At school, geometry caused me severe distress, untold hours of wasted time and migraines. I did not know how to do it. Give me a protractor to measure an angle, or a pair compasses to bisect it, and I was fine. Give me a problem to solve that some people could do in two minutes, and I was up until eleven at night, having started at 5.30 or 6, and completely, hopelessly, despairingly confused. I did not know what constituted a proof, or how to apply whatever theorems we'd been shown. The misery caused by having to do something, and not being able to do it, has, I'm told, given me some understanding of how people feel when they can't read or write properly, but that, at the time, was a long way off. All I knew was that geometry homework gave me serious grief, and I dreaded Thursday nights.
My school's solution at the time was to introduce a new option at GCSE, statistics, which amounted to illustrated arithmetic, that I enjoyed and could do. I learned the principles of geometric (Euclidian) proof twenty years later through Davis and Hersh's The Mathematical Experience, but that was too late to do anything about the initial problem. I couldn't do geometry, and, like many others, concluded that I was useless at maths.
Sahar Bokosmaty's thesis, open access here, shows how giving students worked examples before expecting them to solve problems on their own, helps free them from their own instinctive approach to problem solving - my protractor - and enables them to understand the principles of mathematical thinking. Part of the process involves reducing the load on working memory by building up longer term knowledge and understanding. This is important work, and has been published in article form here, though unfortunately only the abstract is available freely on the internet. I suggest that people interested read it via a copyright library, as I have thanks to the excellent service at Cambridge University Department of Education Library.