Previous VMAC studies using gaze measures (Le Pelley et al., 2015; Pearson et al., 2015; Pearson et al., 2016) have demonstrated a mean effect size of dz = 0.73 (range = 0.41–1.4). Based on an anticipated effect size of d = 0.73, a power analysis conducted with G*Power (Faul, Erdfelder, Lang, & Buchner, 2007) indicated that 48 participants per group would provide adequate power (~.95 for between-subjects t-tests, >.99 for within-subjects t-tests). We therefore tested for enough days to collect data from 96 participants (University of New South Wales (UNSW) Sydney students), with a final sample size of 98 (n = 49 per group, 62 females, age M = 20.4, SD = 3.85). Participants received a monetary bonus dependent upon their performance (M = 20.9 AUD, SD = 2.57 AUD).
This study was approved by the UNSW Sydney Human Research Ethics Advisory Panel (Psychology). Participants were tested using a Tobii TX-300 eye-tracker (sampling frequency 300 Hz), mounted on a 23-in. monitor (1,920 × 1,080 resolution, 60 Hz refresh rate) with a chin rest ~60 cm from the screen. Gaze-contingent calculations used down-sampled gaze data (100 Hz). Stimulus presentation was controlled by MATLAB using Psychophysics Toolbox extensions (Brainard, 1997).
All stimuli were presented on a black background. Each trial consisted of a fixation display, search display and feedback display (see Fig. 2a). The fixation display consisted of a central white cross (0.5° × 0.5° visual angle) inside a white circle (diameter 3.0°). The search display comprised six filled shapes (2.3° × 2.3°) arranged evenly around the screen centre at 5.1° eccentricity (see Fig. 2). Five of the shapes were circles, and one (the target) was a diamond. The diamond target and four of the circles were grey (CIE x, y chromaticity coordinates: .304/.377, luminance ~32 cd/m2), with the remaining circle (the distractor) either red or blue (CIE coordinates: red .595/.360, blue .160/.116, luminance ~42.5 cd/m2), or the same shade of grey as the other shapes. The feedback display showed the reward for the previous trial, as well as the total reward accumulated.
Participants were alternately assigned to group Omission or group Yoked. For 25 of the participants in each group, red was the high-reward colour and blue was the low-reward colour; these relationships were reversed for the remaining participants. There were three types of trial: (1) high-reward trials, in which the distractor was rendered in the high-reward colour and a 10¢ reward was available; (2) low-reward trials, in which the distractor was rendered in the low-reward colour and a 1¢ reward was available; and (3) distractor-absent trials, in which all shapes were rendered in grey and there was an equal likelihood of 10¢ or 1¢ being available. The experiment comprised ten blocks of 48 trials. Each block consisted of 20 high-reward trials, 20 low-reward trials, and eight distractor-absent trials. For participants in group Omission, trial type was randomly determined within these constraints, and the location of the target and distractor were randomly determined with the constraint that the distractor was never presented adjacent to the target. For participants in group Yoked, trial type, target location, and distractor location on each trial were yoked to those of one participant in group Omission, such that each pair of matched participants experienced the exact same sequence of trials (i.e., the same sequence of trial types, with the target and distractor in the same positions across all trials; see Fig. 2b).
A circular region of interest (ROI) with a diameter of 3.5° was defined around the target, and a larger ROI (5.1°) was defined around the distractor. On distractor-absent trials, one of the non-target circles that was not adjacent to the target was randomly chosen to act as the ‘distractor’ location.Footnote 2 A response was registered when 100 ms of gaze dwell-time was detected within the target ROI. Response times (RTs) slower than 600 ms were not rewarded. For group Omission, the reward for the trial was cancelled if any gaze was detected within the distractor ROI (hereafter called omission trials). By contrast, for group Yoked, omission trials were yoked to those of the participant’s matched pair (i.e., regardless of whether or not the Yoked participant looked at the distractor, a reward omission would be triggered on trial n if their matched pair from group Omission had looked at the distractor and triggered an omission trial on trial n; see Fig. 2b). Yoking the trial sequence and omission trials ensured that participants from each group had near-identical signalling relationships between each distractor and its associated reward (i.e., each distractor was paired with its associated reward equally often, disregarding timeouts), but different response relationships between each distractor and its associated reward (i.e., looking at the distractor resulted in the omission of reward for group Omission, whereas looking at the distractor had no direct effect on reward delivery for group Yoked).
Participants were told that their task was to move their eyes to the target as quickly and directly as possible on each trial, and that they could earn 0¢, 1¢ or 10¢ on each trial “depending on how fast and accurate” their response was, with responses slower than 600 ms receiving no reward.
Each trial began with the fixation display. Once 700 ms of gaze dwell-time was recorded within the circle surrounding the fixation cross, or after 5,000 ms, the cross and the circle turned yellow. After 300 ms the screen blanked, and after a random interval of 600, 700 or 800 ms the search display appeared and remained on screen until a response was recorded, or until 2,000 ms had passed. The feedback display then appeared for 1,400 ms. The inter-trial interval was 1,400 ms. Participants took a short break after every second block.
In line with previous protocols (Le Pelley et al., 2015; Pearson et al., 2015), the first two trials of the task and the first two trials after each break were discarded. Hard timeouts (1.3% of all trials) were also discarded. Valid gaze data were registered on an average of 96.6% (SD = 5.9%) of samples.
Statistical analyses were conducted in R (Version 3.5.1). Greenhouse-Geisser corrected degrees of freedom are reported where appropriate. When conclusions are drawn on the basis of a null effect, we report the Bayes factor that corresponds to a Bayesian t-test using the default Cauchy prior.
Due to experimenter error, raw gaze coordinates were not saved during the experiment. However, data files containing aggregated gaze measures (i.e., a trial-by-trial recording of whether or not gaze was detected on the distractor, as well as gaze dwell-times on each distractor location) are available at https://osf.io/dy5kj/.