A number of computational techniques have been proposed that aim to detect mimicry in online conversations. In this paper, we investigate how well these reflect the prevailing cognitive science model, i.e. the Interactive Alignment Model. We evaluate Local Linguistic Alignment, word vectors, and Language Style Matching and show that these measures tend to show the features we expect to see in the IAM, but significantly fall short of the work of human classifiers on the same data set. This reflects the need for substantial additional research on computational techniques to detect mimicry in online conversations. We suggest further work needed to measure these techniques and others more accurately.