Guidance of attention by statistical learning when multiple guidance mechanisms compete

Abstract

Previous research has shown that people are sensitive to statistical regularities and will implicitly learn to bias attention toward locations and features frequently associated with search targets. However, most prior work has involved a single biasing contingency. In Experiment 1, we investigate whether individuals can simultaneously learn and implement two distinct contingencies: one based on location and another based on color. The results suggest that both contingencies are learned implicitly and exert independent effects on attentional allocation. Experiments 2 and 3 examine whether shifting one feature to the volitional system via endogenous cues affects the implicit learning and implementation of contingencies for the other feature. The findings indicate that endogenous cues for one feature do not block implicit learning of contingencies in the other. However, the influence of implicit contingencies on attentional allocation depends on the validity of the volitional cues: they are effective when cues are neutral or valid, but not when invalid. This pattern suggests that the implicit and volitional systems operate hierarchically, with the volitional system taking precedence. Efficient attentional guidance is key to effective visual search. While both volitional control and statistical learning can guide attention, little is known about how attention is influenced when multiple guidance mechanisms compete. Using a task with competing statistical regularities, one for target location and one for color, we show that both statistical learning biases independently influence attention. Shifting a feature to the volitional control system does not disrupt the statistical learning of the other, but invalid volitional cues eliminate statistical learning effects, suggesting a hierarchical relationship. These findings offer new insight into how multiple guidance systems shape attentional allocation. Subject level data and SPSS analysis syntax available on the Open Science Framework at: https://osf.io/5vt74/overview. The experiments were not preregistered, but the data, and analysis syntax are publicly available at https://osf.io/5vt74/overview.