Effective decision making across contexts is essential for successful navigation of a complex world. Decision making styles vary greatly between individuals, and with context and state. Theyy are altered in a range of psychopathologies. However, the development of a systematic understanding of variation in decision making has been hampered by variable and limited characterization of decision-making parameters, small samples in most individual studies, and a lack of robust normative data. Computational models of decision making, such as the drift-diffusion model (DDM), can be fitted to behavioral data from individual participants to reveal variation in underlying processes. Parameters of such computational models may serve as ?cleaner? measures of processes of interest than unmodeled behavioral data, or self-report measures. They can also be used as correlates of neural activation patterns during decision making. The validation of computational models and the identification of model parameters that correlate robustly with brain activation sets the stage for parallel studies in animals, in which causal relations can be more readily probed. We propose to conduct a large-scale online data collection of two DDM-compatible tasks, which probe perceptual and value-based decision-making processes. We will use best practices developed for Amazon Mechanical Turk (MTurk) to generate a reference distribution of DDM parameters. Since DDM relies on precise measurements of reaction time, it is critically important to establish validity of online instruments, which we propose to do by collecting parallel in-lab and online data in an initial medium size sample; this will permit robust hypothesis-driven and exploratory analyses, as well as allowing us to optimize and validate online data collection for the collection of online-only data in a larger sample (N = 500). If successful, this validation will allow large-scale behavioral data collection powered to detect small to medium effect size associations and will provide a reference distribution and cutoff levels for extreme cases of DDM parameters. We will investigate relations between continuous measures of selected clinical tendencies in general population and DDM parameters in a large sample. We will also investigate relations between DDM parameters and individual approach and avoidance tendencies, which are hypothesized to underlie individual variations in decision making styles and have been translationally validated. This will generate new hypotheses as to the role of decision-making abnormalities in psychopathology. The use of computational modeling approaches like DDM and large general population samples may be more powerful for the elucidation of such relationships than simple correlations of behavioral measures with symptomatology. This approach is consistent with the RDoC framework and can be extended in future work to transdiagnostic and translational studies of psychopathology.
Decision-making styles vary greatly between individuals and with context and state, and have been shown to be altered in psychopathology; the development of a systematic understanding of variation in decision making has been hampered by variable and limited characterization of decision-making parameters, small samples in most studies, and a lack of robust normative data. We propose to conduct large-scale online data collection to characterize a reference distribution of the parameters of an elegant and robust framework for the analysis of decision making, the Drift Diffusion Model (DDM) of choice; we will employ DDM-compatible versions of tasks that probe both perceptual and value-based decision-making process. The validation of computational models and the identification of model parameters that correlate robustly with brain activation sets the stage for parallel studies in animals, in which causal relations can be more readily probed.
