We investigated the origins of analogical ability in 7-month-old infants, using the simplest and most basic relation – that of sameness and difference between two things. Experiment 1 showed infants were unable to detect and generalize these relations from a single exemplar (as suggested in Tyrrell et al., 1991). Experiment 2 used a habituation-dishabituation paradigm and found that infants could generalize the same-different relation to novel objects with six to nine training trials. Experiment 3 demonstrated that labels influenced performance: labeling the relation enhanced performance, but labeling the individual objects hindered performance. In addition, we varied infants’ prior experience with the objects and found signatures of relational learning have continuity across development. In summary, abstraction of relations can be facilitated by comparison across exemplars, disrupted by the saliency of individual objects, and manipulated by labeling. These findings are discussed in light of recent debates about phylogenetic continuity in relational abilities.