The face recognition literature has shown that it is easier to remember a full face and an upper region of a face than it is to remember a lower region. A major thesis underlying this study was that this difference in memory is due to greater information based upon greater variations in the universe of faces involving the upper region rather than the lower region. Hence, the lower region is less discriminable than the upper region. Two experiments were conducted to test this hypothesis. Experiment 1 investigated the discriminability of parts of face (full, upper, lower) and expressions (neutral, smile, pout) in memory. Thus, ninety adult subjects saw a series of 50 pictures and were required to keep a mental tally of each new face that appeared in the series. A subtally was recorded after the first 25 pictures and a final tally after the second 25 pictures. More individuals were discerned when the subjects saw a series of 50 full faces than when they saw 50 upper regions (p < .01) and 50 lower regions (p < .01). Upper region faces produced a higher tally than lower region faces (p < .01). Also, more individual faces were discerned during the first half of the series than during the second half (p < .01). Part of face interacted with series length (p < .01). Series length affected the lower region most, full face least, and the upper region intermediately. Experiment 2 investigated whether the results of Experiment 1, concerning parts of face and expressions, were attributable to memory or would occur with direct sorting. Ninety new adult subjects sorted 25 pictures into categories containing pictures of the same person. The results again showed that more individuals were discerned for full faces than upper regions (p < .05) and lower regions (p < .01). Upper regions were sorted into more categories than were lower regions (p < .01). Although the individual analyses for each study failed to produce an effect of expression, combining the data for the two experiments yielded a significant main effect (p < .05). More individuals were discerned for neutral and smiling faces than for pouting faces (p < .05 for all comparisons). The findings suggest that recognition differences between facial regions are based upon differences in discrimination at input.
展开▼
