The present paper explores the role of sonority and other perceptual constraints in governing syllable structure constraints. One of the most important issues in phonology today is the formalization of the phonetic grounding of markedness constraints (Hayes and Steriade 2004). Sonority constraints have been particularly controversial because there is no formalized definition of sonority, but rather several different contributing factors, such as intensity, constriction and formant transitions, that all vary depending on context (Henke, Kaisse, and Wright 2012; Wright 2004). This paper makes use of an artificial grammar learning paradigm, whereby adult English speakers were exposed to a consonant-consonant metathesis pattern that either improved sonority at a syllable boundary, or worsened sonority at a syllable boundary. Learners did not show generalization in line with sonority-based syllable contact laws, but instead showed generalization in accordance with avoidance of a voiced obstruent in coda position, thus supporting a theory of sonority and syllable contact that makes use of the interaction of perceptual cues, rather than a strict, abstract sonority hierarchy.