11.4: Design of the N170 Experiment
selected template will load here
This action is not available.
This section will provide a brief overview of the experimental design and main results from the ERP CORE N170 experiment, which was based on a prior study by Rossion and Caharel (2011). The N170 is typically found to be larger for faces than for almost any class of non-face stimuli. In this experiment, cars were used as the non-face stimuli. We also presented phase-scrambled faces and cars, which contain the same low-level information as the faces and the cars but without the higher-level features that are essential for discriminating between the broader classes of faces and cars.
The N170 effect is largely independent of what task the participants are performing, but it can be helpful to have participants perform some kind of task to keep them alert and attentive. In this experiment, participants were instructed to discriminate whether a given stimulus was an intact image or a scrambled image. Specifically, they pressed one button for faces and cars and another button for scrambled faces and scrambled cars. These four stimulus classes were presented in random order. Stimulus duration was 300 ms, and a blank interstimulus interval of 1100-1300 ms occurred between stimuli.
We used 40 different face images, 40 different car images, and scrambled versions of each of these images. Each of the 160 images was presented once. A different event code was used for each of the 160 images, as indicated in Table 11.1.
|
|
|
Event Code |
|
Stimuli |
Faces |
1–40 |
|
Cars |
41–80 |
|
|
Scrambled Faces |
101–140 |
|
|
Scrambled Cars |
141–180 |
|
|
Accuracy |
Event Code |
|
|
Responses |
correct |
201 |
|
incorrect |
202 |
Of the 40 participants who were tested, three had to be excluded because of artifacts. The grand average waveforms from the remaining 37 participants are shown for the face and car stimuli in Figure 11.1.B. Note that, for the sake of simplicity, the exercises in the present chapter include only Subjects 1-10. Subject 5 was one of the excluded subjects, so the final analyses in this chapter are based on the data from only 9 participants.
As illustrated in Figure 11.1.B, the peak of the N170 was slightly greater (more negative) for faces than for cars. The main difference between the waveforms appeared to be a faster onset latency for the faces. However, this means that the amplitude was greater for faces than for cars between approximately 110 and 150 ms. This is illustrated in the faces-minus-cars difference wave shown in Figure 11.1.C. When I first saw these results, I thought we had done something wrong. However, this pattern is actually quite common in N170 experiments.
When you compare the ERPs elicited by faces and cars, it’s possible that any differences in the waveforms could be a result of differences in low-level features (e.g., luminance, spatial frequency) that are not actually important in perceiving whether a stimulus is a face or a car. This is why the experiment included phase-scrambled face and car images, which contain the same low-level features as the faces and cars but do not contain the higher-level features that we used to determine whether an image is a face or a car.
Figure 11.1.D shows the ERPs elicited by the faces and the scrambled faces, along with the faces-minus-scrambled-faces difference wave. This difference wave is designed to subtract out any brain activity related to the low-level features shared by faces and scrambled faces. You can see a nice N170 in this difference wave. Figure 11.1.E shows the corresponding waveforms for cars, scrambled cars, and the cars-minus-scrambled-cars difference. You can again see an N170, but it’s smaller and later than the N170 in the faces-minus-scrambled-faces difference wave.