The surprise-attention link: a review

The surprise-attention hypothesis assumes a strong connection between surprise-expectancy discrepant events- and attention. Attention is easily engaged with surprising events, leading to long dwell times. In addition, if the expectancy discrepancy can be determined on the basis of simple, preattentively available information, attention can be captured by the surprising stimulus. This review summarizes different lines of research relevant to the proposed surprise-attention link: shifts of attention as indexed by accuracy gains and efficiency gains, validity effects, shifts of gaze, discrepancies in natural scenes, surprise-induced blindness, and action interruption. It is argued that there is convergent evidence for the surprise-attention link in general, and for the particular hypothesis that the underlying mechanism constantly tests expectancies on different levels of representation. Evidence also converges on a latency of an attentional engagement of nearly 400 ms. This seems to be a unique feature of surprise capture that also questions the validity of models proposing that saliency is an early automatic attractor of attention. Mechanisms possibly underlying the surprise-attention link are discussed.