The Face Inversion Effect (FIE), the finding that inversion disproportionately affects face recognition, is one of the primary pieces of evidence suggesting that faces are encoded in a qualitatively different way to other stimuli (e.g., along configural as well as featural dimensions). However, when Loftus, Oberg and Dillon (2004) tested the FIE using state-trace analysis (Bamber, 1979), they found evidence for a one dimensional encoding of unfamiliar faces when inversion only occurred during the study phase of a recognition memory test. We report experimental results that replicate Loftus et al.’s findings and rule out several potential problems with their experimental manipulations and state-trace analysis.