Analysis

All analyses were conducted using R \cite{Team2017}; mixed effects models were implemented using the lme4 package \cite{Bates2015} and generalized additive models were implemented using the mgcv package \cite{Wood2011}.

Prior to analysis, trials were removed if the eye tracker detected either a saccade (14%) or blink (3%). Additionally, any trials on which a response was detected prior to appearance of the target (1%) were removed. As a check that participants were engaging the go/no-go task as instructed, and to evaluate whether cuing affected the presence/absence of responses, a binomial mixed effects model was performed with response presence/absence as the outcome variable predicted by cuing, the expected response (present/absent) and the interaction of these variables, with subjects as random effects on the intercept and all predictors. In such models, the main effect of the expected response variable measures overall performance at discriminating the go/no-go stimuli, the main effect of cuing measures any response bias induced by the appearance of the target at the cued location, and the interaction between the expected response and cuing measures any change in discrimination performance induced by the appearance of the target at the cued location. Consistent with participants engaging the go/no-go task as instructed, there was a considerable effect of expected response (log-odds coefficient: 8.9; 95% CI: 9.8 – 7.9), while cuing had neither a main effect (log-odds coefficient: 0.7; 95% CI: -0.3 – 1.6) nor interacted with expected response (log-odds coefficient: 0.06; 95% CI: -2.0 – 2.2), suggesting that cuing did not induce any response bias nor affect gross go/no-go discrimination performance