Errors, fast and slow: an analysis of response times in probability judgments

Abstract

Based on the Dual-Process Diffusion Model, we tested three hypotheses about response times of errors and correct responses in probability judgments. We predicted that correct responses were (1) slower than errors in the case of conflicting decision processes but (2) faster than errors in the case of alignment; and that they were (3) slower in the case of conflict than in the case of alignment. A binary-choice experiment was conducted in which three types of decision problems elicited conflict or alignment of a deliberative decision process and a heuristic decision process. Consistent with the traditional dual-process architecture, the former captured computational-normative decision strategies and the latter described intuitive-affective aspects of decision making. The hypotheses (1) and (3) were supported, while no statistically significant evidence was found for (2). Implications for the generalisability of the Dual-Process Diffusion Model to slow probability judgments are discussed.

Is there more than one ratio bias? If so, why?

“…Contrary to our expectations, we found a tendency that decisions exhibiting a bias take longer than rational decisions. This finding is in accordance with some observations in the recent literature (De Neys & Glumicic, 2008; Ludwig et al., 2020; Pennycook & Thompson, 2012), which showed that correct answers in alignment with intuitive responses are generated rapidly. However, these effects were rarely studied in the context of the ratio bias.…”

“…An exception is Ludwig et al. (2020), but in their research, the ratio bias was one type of problems in which this result did not hold.…”

“…However, these effects were rarely studied in the context of the ratio bias. An exception is Ludwig et al (2020), but in their research, the ratio bias was one type of problems in which this result did not hold.…”

The Zero Effect: An Eye‐Tracking Study of Affect and Motivation in Risky Choices

“…Previous research argued that decision times indicated decision difficulty (Achtziger and Alós-Ferrer 2014;Ludwig, Ahrens, and Achtziger 2020;Schotter et al 2010). If zero outcomes were processed more easily than other gamble attributes, decisions should be quicker for lotteries with zero outcomes (Types IV and V) compared to lotteries without zero outcomes (Types I-III).…”

Multiple Behavioral Rules in Cournot Oligopolies

“…As an extreme example, study the responses to questions as in the Cognitive Reflection Test, with median response times typically above 30 seconds, and find differences depending on whether the questions elicited more or less intuitive responses.4 Achtziger and Alós-Ferrer (2014) applied the model to reinforcement and Bayesian updating in a belief updating task Spiliopoulos (2018). used the model to study win-stay, lose-shift vs. more sophisticated (cognitive) heuristics in a repeated game played against computer algorithms Ludwig et al (2020). applied it to heuristic decisions in probability judgments.…”