According to the argumentative theory of reasoning, the primary function of reasoning is the exchange of arguments between the interlocutors. In the context of a group discussion, the exchange of arguments can result in increased performance of various tasks, including reasoning problems. The aim of this study was to examine whether such an improvement in reasoning accuracy occurs across eight tasks used to study different facets of reasoning and to examine the effects of accuracy and confidence before discussion on response changes after discussion. A total of 108 psychology students participated in the study, grouped into 36 online triads. The participants first solved eight reasoning tasks individually and assessed their confidence. They then discussed the tasks in groups. Finally, they solved the problems again individually. In five tasks, higher accuracy was achieved after a group discussion, and in three tasks there was no increase in accuracy. In the groups in which participants had given both correct and incorrect answers before the discussion, 54% of the incorrect answers were changed after the discussion, but 18% of the correct answers were also changed. Most of the correct answers were changed in groups where the correct answer was initially given by only one member of the group. The changes in response were not systematically related to confidence. Confidence judgments after the discussion were higher in groups with unanimous answers than in groups with different answers, and the highest confidence was achieved in groups in which all participants gave correct answers. The results obtained are discussed in the context of the argumentative theory of reasoning.