Many educationalists advocate that students should be given frequent opportunities to learn through problem solving and learn collaboratively. Over many decades, researchers into Cognitive Load Theory (CLT) have demonstrated that for novel content, problem solving is a very inefficient method to learn. However, little research into collaborative learning has been conducted from the perspective of CLT. Consequently, the motivation of this thesis was to investigate the effectiveness of combining problem solving and collaboration through the lens of CLT and principles of evolutionary educational psychology. Four experiments were completed with Grade 7 Indonesian students using Mathematics as the learning materials. Each experiment had an acquisition phase, where the interventions were introduced, followed by tests of similar content and transfer. All experiments compared individual and collaborative learning. During acquisition, students in collaborative learning groups completed the learning task together in face-to-face interactions but completed the test phases individually. In the first three experiments, collaborative learning was structured by grouping students with the same content knowledge. In the last experiment, students with different content knowledge were grouped and required to share their knowledge to solve tasks that depended on the shared knowledge. In addition, experiments included comparisons between problem solving and worked examples and used problem complexity as a potential moderator. The overall results demonstrated that a worked example strategy was superior to a problem solving strategy. It was found that for high-complexity tasks, students who originally learned individually scored significantly higher than those who learned collaboratively. However, when learning by problem solving, collaborative learning was more effective than individual learning, but for low-complexity problems only. Experiment 4 showed that when students had to share different content knowledge, collaborative learning was found to be superior to individual learning in a problem solving context. In contrast, when all students had access to the relevant content knowledge, individual learning was superior to collaborative learning. In summary, the study found that some caution should be shown when advocating the use of problem solving and collaboration. Both strategies can be effective, but according to quite strict conditions.