Methods for investigating the neural components of insight

Abstract

The authors describe how they have used visual-hemifield and event-related neuroimaging approaches to study their theory specifying some of the neural components of insight. A set of problems developed by the authors, and the use of solvers’ self reports of insight, are presented to argue that advances in our understanding of insight are being unnecessarily stifled by over reliance on traditional insight problems and a widespread failure to determine whether insight has occurred on a solution-by-solution basis.

Introduction

Since the beginning of psychology as a science more than a century ago, one of the central areas of interest has been in understanding problem solving (e.g., [1], [2]). A distinction has been made between step-by-step analytic problem solving and a distinct “all-at-once” process called insight, in which the solution to a problem appears in consciousness suddenly and without conscious access to the solving process [3], [4]. Instances of insight-like phenomena can also be found in perception (e.g., sudden recognition of an object in a blurred or ambiguous picture; [5]) and language comprehension (e.g., sudden comprehension of the meaning of a joke or metaphor, [6]). Thus, insight is not limited to problem solving but appears to be a general phenomenon or mechanism of cognition.

When discussing insight researchers have used terms such as representational change, restructuring, and chunk decomposition to describe the cognitive processes necessary for insight solutions [7], [8], [9], [10], [11], [12], [13], [14], [15]. However, until very recently almost no attempts had been made to link these descriptions to the underlying brain structures and neural activity.

Advances in theories of brain function and neuroimaging techniques have made it possible to predict and view the neural activity associated with differences between insight and non-insight problem solving. For example, Functional Magnetic Resonance Imaging (fMRI) allows the entire brain to be imaged repeatedly while a person is engaged in a task, providing a view of all regions involved in the task. Thus, by making it possible to view differences in the neural activity underlying insight and non-insight solutions, the use of fMRI provides a significant addition to the information that can be gained with traditional behavioral techniques. However, many fundamental issues remain unresolved [16], and the cognitive and neural mechanisms by which people solve problems with insight remain under-specified.

It is our goal to move toward demystifying insight by revealing the neural components of problem solving. To this end we have used both behavioral and neuroimaging methods (see [17], [18], [19], [20], [21]). Other researchers are also using neuroimaging to investigate the neural components of insight (e.g., [22], [23], [24]).

Of course, the methods that researchers use should always be determined by the questions they are attempting to answer. We began with the hypothesis that the cerebral hemispheres process information differently. Specifically, both hemispheres process all types of information but the right hemisphere (RH) engages in coarser coding of information while the left hemisphere (LH) engages in finer coding of the same information. This hypothesis predicts differences in the kind of information each hemisphere should have available, therefore we began our research by using a visual-hemifield presentation paradigm, which can reveal hemispheric differences. We also have hypotheses that are more precise in their predictions about where, when, and how the brain produces solutions with insight, therefore we have used fMRI, which can provide precise information regarding the locations of activation, and electro-encephalography (EEG), which can provide precise information regarding the timing of cognitive events.

In practical terms the methods we have used can be placed into two categories: inexpensive, easy, but inexact, and expensive, complicated, but more precise. Visual-hemifield presentation is an inexpensive and relatively easy approach that can provide gross information regarding whether critical solution information is more available in one hemisphere than the other. In contrast, fMRI and EEG are expensive and complex methods that can provide far more detailed information about the location and timing of neural activity.

The aim of this paper is to provide a description of the methods we have used at a level that will be appropriate for researchers just venturing into studies of the neural mechanisms of insight and creativity. We will attempt to provide enough detail so that common and easily made mistakes can be avoided. We also will provide more details for visual-hemifield than for neuroimaging methods because individual researchers with limited resources can carry out visual-hemifield experiments. In contrast, neuroimaging experiments require a team of researchers and extensive resources therefore it is not necessary for any individual to possess all of the required technical knowledge.

We will begin by providing some background so that the rationale behind our use of these methods is clearer.