Granivorous animals may prefer to predate or cache seed of certain plant species over others. Multiple studies have documented preference for larger, non-native seed by granivores. To accomplish this, researchers have traditionally used indirect inference by relating patterns of seed removal to the species composition of the granivorous animal community. To measure seed removal, researchers present seed to granivorous animals in the field using equipment intended to exclude certain animal taxa while permitting access to others. This approach allows researchers to differentiate patterns of seed removal among various taxa (e.g., birds, small mammals, and insects); however, it is unclear whether the animals of interest are freely using the exclusion devices, which may be a hindrance to discovering the seed dishes. We used video observation to perform a study of seed predation using a custom-built, infrared digital camera and recording system. We presented native and non-native seed mixtures in partitioned Petri dishes both within and outside of exclusion cages. The exclusion cages were intended to allow entrance by rodent taxa while preventing entrance by rabbits and birds. We documented all seed removal visits by granivorous animals, which we identified to the genus level. Genera exhibited varying seed removal patterns based on seed type (native vs. non-native) and dish type (open vs. enclosed). We documented avoidance of the enclosed dishes by all but one rodent taxa, even though these dishes were intended to be used freely by rodents. This suggests that preference for non-native seed occurs differentially among granivorous animals in this system; however, interpretation of these nuanced results would be difficult without the benefit of video observation. When feasible, video observation should accompany studies using in situ equipment to ensure incorrect assumptions do not lead to inappropriate interpretation of results.