Testing the validity of conflict drift-diffusion models for use in estimating cognitive processes: A parameter-recovery study

Abstract

Researchers and clinicians are interested in estimating individual differences in the ability to process conflicting information. Conflict processing is typically assessed by comparing behavioral measures like RTs or error rates from conflict tasks. However, these measures are hard to interpret because they can be influenced by additional processes like response caution or bias. This limitation can be circumvented by employing cognitive models to decompose behavioral data into components of underlying decision processes, providing better specificity for investigating individual differences. A new class of drift-diffusion models has been developed for conflict tasks, presenting a potential tool to improve analysis of individual differences in conflict processing. However, measures from these models have not been validated for use in experiments with limited data collection. The present study assessed the validity of these models with a parameter-recovery study to determine whether and under what circumstances the models provide valid measures of cognitive processing. Three models were tested: the dual-stage two-phase model (Hübner, Steinhauser, & Lehle, Psychological Review, 117(3), 759-784, 2010), the shrinking spotlight model (White, Ratcliff, & Starns, Cognitive Psychology, 63(4), 210-238, 2011), and the diffusion model for conflict tasks (Ulrich, Schröter, Leuthold, & Birngruber, Cogntive Psychology, 78, 148-174, 2015). The validity of the model parameters was assessed using different methods of fitting the data and different numbers of trials. The results show that each model has limitations in recovering valid parameters, but they can be mitigated by adding constraints to the model. Practical recommendations are provided for when and how each model can be used to analyze data and provide measures of processing in conflict tasks.

Cueing Effects in the Attentional Network Test: a Spotlight Diffusion Model Analysis

“…Because of the low number of errors, only the median RT for incorrect responses was used in the model fitting. A recent simulation-recovery study on the SSP showed that this approach provides accurate estimation of all of the parameters with the exception of the spotlight width (sd a ) and shrinking rate (r d ) (White, Servant, and Logan 2018). The simulation study showed that these two parameters cannot be estimated independently because they trade off with each other: a wide spotlight that shrinks rapidly can produce similar interference as a narrow spotlight that shrinks slowly.…”

“…Based on a recent simulation study that did not observe such correlations from synthetic data, we believe that these negative correlations stem from the participants and data rather than trade-offs in the model (see White, Servant, and Logan 2018). Note that the positive within-participant correlations were not guaranteed because the model was fitted separately to each condition; thus, the strong correlations across cueing conditions for each parameter suggest the model fitting captured meaningful trends in the data.…”

“…This means that the SSP measure of attentional control, interference time (see "Results"), is driven by estimates of the shrinking rate. Readers who are interested in additional details about implementing and fitting the SSP are directed to Grange (2016), White et al (2011), or White et al (2018.…”

“…The starting point was fixed to be halfway between the top and bottom boundaries because right and left responses were collapsed for the analysis. The spotlight width was fixed at a standard value of 1.8, which distributes roughly equal attention to each item in the display at stimulus onset, because this parameter trades off with the shrinking rate (see White, Servant, and Logan 2018). An exploratory analysis was performed in which both the spotlight width and the shrinking rate were allowed to vary, and the results were nearly identical to the fits with the spotlight width fixed (due to the parameter tradeoff).…”

“…Other models exist that accomplish the same goal for flanker stimuli, including the dualstage two-phase model (Hübner et al 2010) and the diffusion model for conflict tasks (Ulrich et al 2015). However, this study focuses on the spotlight model over the others because (1) it was designed specifically for visual attention tasks with flanker stimuli, and (2) it has been shown to provide better parameter estimation from behavioral data (White, Servant, and Logan 2018), which is elaborated upon later. Importantly for the present study, the SSP successfully accounts for effects of manipulations that are meant to target specific components of the model.…”