Adaptive models are complex systems that are especially difficult to investigate. In particular, when researching such systems it easy to blur the line between model and domain, given that the former, the later, and both are vital adaptive system research pursuits. When investigating a model, one may become ‘caught-up’ on model performance resulting in a vicious cycle of increasing the complexity of benchmark problem instances, and losing site of the characteristic model behaviours that one is seeking. This work proposes a simple ‘colour space’ domain from which trivial problem instances may be derived to investigate complex systems. Example optimization and patternrecognition problem instances are described, and the paper is rounded off with a discussion of the potential limitations of using the domain to investigate adaptive models.