In everyday life, people constantly classify objects, individuals and situations in categories, in order to make the world more comprehensible and predictable. While most researchers agree that an item is categorized based on its similarity to a category representation, they do not agree on the exact nature of the category representation. Some believe that a category is represented by all of the members of the category they have previously seen (i.e., the exemplar representation) others believe that a category is represented by an abstract summary of the members of the category (i.e. the prototype representation). An alternative account that received less attention is the view that categories can be cognitively stored by a combination of both: representations were some exemplars are merged into a set of subprototypes and where other exemplars are represented individually. These representations are also called subgroup or intermediate representations. Within the dissertation we used the Varying Abstraction Model (VAM) to systematically investigate all the possible subgroup representations of a particular category. In addition, within the prototype and exemplar discussion it was implicitly assumed that individuals do not differ in the representation that they use to represent a category in semantic memory and that the same representation was always used irrespective of situational differences. Recent studies show that this assumption is not correct. Both aspects, the idea of subgroup representations and the idea of individual differences and situational characteristics that influence the degree of abstraction in the category representations formed the basis of this dissertation. The first part of the dissertation starts with a description of an extension of the VAM that can account for categorization response times. It then describes a simulation study in which we tried to provide an explanation for the fitting advantage of the exemplar representation in categorization studies based on the intermediate representations of the VAM. In the second part, the VAM was used to further investigate interindividual differences in categorization strategy and the factors that might influence these differences. In the first chapter of the second part, the effect of category size and instructions on representational abstraction was investigated. The second chapter describes a study in which the stability of individual differences in representational abstraction and their relation with working memory capacity was assessed. The last chapter investigated the same relationship using a dual-task approach.