An Empirical Exploration of Fast and Slow Errors at Group and Subject Levels in Speeded Decision-making

The time taken to make decisions, in combination with the choices made, form the basis of our understanding of speeded human decision-making. A key basis for developing existing theories of speeded decision-making have been the nuanced trends contained within choice response time data, such as the ubiquitous right skew. Here, we provide an empirical exploration of one of these nuanced trends: the patterns of relative mean response time between correct and error responses (i.e., fast/slow errors). Specifically, we investigate both the group level patterns in fast/slow errors across three brightness discrimination experiments with speed and accuracy emphasis conditions, and the individual level consistency in these patterns within each experiment. Our findings showed inconsistencies in the patterns of fast/slow errors at the group-level, with each experiment showing a different pattern of relative error speed; only fast errors, only slow errors, and fast errors in one condition and slow errors in the other condition, respectively. Furthermore, all experiments showed inconsistency between participants in whether they display fast or slow errors in at least one condition. Our findings show that even within near-identical experimental paradigms – or the same experiment – there is inconsistency in patterns of and conclusions about fast/slow errors.