Neural correlates of decision making in quasi-realistic binary decision situations : an EEG and fMRI study

Abstract

A central aspect of daily life is decision making, often under uncertain and ill-defined conditions. This may be why there is huge and continuously growing scientific interest in human decision making. Research of decision making was dominated by mathematical and economic models at first, which led to abstract problems and decision situations, but recent years have seen a growing trend to orient research towards realistic scenarios. The relatively young branch of Naturalistic Decision Making (NDM) placed analyzing real world decision processes of experts at its very core and emphasizes the importance of recognizing the situation to come to a quick yet reliable decision. Uncertainty is a constant aspect of decision making and can be defined as a feeling of doubt that delays or impedes the decision. The aim of this study was to investigate the neural correlates of decision making under certainty and uncertainty in a quasi-realistic setting, that approached the principles of NDM as much as possible in neuroscience. Participants were presented with variations of a simple binary choice. Two types of information from different sources were presented to them: a weather forecast on the probability of rain and a picture of the sky. Based on this information the participants decided whether they would want to take an umbrella with them in a hypothetical context situation. This information varied in its conclusiveness and in its congruency, evoking either certainty or uncertainty in the decision makers, depending on their decision strategies. Each participant undertook the experiment twice; their neural activity was measured once with the EEG and once with fMRI. To account for said individual decision strategies, uncertain and certain decision making during the experiment was identified for each participant individually. This categorization was based on the congruency of the participants' choices. Multiple analyses were conducted to investigate the data, namely: statistical analyses of the behavioral data, including visualization and parameterization of decision strategies; a contrast and conjunction analysis of the fMRI data, and an additional analysis of individual voxel based activation; an ERP analysis, and a frequency analysis of the EEG data. Overall, this study has shown two things. One concerns the neural correlates of decision making: The processing of uncertain and certain decision making seems to have some differences, as addressed in the relevant literature, but seems to be mostly driven by the same fronto-parietal network. The second concerns the nature of quasi-realistic research, which seems to be possible in a laboratory context and offers much needed ecological validity, but requires attention to detail and individual variations. To properly use quasi-realistic designs, the standard procedures of analysis have to be adapted. During the analysis of the data, a number of limitations of the experimental design became apparent, mostly stemming from an initial underestimation of the participant's individually varying decision strategies. Many possible improvements and alternative approaches to design and analysis could be devised based on these realizations, which may be beneficial to future studies in this general field.