In view of the criticism of ‘forced distinction between mainstream economics and non-mainstream economics’, this paper points out that ‘mainstream’ economics and ‘non-mainstream’ economics exist objectively in current economic academia, and mainstream economics embodies the normal paradigm in modern economic analysis. This paradigm also has affected economic experiments, and ‘mainstream’ is reflected in the basic ways and typical procedures adopted by most economists or professors in economics schools when doing behavioral experiments. Obviously, this mainstream orientation of experimental economics tends to combine mathematical logic with economic experiments, thereby constructing a simple situation like ‘double blindness’. However, ‘double blind’ experiments ‘verify’ the theories of mainstream economics since they construct the social context fit for economic man. However, it does not meet the scientific requirements because of strong tautology. At the same time, in view of the argument that ‘double blind design’ helps to achieve the internal validity, this paper points out that the quality and progress of economic theories mainly depend on their explanatory ability, and the explanatory ability is rooted in the internal validity of a theory. However, the internal effectiveness of economic theories is fundamentally different from natural science, and it reflects the reality of behavior logic rather than the consistency of formal logic. No matter of this, the mainstream experimental economics has designed a ‘double blind’ program, in order to achieve internal validity through strict condition control, but what is pursued and achieved by this internal validity is just the consistency of formal logic. Furthermore, the ‘double blind’ experiments try (as far as possible) to exclude the pro-sociality of actors, so as to cater to, in essence, the paradigm of economic man in modern mainstream economics. Through the above logic textual research in two aspects, we can clearly understand that to promote economic experiments more effective, context settings should be gradually materialized and actualized rather than stick to ‘double blind’ experiments. In fact, the results of an experiment should be determined by the world rather than by a theory, and the results of the same theory in different experimental settings would also be distinct, so the real gist of an experiment lies in that a theory is tested by reference to the world. Meanwhile, non-mainstream experiments led by behavioral economists like Kahneman and Sall always relax conditional control and also obtain a series of behavioral effects against mainstream economics; in this way, it can methodize and systemize these ‘isolated’ findings by extraction and synthesis methods, and thus form a more perfect cognitive system, helping to combine internal and external effectiveness and further promoting the progress of economic theories.