Is Insight Always the Same? An fMRI Study of Insight

An flVIRI Study of Insight Using Compound Remote Associate Problems Edward A. Cranford (eac53@msstate.edu) Jarrod Moss (jarrod.moss@msstate.edu) Department of Psychology, Mississippi State University Mississippi State, MS 39762 USA Abstract Prior neuroimaging studies of insight problem solving using Compound Remote Associate (CRA) problems provide consistent results. However, in a prior study (Cranford & Moss, 2010), we found that participants derive solutions by insight in at least two different ways. In the present study, we attempted to extend upon prior studies by dividing insight solutions into two categories: immediate and delayed. The results show a large difference between the pattern of activation for immediate-insight solutions and delayed-insight solutions. Future research may benefit from distinguishing between types of insight in CRA problem solving. Keywords: Insight; Problem Solving; Restructuring; fl\/IRI Introduction Problem solving as a process is what enables humans to discover solutions to even the most difficult of problems. One way to obtain a solution is through a ‘search’ process. The solver engages in a structured exploration of possible solution paths in order to find a solution. Sometimes the solution appears suddenly, without warning, and is termed insightful. Insight solutions often appear from nowhere and solvers experience an affective response of suddenness and surprise (Aha! experience), sometimes resulting after an impasse; insight solutions are obtained through processes known as restructuring, whereby an incorrect representation of the problem is changed, leading to the access of an insightful, correct representation of the problem (Bowden, & Jung—Beeman, 2007; Ohlsson, 1992; Schooler, Fallshore, & Fiore, 1995). Solutions obtained through search do not generate the ‘Aha’ experience, and the solution idea seems to be a continuation of previously generated ideas. The key components of insight are often described as impasse, restructuring, and ‘Aha!’. However, the process is still not fully understood. Neuroimaging techniques and more consistent experimental tasks have the potential to further our understanding of the neural correlates of insight. Utilizing neuroimaging methods such as functional magnetic resonance imaging (fMRI) limits the type of task that can be performed. Classic insight problems usually take a long time to solve and are often different than noninsight problems. Therefore, the Compound Remote Associate (CRA) task was developed by Bowden and Jung—Beeman (2003). The CRA problem consists of three words presented to the solver. The solver must come up with a single fourth word that can be combined with each of the other three words to form new compound words or common phrases. For example, if three words—tree, sauce, and big—are presented, the solution is apple. CRA problems can be solved quickly so many problems can be presented in one fMRI scanning session. Also, CRA problems can be solved by insight or by noninsight, search processes (i.e., generate- and-test or trial-and-error), and individual problems can be solved with insight regardless of learning effects over multiple trials (Bowden & Jung—Beeman, 2007). Prior fl\/IRI studies using the CRA problem to investigate the neural correlates of insight yield somewhat consistent results and offer compelling theoretical arguments for the activation seen in these brain regions. Jung—Beeman and colleagues (2004) expanded on their theory that course semantic coding occurs in the right hemisphere (RH) while fine coding occurs in the left hemisphere (LH) and found that an area in the RH anterior superior temporal gyrus (RH- aSTG) was more active at solution for insight than noninsight solutions. Activity in this region creates broad associations in memory so that seemingly disparate concepts converge on a solution and suddenly emerge into consciousness as an insight. Subramaniam, Kounios, Parrish, and Jung—Beeman (2008) reported a similar region in their results (RH middle temporal gyrus; MTG). Both studies reported activity in the anterior cingulate cortex (ACC). The function of the ACC in insight problem solving is to monitor for competing responses for attention, as is its role in the cognitive control network (i.e., Cole & Schneider, 2007). Other ‘insight’ areas noted in the two studies are the posterior cingulate cortex (PCC), parahippocampus (PH), right superior frontal gyrus (SFG), and right inferior parietal lobe (IPL). Even prior to seeing a problem, certain brain states have been found to predict future solution by insight (Kounios et al., 2006). Kounios and colleagues observed activity in the ACC, PCC, and bilateral middle/superior temporal gyrus (M/STG) prior to seeing a problem. These same areas were observed by Subramaniam and colleagues (2008). The ACC may be active due to an increased readiness to apply cognitive control in order to suppress thoughts, initially select a solution space, and, if needed, to switch attention. The M/STG may be active due to a preparation for semantic activation and retrieval of associations. Finally, the PCC may simply reflect differences in attentional demands between preparation periods preceding insight and those preceding noninsight solutions. Although the results from these three studies are informative, they often report many regions of activation that do not overlap across studies using very similar methodology. Clearly some regions such as ACC and right MTG are consistently active, but one possible explanation for areas that do not overlap across 3558