Mathematics students almost exclusively use pencil and paper - that is, they learn without computational support. In this research, 16 high school students varying in ability from low to high participated in a comparative assessment of geometry problem solving using: (1) pencil and paper, (2) an Anoto-based digital stylus and paper interface, (3) a pen tablet interface, and (4) a graphical tablet interface. Cognitive Load Theory correctly predicted that as interfaces departed more from familiar work practice, students experienced greater cognitive load and corresponding reductions in their expressive fluency and planning. The results of this study indicate that students' communication patterns and meta-cognitive control can be enhanced by pen-based interfaces during math problem solving activities. In addition, low-performing students do not automatically reap the same advantage as high performers when new interface tools are introduced, which means intervention may be required to avoid expanding the achievement gap between groups unless intervention is undertaken.