Dual-coding theory proposes that superior recall for pictures relative to words (the picture superiority effect; PSE) results from encoding information both verbally and visually, creating dual memory codes that aid subsequent retrieval. The current study tested this dual-coding theory by examining recall in individuals with aphantasia, who cannot voluntarily generate mental imagery and thus should not benefit from a secondary visual memory code. We compared recall performance between aphantasics and typical imagers across four stimulus types: pictures, symbols, picture names, and symbol names. Contrary to what dual-coding theory would predict, aphantasic individuals showed a robust picture superiority benefit in memory. While typical imagers recalled pictures and symbols at similar rates, aphantasics recalled symbols significantly better than pictures. These findings challenge dual-coding theory’s account of the PSE, revealing that its core assumption—that conscious access to image codes is required for enhanced picture recall—is flawed. Alternative accounts based on distinctiveness are offered, as well as potential substantive revisions that would be required for dual coding to remain a tenable explanation of picture superiority.