One of the great challenges in user modelling is the question how to gain personal information about the user. Of course, the most direct way is to ask him or her. Nevertheless, this option is not much appreciated by the user: It takes time and effort to fill in questionnaires and the user might quickly get a feeling of being kept from his or her original goal. On the other hand, the user is not always aware of the information needed by a system, especially when it gets to complex aspects of his or her personality or preferences. Perceptual preference can be considered as an aspect of cognitive style and signifies an individuals preference or preference pattern regarding perceptual channels of information presentation and information processing. This thesis develops and evaluates three different methods for testing perceptual preference. The first one is an explicit method, eliciting perceptual preference via the Perceptual Preference Questionnaire (PPQ). ^Perceptual preference is assessed in regard to information processing, knowledge gain, and learning. It is analysed via the scales visual, auditory, kinaesthetic and olfactory/gustatory preference and examined in regard to distribution as well as co-occurrence patterns of perceptual preference and user interests. The second method is an implicit method, investigating user-generated text as an indicator of perceptual preference. In order to study the use of sensory vocabulary within German text, we develop the Signal Term Extraction Method (STEM) Algorithm and integrate it into an analysis pipeline. The analysis of a huge forum corpus is focused on the use of sensory vocabulary patterns as well as response behaviour. The third method is an embedded method. The Game Embedded Testing of Learning Strategies (GETOLS) method embeds the testing procedure into a didactic adventure game. ^The results obtained automatically by our test environment are compared to the ones elicited via conventional testing by a psychologist. Overall, the evaluation results show a high potential for considering perceptual preference as a valuable extension to existing user models.