Is the Thatcher Illusion Modulated by Face Familiarity? Evidence from an Eye Tracking Study

Thompson (1980) first detected and described the Thatcher Illusion, where participants instantly perceive an upright face with inverted eyes and mouth as grotesque, but fail to do so when the same face is inverted. One prominent but controversial explanation is that the processing of configural information is disrupted in inverted faces. Studies investigating the Thatcher Illusion either used famous faces or non-famous faces. Highly familiar faces were often thought to be processed in a pronounced configural mode, so they seem ideal candidates to be tested in one Thatcher study against unfamiliar faces–but this has never been addressed so far. In our study, participants evaluated 16 famous and 16 non-famous faces for their grotesqueness. We tested whether familiarity (famous/non-famous faces) modulates reaction times, correctness of grotesqueness assessments (accuracy), and eye movement patterns for the factors orientation (upright/inverted) and Thatcherisation (Thatcherised/non-Thatcherised). On a behavioural level, familiarity effects were only observable via face inversion (higher accuracy and sensitivity for famous compared to non-famous faces) but not via Thatcherisation. Regarding eye movements, however, Thatcherisation influenced the scanning of famous and non-famous faces, for instance, in scanning the mouth region of the presented faces (higher number, duration and dwell time of fixations for famous compared to non-famous faces if Thatcherised). Altogether, famous faces seem to be processed in a more elaborate, more expertise-based way than non-famous faces, whereas non-famous, inverted faces seem to cause difficulties in accurate and sensitive processing. Results are further discussed in the face of existing studies of familiar vs. unfamiliar face processing.