Individual differences in detecting rapidly presented fearful faces

Abstract

Rapid detection of evolutionarily relevant threats (e.g., fearful faces) is important for human survival. The ability to rapidly detect fearful faces exhibits high variability across individuals. The present study aimed to investigate the relationship between behavioral detection ability and brain activity, using both event-related potential (ERP) and event-related oscillation (ERO) measurements. Faces with fearful or neutral facial expressions were presented for 17 ms or 200 ms in a backward masking paradigm. Forty-two participants were required to discriminate facial expressions of the masked faces. The behavioral sensitivity index d' showed that the detection ability to rapidly presented and masked fearful faces varied across participants. The ANOVA analyses showed that the facial expression, hemisphere, and presentation duration affected the grand-mean ERP (N1, P1, and N170) and ERO (below 20 Hz and lasted from 100 ms to 250 ms post-stimulus, mainly in theta band) brain activity. More importantly, the overall detection ability of 42 subjects was significantly correlated with the emotion effect (i.e., fearful vs. neutral) on ERP (r = 0.403) and ERO (r = 0.552) measurements. A higher d' value was corresponding to a larger size of the emotional effect (i.e., fearful--neutral) of N170 amplitude and a larger size of the emotional effect of the specific ERO spectral power at the right hemisphere. The present results suggested a close link between behavioral detection ability and the N170 amplitude as well as the ERO spectral power below 20 Hz in individuals. The emotional effect size between fearful and neutral faces in brain activity may reflect the level of conscious awareness of fearful faces.

Figure 1. Schematic diagram of one experimental trial.

Figure 2. The grand-mean ERPs of all 42 subjects for the frontal N1 and occipital P1 components.

The scalp topographies were computed at the time interval of 115 to 135 ms. The three white dots in each topography indicate which three electrodes recorded the most prominent electrical activity. F, fearful faces; N, neutral faces.

Figure 3. The grand-mean ERPs of all 42 subjects for the N170 component.

The scalp topographies were computed at the time interval of 160 to 180 ms. The two white dots in each topography indicate which two electrodes recorded the most prominent electrical activity.

Figure 4. The scatter plots between the behavioral measure d' and the amplitude differences between fearful and neutral faces of the N1 (recorded at FCz), P1 (recorded at O1 and O2), and N170 (recorded at P7 and P8) in response to 17-ms presentations.

Significant linear correlation is indicated by dashed line in associated color.

Figure 5. The grand-mean ERSPs of all 42 subjects at P7 and P8 electrodes.

Single-participant ERSP was computed for neutral, fearful and fearful-neutral conditions, averaged across 42 subjects. Nongreen areas in the time/frequency plane show significant (p<.05) post-stimulus increases (red) or decreases (blue) in log spectral power relative to mean log power the 200-ms pre-stimulus baseline. Vertical dotted lines indicate the N170 latency in response to neutral (magenta line) and fearful faces (black line).

Figure 6. The scatter plots between the behavioral measure d' and the ERSP differences (fearful ERSP – neutral ERSP) at P7 and P8 electrodes in response to 17-ms presentation.

Significant linear correlation is indicated by dashed line in associated color.

Figure 7. Statistical ERSP comparison between fearful and neutral faces in response to 17-ms facial presentations (five or six participants in each group).

The 42 subjects were divided into 8 groups according to d' values at 17 ms. The number of participants and the mean d' value in each group are displayed in the right column. Single-participant ERSP differences were computed by subtracting the ERSP in response to fearful faces by the ERSP in response to neutral faces, averaged across five or six participants.