Visual detection of cryptic displays in jumping spiders

Most animals can segment camouflaged targets via static cues, such as luminance or contrast differences against a background. However, targets can be cryptic and merge with background clutter, making static cues unavailable until the camouflaged object starts moving: the target can then be segmented by identifying spatially coherent luminance changes over time. Jumping spiders represent a unique model for the study of spatiotemporal segmentation, as they process moving signals and static figures separately through distinct pathways associated with different pairs of eyes: the secondary eyes detect moving objects and trigger full body pivot responses, bringing the target within focused view of the principal eyes for subsequent figure processing. While secondary eyes can be selective for specific stimulus characteristics that cause the animal to pivot, it remains unclear whether their selectivity requires full motion segmentation or relies on simpler stimulus characteristics. Here, we tested whether the pivoting response triggered by secondary eyes is specifically linked to motion segmentation, or whether it is also elicited by less specific stimulus characteristics. We presented jumping spiders with several moving and non-moving stimuli, including clearly delineated objects, cryptic objects visible only through motion, and non-segmentable spatio-temporal luminance patterns. Based on the measured responses, we developed a computational model that captured all salient features of the pivoting behavior produced by the animal. We found that jumping spiders do not require a discrete, segmented object to trigger a pivoting response. Instead, they attend to luminance patterns that change across both time and space, while disregarding changes that repeat over the same spatial location. This mechanism constitutes an effective solution in terms of neural simplification: it offloads the task of complex motion segmentation from the secondary eyes, thus maximizing the efficiency of visual modularity—a unique evolutionary solution to the problem of brain miniaturization.