This paper discusses the divide between theory-based and test-based understanding of intelligence. Regardless of which intelligence theory might be true or might represent the academic state-of-the-art, intelligence test-batteries are commonly administered because of their existence and availability, and in that sense establish their own 'theory'. The question is which representation of human intelligence actually does apply in practice. As a matter of fact, Wechsler-based intelligence test-batteries are widely used all over the world. In this paper a psychometrically sound representative of Wechsler-based intelligence test batteries, the Rasch model scaled Adaptive Intelligence Diagnosticum (AID 2), is used as an example. Exploratory factor analyses and confirmatory factor analyses were performed with respect to the theories of Spearman, Wechsler, Thurstone, Cattell, and Jäger. Confirmatory factor analyses were also performed with respect to a specific model of learning disorders. The data of 662 6- to 16-year old children and adolescents were analyzed. The results indicate a surprising finding: it was not the four factor solution yielded from exploratory factor analysis of the AID 2 standardization sample which fits the data best - Wechsler's two factor model (verbal vs. performance factor) fits almost equally well - but a simple model of learning disorders. This is a hierarchical model with 'perception' as its base component, followed by 'utilization', and 'retrieval' of information.