Two experiments explored ways in which novel stimuli might be represented in associative learning, focusing on (1) representations in which novel stimuli embody novelty as a stimulus feature that acquires associative strength in the same fashion as color, shape, texture, or other frequently used stimulus features; and (2) representations in which novel stimuli embody common elements, that is, the stimulus elements shared among other stimuli in an experimental setting. Both experiments examined the effects of reinforcing or nonreinforcing separately presented novel stimuli on learning about compound stimuli that included novel stimuli as part of the compound. Experiment 1 found that the relative ease of learning positive and negative patterning discriminations could be reversed, depending on whether novel stimuli were separately reinforced or nonreinforced. This result is consistent with predictions from a model that assumes novelty is a stimulus feature. Experiment 2 found that the additive effects of combining relevant and irrelevant cues in compound were not obtained when novel stimuli were used as irrelevant cues. This result is consistent with predictions from a model that assumes novel stimuli are represented as common elements. Implications for understanding how both types of representation account for the role of novelty in complex discrimination learning in general are discussed.