Even though Mayer's cognitive theory of multimedia learning is supported by empirical evidence and is theoretical and practical appealing to instructional designers, recent research is not able to replicate the research findings in other knowledge domains (Mayer, 2001a; Mayer, 2001b, Mayer, 2003; Cox, 1999; Lowe, 2003; Schnotz & Bannert, 2003; Dobson, 1999). The formal iconic sign system used in knowledge domains can explain these results. Former research of the authors suggested extending CTML with additional design principles (De Westelinck, K., Valcke, M., De Craene, B. & Kirschner, P., in press). In the three researches reported here, two additional principles - based on 'activation' and 'collaboration' - are included in experimental designs. Next, 'training' in the use of external graphical representations is added as a research condition. Freshmen received a package consisting of a pre- test, printed learning materials and post-tests. In the first research 217 students were randomly assigned to four conditions with 'activation' as the key element: no external graphical representations, with external graphical representations, developing external graphical representations, completing pre-worked external graphical representations.In the second research again 'activation' was the core element but 'training' was added as an extra condition. This research assigned 217 students to four conditions: no external graphical representations, external graphical representations, developing external graphical representations and developing external graphical representations after receiving training. During the training the use of iconic sign systems was highlighted. 220 students were involved in the third research. Also here 'activation' was a central element combined with 'training' and 'collaboration'. Collaboration was defined as working together in groups of three. The students were randomly assigned to four different conditions. The conditions were the same as in the second research. The first research confirms earlier findings (De Westelinck, K., Valcke, M., De Craene, B. & Kirschner, P., in press). The design principles, as formulated by the CTML-theory, cannot be extrapolated to the social sciences knowledge domain in a straightforward way. In the first research, students who had to develop their own external graphical representations scored significant higher on retention and transfer tests than students in the other conditions (effect sizes d = 0.63, d = 0.66 and d = 0.54; d = 0.86). These results are in line with Lowe (2003) and Stern, Aprea and Webner (2003).The external graphical representations developed by the students in the second research were categorized based on the theory of Wileman (1993). The high usage level of certain representation categories was considered a predictor for higher test performance. Results of a linear regression did not confirm this hypothesis. Future analyses on the second and third research will be included in the presentation. Overall methodological remarks can be made. First, the quality of the representations can be questioned? These representations were developed by a team of educational scientists and were in line with examples of textbooks in the social sciences. A second remark is related to the themes that might have been too difficult and complex for freshman students. But this level is not exceptional for this audience in this particular university program. Thirdly, the researches of Mayer represent studies on very short study periods (e.g. 120 seconds). It is possible that the learning processes in this research have been more demanding than these in the researches of Mayer (2001a). A related remark is that the students did not have prior knowledge of the knowledge domain while CTML-research shows that prior knowledge is an important determining factor. Individual differences - other than prior knowledge - were not taken into account.Directions for future research will be presented.References:Cox, R. (1999). Representation construction, externalised cognition and individual differences. Learning and Instruction, 9, 343- 363.Dobson, M. (1999). Information enforcement and learning with interactive graphical systems. Learning and Instruction, 9, 365-390.Lowe, R.K. (2003). Animation and learning: selective processing of information in dynamic graphics. Learning and Instruction, 13, 157-176.De Westelinck, K., Valcke, M., De Craene, B. & Kirschner, P. (in press). Multimedia Learning in Social Sciences: Limitations of External Graphical Representations. Computers in Human Behavior.Mayer, R. E. (2001a). Multimedia Learning. Cambridge: University Press.Mayer, R. E. (2001b). Cognitive, metacognitive and motivational aspects of problem solving. In H. Hartman, ed., Metacognition in Learning and Instruction, pp. 87-102, Dordrecht: Kluwer Academic Press.Mayer, R.E. (2003). The promise of multimedia learning: using the same instructional design methods across different media. Learning and Instruction, 13, 125-139.Schnotz, W. & Bannert, M. (2003). Construction and interference in learning from multiple representations. Learning and Instruction, 13, 141-156.Stern, E., Aprea, C. & Ebner, H.G. (2003). Improving cross-content transfer in text processing by means of active graphical representation. Learning and Instruction, 13, 191-203.Wileman (1993). Visual Communicating. Educational Technology Publications Inc.