Recent work with human infants and toddlers suggests a dissociation between performance on looking and reaching tasks. Specifically, infants appear to generate accurate representations of occluded objects and their actions when tested in expectancy violation looking tasks but often fail to use this information when reaching for occluded objects. We explore a similar dissociation in cotton-top tamarin monkeys ("Saguinus oedipus"). We presented adult tamarins with an event in which a piece of food rolled behind an occluder and into a solid barrier. In Experiment 1, subjects were required to retrieve the hidden food using the location of the solid barrier. Like human toddlers, adult tamarins failed to take into account solidity information when reaching for an invisibly displaced object. In Experiments 2 and 3, we presented subjects with expectancy violation looking versions of the same solidity problem using an identical apparatus and setup. We presented subjects with an event in which a piece of food appeared to roll unexpectedly through a solid barrier or stopped at the appropriate spot. Although tamarins failed to locate the food in Experiment 1, the same subjects successfully detected violations of solidity in these 2 looking studies. This performance dissociation is discussed in light of similar dissociations in human toddlers and other primate species. (Contains 4 figures, 2 tables and 1 footnote.)