Introduction
Although risky drinking and alcohol dependence have been associated with spatial attentional biases, concerns have been raised about the reliability of the frequently-used dot-probe task. A form of anticipatory bias related to predictive cues has been found to be related to alcohol-related processes, and to have high reliability in the context of threat stimuli. It remains to be determined whether this anticipatory attentional bias also has good reliability for alcohol stimuli. Further, correlations with drinking-related individual differences need to be replicated.

Material and Methods
83 healthy adult participants were included, who completed the task and questionnaires on risky drinking (AUDIT-C), drinking motives (DMQ-R), reasons to abstain from drinking (RALD), and alcohol craving (ACQ). The task used a 400 ms Cue-Stimulus Interval, based on previous work. The Spearman-Brown split-half reliability of reaction time-based bias scores was calculated. The within-subject effect of probe location (predicted-alcohol versus predicted-non-alcohol) was tested using a paired-sample t-test. Correlations were calculated between bias scores and questionnaire scales; tests were one-sided for predicted effects and two-sided for exploratory effects.

Results
The reliability was .81 (.74 after outlier removal). There was no overall bias. A predicted correlation between risky drinking and anticipatory bias towards alcohol was found, but no other predicted or exploratory effects.

Discussion
The anticipatory attentional bias for alcohol is a reliably measurable individual difference, with some evidence that it is associated with risky drinking.

Conclusions
Implicit behavioural measures of spatial attentional bias can in principle achieve high reliability. Further study of attentional biases using predictive cues would appear to be promising.