Background: Psychological research has largely overlooked that there is variability between photographs of the same person and as a result facial recognition studies have used randomly selected images to represent individuals. Thesis aim: This study therefore aimed to investigate if the variations between images are significant and if these differences influence the caricature and prototype effects. Method: Experiment 1 was designed to investigate image likeness and it found that variations between photographs of the same person were significantly different. This enabled photographs to be classified into good or bad likeness categories and allowed the main thrust of the research to be completed. Experiments 2 and 3 were name verification tasks. These tasks investigated accuracy and response time in relation to likeness, (good, bad) image type (prototypes, photographs) and caricaturing (caricatured, veridical (not caricatured)). Predictions: The first expectation of the study was that likeness would affect the caricature effect; the second was that prototypes would not be affected by likeness. Results: The experiments demonstrate that the degree of image likeness affects both caricature and prototype effects. Accuracy benefits were found for good likeness caricatures compared to veridical at 33ms (Exp 2) however bad likeness caricatures improved accuracy at 100ms (Exp 3) when compared to veridical images. Good likeness photographs were more accurately identified than prototypes at 33ms (Exp 2) and 100ms (Exp 3). Bad likeness photographs were more accurately identified than prototypes at 33ms, but prototypes were more accurately identified at 100ms (Exp 3). Conclusions: This study demonstrates that photographs of the same person cannot be classified as equivalent to one another. Degrees of image likeness must be taken into consideration in subsequent studies as likeness influences aids of facial recognition at different timings and to different degrees. Keywords: Caricatures, facial recognition, likeness, photographs, prototypes, veridical.