The ability to assess learning hinges on the quality of the instruments that are used. This paper reports on the first stage of the design of software to assist educators in ensuring assessment questions meet educational outcomes. A review of the literature within the field of instructional psychology was undertaken with a view towards establishing a theoretical foundation for the design of an autonomous system to classify instructional statements according to learning complexity. This review focussed on taxonomies of learning outcomes commonly used by educators and instructional designers to design and deliver instructional content within the cognitive domain. An Instructional Activity Matrix was synthesised from this review to provide an informed basis for the autonomous classification of instructional statements according to the nouns and verbs used to describe the types of knowledge and cognitive processes to be elicited. This paper reports on the theoretical development of the Instructional Activity Matrix and how it will be implemented within the autonomous system.