This document summarizes research on the Thatcher effect and its impact on face recognition. It begins with background on the Thatcher effect optical illusion and selecting the VGGFace2 dataset. It then describes two problems studied: 1) Detecting whether an image is "thatcherized" or normal using a neural network, achieving over 99% accuracy. 2) Assessing how thatcherized images impact face recognition training, finding that including more thatcherized images improves test accuracy. The document concludes future work could use generative models and other illusions to further understand robust face recognition.