Introspection in Problem Solving

Journal of Problem Solving Research Article Introspection in Problem Solving Frank Jäkel1 and Cornell Schreiber2 1 2 Institute of Cognitive Science, Universität Osnabrück Department of Philosophy, Universität Wien Correspondence: Frank Jäkel, Institute of Cognitive Science, Universität Osnabrück. Email: Keywords: introspection, problem solving, metacognition, verbal reports, think aloud Problem solving research has encountered an impasse. Since the seminal work of Newell und Simon (1972) researchers do not seem to have made much theoretical progress (Batchelder and Alexander, 2012; Ohlsson, 2012). In this paper we argue that one factor that is holding back the field is the widespread rejection of introspection among cognitive scientists. We review evidence that introspection improves problem solving performance, sometimes dramatically. Several studies suggest that self-observation, self-monitoring, and self-reflection play a key role in developing problem solving strategies. We argue that studying these introspective processes will require researchers to systematically ask subjects to introspect. However, we document that cognitive science textbooks dismiss introspection and as a consequence introspective methods are not used in problem solving research, even when it would be appropriate. We conclude that research on problem solving would benefit from embracing introspection rather than dismissing it. In contemporary cognitive science introspection is widely regarded as a mysterious and problematic method for uncovering the operations of the mind. Most researchers will typically avoid the conceptual and methodological minefield of introspection. Johnson-Laird (2008, p. 17), in the context of reasoning, expresses an attitude towards introspection that many cognitive scientists share: “To reason is to carry out a mental process. Introspection doesn’t reveal to us how that process works. (If it did, then psychologists would have understood how it worked long ago.) Hence the process is unconscious.” There are good reasons to be skeptical about the usefulness of introspection as a method in many areas of perception and cognition. Many processes are unconscious and hence, by definition, unaccessible for introspection. From this, however, it does not follow that introspection has no role to play in psychological research or that introspection is not an interesting process worthy of study in itself. In the context of problem solving it is important to stress that introspection is, at least sometimes, an essential part of problem solving processes. When dealing with new or hard problems that cannot be solved by standard methods, good problem solvers introspect while they engage with their task: they examine their strategies and representations, they evaluate their progress, and they realize that they are stuck or that they have had an insight. Understanding often seems to depend on some form of conscious and deliberate reflection on one’s own thoughts. These are salient phenomena that we should not ignore. In the literature, the dubious term introspection is often avoided and is replaced with notions such as self-observation, self-monitoring, self-reflection, or metacognition, but also these terms remain mysterious (Brown, 1987). We, thus, believe that research on problem solving, at some point, will have to face squarely the conceptual muddle surrounding introspection, metacognition, and consciousness. Problem solving research has encountered an impasse. We are fixated on problem solving as search and we are spending most of our (re)search time in problem spaces that do not seem to bring us closer to the ill-defined goal of understanding the general principles underlying problem solving. The work by Newell and Simon (1972) was a real breakthrough that allowed us to investigate the mechanics of well-defined search problems. Contrary to the high hopes in the early days of cognitive science no general problem solving theory emerged. While we can formalize the representations and the steps subjects take in a search problem, we do not understand how they choose between different problem representations, different search algorithms, and different heuristics. An accepted formal theory that can describe how representations and heuristics are attained and adapted is still missing. The same is true for a formal theory of insight. Several people have commented on this state of affairs at a recent workshop that inspired this paper (van Rooij et al., 2011). Two recent papers in this journal also offer an analysis of the current impasse in problem solving research (Batchelder and Alexander, 2012; Ohlsson, 2012). We would like to add to these commentaries that, perhaps, the widespread rejection of introspection among cognitive scientists is one of the factors that is holding back the field. A computer scientist who had started to do experiments on problem solving told the first author that they wanted to interview their participants about their problem solving strategies for a newly developed set of problems. However, when docs.lib.purdue.edu/jps http://dx.doi.org/10.7771/1932-6246.1131 20 October 2013 | Volume 6 | Issue 1 F. Jäkel and C. Schreiber Introspection in Problem Solving they consulted with an experimental psychologist they were advised to focus on collecting hard data, like performance measures, and to avoid verbal reports and introspection. We think that most cognitive scientists will react in the same way if a computer scientist who wants advice on running experiments approaches them. But is this good advice? The argument that we would like to put forward in this opinion paper is the following: cognitive scientists have good reasons to be skeptical about introspection as a research method but selfimposed experimental rigor can get in the way of exploratory research. A systematic introspective study is a good way to get experience with a new experimental paradigm and to explore possible problem solving strategies that subjects might be using in the task. In this way, introspection can help develop new hypotheses and ultimately, perhaps, help overcome the impasse the field is facing. Furthermore, in problem solving research the rejection of introspection is particularly harmful because introspection, in the sense of self-observation, is a crucial component of some problem solving processes. Hence, we will argue that research on problem solving could benefit from being more open-minded about introspection, both as a method and as a cognitive phenomenon. To avoid misunderstandings right from the start, we are very aware of the big conceptual and methodological problems that introspection raises and readers should not expect that we offer any solutions or easy-to-follow recommendations for running introspective studies. We do not think that there are new arguments or new suggestions in this paper. After all, the controversies surrounding introspection have been raging (...truncated)