Obstacles in Investigating the Role of Restructuring in Insightful Problem Solving

Abstract

In the present article, we articulate three assumptions underlying theories proposing that restructuring processes play a key role in insightful problem solving: representational difficulty, representational change, and discontinuity in solution processes. We argue that these assumptions need empirical validation to justify the proposition of restructuring mechanisms that are unique from those involved in classic information-processing theories of problem solving. To this end, we review some theoretical and methodological obstacles that are inherent in the investigation of the existence and nature of restructuring processes. We then offer some recommendations on how to overcome or avoid these obstacles in future studies. Finally, we discuss some questions to help motivate new research.

Investigating Insight as Sudden Learning

“…Thus, these results replicate Fleck and Weisberg (2004) by showing not all solvers come to impasse before solving classic insight problems. Furthermore, they highlight the importance of taking a process-oriented approach to studying insight phenomenon instead of a problem-oriented approach (Ash, Cushen, & Wiley, 2009;Ellen, 1982;Fleck & Weisberg, 2004). Second, people were regularly able to overcome impasse and go on to solve the puzzle problems.…”

“…For the purposes of studying insight learning, it is important to have problems that people can solve without hints or experimenter interventions. Hints and training interventions are useful methods for studying fixation and other sources of problem difficulty (Ash, Cushen, & Wiley, 2009). However, according to Gestalt insight learning theory, there is a need to differentiate between new information obtained from the external environment, and new information resulting from internal problem solving processes.…”

“…There are many reasons why one may expect faster solution times on a second problem solving attempt and only some may reflect insight learning. Other alternatives include effects of re-exposure such as reducing the effort needed to re-read the problem, reduced time in formulating a solution strategy, or excluding branches of the search space that were previously found to be ineffective (Ash, Wiley, & Cushen, 2009). Ash and Wiley (2008) also provided some evidence supporting the Gestalt notion of insight learning.…”

“…In other words, there is no such thing as an "insight problem, " and one cannot assume that any puzzle will always be solved via the attempt-impasse-solution pattern. Classic insight problems are laboratory stimuli that have been designed by researchers to be highly likely to lead to solvers adopting inappropriate initial solution strategies (Ash, Wiley, & Cushen, 2009) and to lead to fixation or impasse. However, despite their design, there is no guarantee that solvers will actually experience impasse during solution (see Cranford & Moss, 2012).…”

“…Therefore, current ACT-R learning mechanisms would not be able to accommodate these types of processes. Developing computational models that can account for creative leaps, novel discoveries, and instances of one-trial learning still represents an interesting challenge for association-based models (Ash, Cushen & Wiley, 2009;Langley & Jones, 1988).…”