Early identity recognition of familiar faces is not dependent on holistic processing

Abstract

It is widely accepted that holistic processing is critical for early face recognition, but recent work has suggested a larger role for feature-based processing. The earliest step in familiar face recognition is thought to be matching a perceptual representation of a familiar face to a stored representation of that face, which is thought to be indexed by the N250r event-related potential (ERP). In the current face-priming studies, we investigated whether this perceptual representation can be effectively activated by feature-based processing. In the first experiment, prime images were familiar whole faces, isolated eyes or isolated mouths. Whole faces and isolated eyes, but not isolated mouths, effectively modulated the N250r. In the second experiment, prime images were familiar whole faces presented either upright or inverted. Inverted face primes were no less effective than upright face primes in modulating the N250r. Together, the results of these studies indicate that activation of the earliest face recognition processes is not dependent on holistic processing of a typically configured face. Rather, feature-based processing can effectively activate the perceptual memory of a familiar face. However, not all features are effective primes as we found eyes, but not mouths, were effective in activating early face recognition.

Fig. 1

Study 1 Stimuli and ERP results. (FIRST ROW) The nine trial types used in Study 1 were three levels of prime (same, different, none) × three levels of feature (faces, eyes, mouth). (SECOND ROW) Grand-average ERPs (N = 24) across electrodes P₈, P₁₀, TP₈, P₇, P₉, TP₇. The time period of the P100 is indicated by the area with downward slashes. NOTE: this figure displays the waveform across the electrodes listed above, whereas the P100 statistics were done on the average waveform of electrodes C₁, C_z, C₂. The time period of the N170 is indicated by the solid shaded area. The time period of the N250r is indicated by the area with upward slashes. The waveforms are color coded such that RED is the prime_same condition, GREEN is the prime_different condition and BLUE is the prime_none condition. (THIRD ROW) Topographical maps of the N250r priming effect at each cell of the feature factor. The ‘Same’ maps display the prime_samevs prime_none simple contrast, whereas the ‘Different’ maps display the prime_differentvs prime_none contrast. (FOURTH ROW) N250r mean amplitude differences across electrodes P₈, P₁₀, TP₈, P₇, P₉, TP₇. The contrasts displayed are the same as for the topographical maps. Error bars indicate the 95% confidence interval. Asterisks indicate significant priming effects, P < 0.05.

Fig. 2

Study 2 Stimuli and ERP results. (FIRST ROW) The eight trial types used in Study 2 were two levels of prime (none, different) × two levels of feature (face, eyes) × two levels of orientation (upright, inverted). (SECOND ROW) Grand-average ERPs (N = 27) across electrodes P₈, P₁₀, TP₈, P₇, P₉, TP₇. The time period of the P100 is indicated by the area with downward slashes. NOTE: this figure displays the waveform across the electrodes listed above, whereas the P100 statistics were done on the average waveform of electrodes C₁, C_z, C₂. The time period of the N170 is indicated by the solid shaded area. The time period of the N250r is indicated by the area with upward slashes. The waveforms are color coded such that SOLID BLUE is the prime_difforientation_up condition, SOLID GREEN is the prime_noneorientation_up condition, DASHED BLUE is the prime_difforientation_inv condition and DASHED GREEN is the prime_noneorientation_inv condition. (THIRD ROW) Topographical maps of the N250r priming effect. The maps display the prime_diffvs prime_none simple contrast for each cell of the feature and orientation factors. (FOURTH ROW) N250r mean amplitude differences across electrodes P₈, P₁₀, TP₈, P₇, P₉, TP₇. The contrasts displayed are the same as for the topographical maps. Error bars indicate the 95% confidence interval. Asterisks indicate significant priming effects, P < 0.05.