Flynn limited the possible explanations when he looked careful at the test data and discovered that the improvement in scores was on certain parts of the IQ test. Test- takers didn't do better on the arithmetic or vocabulary sections of the test they did better on sections that required a special kind of reasoning and problem solving. For example one part of the test shows a set of abstract shapes, and test-takers must look for patterns and connections between them and decide which shape should be added to the set.