Inferring Events from Phylogenetic Analysis of Eyewitness Testimony in the Historical and Forensic Record

This project will consider if we can use phylogenetic analysis to correct for imperfections in eyewitness testimony. Eyewitness testimony from events with multiple witnesses is often inconsistent and sometime contradictory. If we think of these inconsistencies as mutations from a common ancestor with perfect recall (but not necessarily perspective) on the event, can we obtain an objective estimate of what the collected witnesses saw? Here we will consider historical and forensic records of events with multiple witnesses, in order to construct a “family tree” of what was reported and estimate the ancestral condition. Can statistics tell us what really happened when JFK was assassinated? Could statistics inform analysis of crimes where there are multiple witnesses? Both manual and machine learning approaches will be considered.