Test–retest reliability of eye-tracking metrics for the measurement and classification of sign- trackers and goal-trackers

Reward-predictive cues shape behavior through Pavlovian conditioning, yet individuals vary in the value they assign to these cues. Sign-trackers (ST) attribute both incentive and predictive values to these cues, while goal-trackers (GT) assign solely predictive value, orienting their attention rapidly toward the forthcoming reward. Although animal studies report sign-tracking and goal-tracking as stable, trait-like learning styles often related to maladaptive behaviors, human research has produced inconsistent results, raising questions about the reliability and the stability of this behavior. To address these issues, we investigated the test-retest reliability and stability of the classification over a four-month period of the eye-gaze index most commonly used in human sign- and goal-tracking literature. Our findings revealed good capacity for detecting sign-tracking behavior, but limited consistency in identifying goal-tracking behavior, resulting in suboptimal overall reliability. These results raise the possibility that goal-tracking may be either genuinely rare in the population or poorly captured by the current index. Overall, while the gaze index holds promise for identifying ST behavior, methodological refinements or alternative approaches may be needed to reliably detect GT behavior and improve the utility of this measure in future research.