Abstract Probe stimuli which elicit activity reflexly provide a means of assessing ‘top-down’ effects of attentional manipulations while minimizing the perhaps insolvable problem of determining whether effects are due to post-perceptual selection or to changes in the input pathway. Parallel experiments on adults and infants presented acoustic and visual probes unpredictably while subjects attended to acoustic and visual foregrounds: Attention was indexed by heart rate deceleration. When probe and foreground modality matched, probe-elicited reflex blinks were significantly facilitated in magnitude (infants) or latency (adults) relative to reflexes elicited when probe and foreground modality mismatched. Further, facilitation was greater when modality-matching probes were presented over foregrounds judged a priori to be more ‘interesting’ than ‘dull’ foregrounds. Because acoustic and visual blink reflexes have a common efferent path, modulating effects must have occurred earlier, in modality-specific paths. As such, the results suggest that attention can influence ‘automatic’ sensory-perceptual analysis.