Evaluating tourism destination images: Integrating survey, neural responses and artificial intelligence

This study introduces a triple-method approach to evaluate tourism destination attractiveness by integrating consumer surveys, neurophysiological measurements, and large language models (LLMs). Drawing on dual-process theory as a conceptual and interpretive framework, we examine how different measurement methods, individually and in combination, can help explain tourists' preferences for destination images. The study involved 96 participants from two countries who were shown in an experiment ten images representing various tourism types while collecting self-reported data, neurophysiological measurements, and LLM-based image assessments. To analyze image rankings and relative preferences, we adopted a multi-stage analytical approach. Our findings suggest that each measurement method captures distinct aspects of tourist decision-making: Survey measurements primarily correlate with fast-affective responses. Neurophysiological data reveal affective and cognitive responses not detected in self-reports. LLMs offer complementary evaluations. However, the incremental explanatory value of LLM-based assessment varies by image-type. Combined measurements also differ in effectiveness across tourism types, underscoring the need for tailored evaluation approaches. • Triple-method approach integrates surveys, neurophysiology, and LLM-based for tourism images preferences. • Each method captures distinct decision-relevant signals, with dual-process theory used as a conceptual lens. • Neurophysiological measures can reveal affective or attentional responses not detected in self-reports. • DMOs gain decision-support for visual content optimization (screening, targeting, and pre-testing). • Tourism image evaluation benefits from tailored measurement strategies by image type.