Commentary: Rethinking fast and slow based on a critique of reaction-time reverse inference.

Abstract

An increasingly common claim among cognitive psychologists is that the human mind is capable of two fundamentally different types of processes (Evans and Stanovich, 2013): Type 1 processing that is triggered autonomously by a stimulus and Type 2 processing that operates on a deliberate level via working memory and that allows for decoupling or override from default (Type 1) outputs. An article recently published in Nature Communications by Krajbich et al. (2015; hereafter, KBHF) focused on the use of response time (RT) differences as evidence for these dual-process theories (hereafter, DPT). KBHF outline how some dual-process theorists argue that if RTs are shorter for some response (“A”) than some other response (“B”), then this supports a DPT wherein Response A is intuitive and Response B is deliberative. Then, using economic games and intertemporal choice as examples, KBHF go on to argue that these RT differences can be better accounted for by sequential sampling models (SSM). Specifically, they argue that RT should increase when participants are presented options that are hard to discriminate (e.g., between two equally preferential choices). Thus, SSMs highlight the importance of conflict between choices as an explanation for RT differences. KBHF conclude by strongly cautioning against the use of RT differences to support DPTs. In this commentary we will briefly discuss two important questions that arise from KBHF's important contribution: (1) Are all cases in which researchers have used RTs to support DPTs undermined by KBHF's analysis? and (2) Are there any RT differences that are more easily explained under DPT than by SSMs? Our overarching goal is to illustrate that the scope of KBHF's analysis is far more limited than would be reasonable surmised by their original article.