This article applies an ontology-based approach to economic experiments, emphasizing their differences with respect to physical science experiments. To contextualize our discussion, a conciliatory Weberian view of the similarities and differences between natural and social sciences is provided. After that, some ontological features of the social sciences' domain are highlighted, together with their problematic effect on experimental economics. Specifically, we focus on human beings' representational capacities and intentionality, their cultural and conventionally mediated forms of social interaction, and the holistic openness, instability and uncertainty of the social world. Finally, we emphasize the severe under-determination of theory by evidence affecting social science, as well as the related problems of empirical ambiguity, confirmatory biases and propensity to pseudoscientific practices in experimental economics.