Fig 1.
Two-choice model representing a decision between left and right responses. Each accumulator begins with an initial amount of evidence (k), drawn from a uniform distribution [0, A], that increases at a noisy rate, with mean v, towards the response threshold (b). Decision time–the time taken for the winning accumulator (in this example, the Left response) to reach its respective threshold, along with nondecision time (t0)–time taken for stimulus encoding and response execution components–determine the overall response time. Adapted with permission (not part of governing open access license) from Springer Nature Customer Service Centre GmbH: Springer Nature, Psychonomic Bulletin & Review, Tillman, G., Zandt, T. Van, & Logan, G. D. Sequential sampling models without random between-trial variability: the racing diffusion model of speeded decision making, The Psychonomic Soceity, Inc. (2020).
Fig 2.
On a computer monitor, participants were presented a 500 ms warning signal which indicated a 70% probability of the upcoming imperative signal pointing either to the right (>) or left (<). Following the warning signal, a blank screen (250 ms), a fixation cross (500 ms), and another blank screen (250 ms) were presented. Following these, a 1000 ms green coloured imperative signal (> or <) was presented either in the same (a: congruent trials) or opposite (b: incongruent trials) direction to the warning signal. Participants were required to respond with a button consistent with the direction of the arrow in the imperative signal.
Table 1.
Model fit.
Table 2.
Posterior predictive analysis for Bvt0 model.
Table 3.
Posterior predictive analysis for Bt0 model.
Fig 3.
Observed (filled) and Bt0 model-predicted (unfilled) 10th (square), 50th (circle), and 90th (triangle) error (top panel) and correct (bottom panel) response time percentiles plotted, in milliseconds, separately for each age group–young (Y) and older (O) adults–as well as each bias type–block-wise (B) and trial-wise (T). Observed data points between congruent (C) and incongruent (IC) trial types are connected via dashed lines. Error bars for model-predicted response times represent 95% credible intervals.
Fig 4.
Observed (filled) and Bt0 model-predicted (unfilled) proportion of correct responses plotted separately for each age group–young (A) and older (B) adults–as well as each bias type–block-wise (1) and trial-wise (2). Observed data points between congruent (C) and incongruent (IC) trial types are connected via dashed lines. Error bars for model-predicted proportion of correct responses represent 95% credible intervals.
Table 4.
Median parameter estimates.
Table 5.
Differences in median parameter estimates.