In a series of experiments, we tested a recently proposed hypothesis stating that the degree of alignment between the form of a mental representation resulting from learning with a particular visualization format and the specific requirements of a learning task determines learning performance (task-appropriateness). Groups of participants were required to learn the stroke configuration, the stroke order, or the stroke directions of a set of Chinese pseudocharacters. For each learning task, participants were divided into groups receiving dynamic, static-sequential, or static visualizations. An old/new character recognition task was given at test. The results showed that learning both stroke configuration and stroke order was best with static pictures (Experiments 1 and 2), while there was no reliable difference between the groups for learning stroke direction (Experiment 3). An additional experiment, however, revealed that learning with sequential pictures was superior when testing was carried out with sequential pictures, irrespective of the learning task (Experiment 4). The combined evidence from all experiments speaks against task requirements playing a role in determining the effectiveness of a visualization format. Furthermore, the evidence supports the view that a high degree of congruence between information presented during learning and information presented at test results in better learning (study-test congruence). Implications for instructional design are discussed.