With the increasing popularity of multimodal information on e-commerce platforms, the impact of text-photo sentiment disparity on online consumer behavior has grown increasingly significant. Based on Cognitive Dissonance Theory, this study explores the underlying mechanism through which text-photo sentiment disparity affects the perceived helpfulness of online reviews. By analyzing secondary data from the Ctrip platform and conducting four laboratory experiments, we find the sentiment disparity between images and text significantly boosts the perceived helpfulness of reviews. This effect is mediated through the perception of cognitive dissonance. Additionally, we explore the moderating effect of the Need for Cognition (NFC) during information processing, showing that for individuals with a high NFC, the impact of cognitive dissonance on review helpfulness is more salient. These findings promote the application of multimodal information processing and Cognitive Dissonance Theory, offering valuable insights for e-commerce platforms to optimize review systems and marketing strategies effectively.