Finding the right fit: A comparison of process assumptions underlying popular drift-diffusion models.

Abstract

Recent research makes increasing use of eye-tracking methodologies to generate and test process models. Overall, such research suggests that attention, generally indexed by fixations (gaze duration), plays a critical role in the construction of preference, although the methods used to support this supposition differ substantially. In 2 studies we empirically test prototypical versions of prominent processing assumptions against 1 another and several base models. We find that general evidence accumulation processes provide a good fit to the data. An accumulation process that assumes leakage and temporal variability in evidence weighting (i.e., a primacy effect) fits the aggregate data, both in terms of choices and decision times, and does so across varying types of choices (e.g., charitable giving and hedonic consumption) and numbers of options well. However, when comparing models on the level of the individual, for a majority of participants simpler models capture choice data better. The theoretical and practical implications of these findings are discussed. (PsycINFO Database Record

Gaze Amplifies Value in Decision Making

“…One finding that distinguishes between the two proposed mechanisms (additive and multiplicative) is the inverse relationship between overall value and RT. This highlights that RTs are not simply a function of the value differences (as is often assumed; e.g., Ashby et al, 2016). These factors need to be accounted for in model comparisons.…”

GLAMbox: A Python toolbox for investigating the association between gaze allocation and decision behaviour

“…Researchers have recently started to systematically investigate the role of visual gaze in the decision making process. By now, it is established that eye movements do not merely serve to sample information that is then processed independently to produce a choice, but that they are actively involved in the construction of preferences (Ashby et al, 2016;Cavanagh et al, 2014;Folke et al, 2017;Konovalov & Krajbich, 2016;Krajbich et al, 2010;Krajbich & Rangel, 2011;Orquin & Mueller Loose, 2013;Shimojo et al, 2003;Tavares et al, 2017;Thomas et al, 2019). The dominant theoretical perspective is that evidence accumulation in favor of each option is modulated by gaze allocation, so that accumulation for non-fixated options is attenuated.…”

“…between gaze allocation and choice behaviour (Smith & Krajbich, 2018;Thomas, Molter, Krajbich, Heekeren, & Mohr, 2019) as well as individual differences in the decision mechanisms used (Ashby et al, 2016). Yet, the majority of model-based investigations of the relationship between gaze allocation and choice behaviour were focused on the group level, disregarding differences between individuals.…”

Inferiority, not similarity of the decoy to target, is what drives the transfer of attention underlying the attraction effect: Evidence from an eye-tracking study with real choices.

“…Despite terms like 'attention' or 'saliency' featuring prominently in these explanations, and despite eye-tracking being often used to investigate other forms of decision bias (e.g., Król & Król, in press), there has been little attempt to use it to study the visual attention patterns accompanying decoy-induced preference reversals. Such an analysis seems fruitful, because research on value-based choice (in the absence of decoys) demonstrated that attention guides choices in a positive feedback loop, whereby a stronger preference for an option causes more attention to be allocated to it, and vice versa (Ashby, Jekel, Dickert, & Glöckner, 2016;Krajbich, Armel, & Rangel, 2010;Reutskaja, Nagel, Camerer, & Rangel, 2011;Shimojo, Simion, Shimojo, & Scheier, 2003). Nevertheless, we are aware of only a very small number of processtracing studies of context effects, most of which were based on functional magnetic resonance imaging (W. Hedgcock, Rao, & Chen, 2009;Hu & Yu, 2014;Li, Michael, Balaguer, Herce Castañón, & Summerfield, 2018;Mohr, Heekeren, & Rieskamp, 2017), and only one of which used eye-tracking.…”