guidance to students, can have the effect of limiting the “problem space” in which students grapple with knowledge construction processes. This is because highly-structured problems tend to narrow students’ focus to a particular or pre-defined solution path, procedure, or conceptual feature of the problem. While providing such guidance may be efficient for enhancing student performance (i.e., reducing failure) in the short-term, expanding the problem space in such a way that requires students to (1) determine how to define a problem, (2) generate their own analytic methods or ways to represent the problem, and (3) explain or critique those methods and representations, creates “desirable difficulties” (Schmidt & Bjorn, 1992) that prepare students to construct knowledge that is more flexible and readily adaptable to novel problem-solving situations (Schwartz & Martin, 2004).
Such effects may, partially, be explained by research on what is known as impasse-driven learning (Van Lehn, 1999; Van Lehn, Siler, Murray, Yamauchi, & Baggett, 2003). Studies of impasse-driven learning have found that students show greater learning gains when they receive guidance or instruction after reaching an “impasse” – that is, when a student can no longer make adequate progress towards a problem solution. The experience of reaching an impasse, followed by instruction that targets learners’ attention to key features of a canonical or conventional solution, helps students discern critical differences between their “failed” solutions and the canonical ones. It is this focus on differences, rather than similarities between student-generated and expert-like solutions, that researchers believe may contribute to flexible and long-term learning (Kapur, 2008; Schwartz & Bransford, 1998). In essence, the effort students put into generating their own solutions, representations, or methods gives meaning to key conceptual features of expert-like or canonical solutions. By definition, highly-structured problems reduce the need for learners to construct such meaning on their own, providing a weaker foundation for long-term or flexible learning.
How does it work?
Well-designed productive failure activities accomplish four key learning principles:
- They activate and differentiate students’ prior knowledge in relation to target concepts,
- They focus students’ attention to key features of the target concepts,
- They require students to explain and elaborate on those key conceptual features, and
- They help students organize those key features into the target concepts.
In terms of actual practice, this means creating problem-solving contexts that (a) are challenging to students, but not beyond their capacity or prior knowledge, (b) allow for students to generate multiple ways of analyzing, representing, or modeling the problem, (c) encourage students to explain and elaborate on their ideas, and (d) compare and contrast the methods, representations, and models that students create with those that are more expert-like or canonical to the domain.
Naturally, this means that instructors must have a sense of the knowledge that students bring with them to the problem space, and be able frame problems in ways that are engaging and interesting to students. Students should have the underlying knowledge needed to solve certain aspects of a problem, but should not be able to apply that knowledge directly to solve the entire problem. This also means that productive failure activities must be sufficiently complex that they require students to consider multiple possible solution paths.
Equally important, learning scientists suggest that productive learning activities should take advantage of collaborative processes. Collaboration has been shown to encourage sharing, critique, elaboration, and evaluation in problem solving activities (Bereiter & Scardamalia, 2014) – all of which are central to the productive aspects of productive failure, and all of which have been shown to support student learning. Further, instructors should encourage groups of students to explore various ways of solving a problem, and to generate multiple possible solutions. Notably, this does not mean necessarily finding the correct answer – an outcome that should be underemphasized in this setting. Rather, the intended result of the activity should be for groups to generate a wide diversity possible ways to model, represent, or analyze the problem.
Finally, students need to attend to the key features of the target concepts, as well as organize those features into their understanding of canonical solutions. Whole class discussions that focus on the affordances and constraints of student-generated methods, models, or representations provide such an opportunity. Students may present their work to the class, guided by the instructor’s questions or requests for clarification with regards to key conceptual features. Students may also prompt each other for similar explanations or clarification through peer reviews. The instructor may then share the canonical ways to solving or representing the problem with the class, in so doing, comparing and contrasting between student-generated solutions and the canonical one.
See also:
References:
Bereiter C., Scardamalia M. (2014) Knowledge Building and Knowledge Creation: One Concept, Two Hills to Climb. In: Tan S., So H., Yeo J. (eds) Knowledge Creation in Education. Education Innovation Series. Springer, Singapore
Holmes, N., Day, J., Park, A., Bonn, D., & Roll, I. (2014). Making the failure more productive: Scaffolding the invention process to improve inquiry behaviors and outcomes in invention activities. Instructional Science., 42(4), 523-538.
Kapur, M. (2008). Productive Failure. Cognition and Instruction., 26(3), 379-424.
Kapur, M., & Bielaczyc, K. (2012). Designing for Productive Failure. The Journal of the Learning Sciences., 21(1), 45-83.
Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3, 207-217.
Schwartz, D., & Bransford, J. (1998). A Time For Telling. Cognition and Instruction., 16(4), 475-5223.
Schwartz, D., & Martin, T. (2004). Inventing to Prepare for Future Learning: The Hidden Efficiency of Encouraging Original Student Production in Statistics Instruction. Cognition and Instruction., 22(2), 129-184.
VanLehn, K. (1999). Rule-Learning Events in the Acquisition of a Complex Skill: An Evaluation of Cascade. The Journal of the Learning Sciences., 8(1), 71-125.
VanLehn, K., Siler, S., Murray, C., Yamauchi, T., & Baggett, W. (2003). Why Do Only Some Events Cause Learning During Human Tutoring? Cognition and Instruction., 21(3), 209-249.