EEG theta and N400 responses to congruent versus incongruent brand logos

Abstract

Neuroimaging and behavioral studies have shown that brands convey meaning to consumers. To investigate the immediate reactions of the brain to brand logos, followed either by congruent or incongruent pictorial brand-related cues, can deepen understanding of the semantic processing of brands, and perhaps how consolidated the logo is in consumers' minds. Participants were exposed to different brand-related image sets, that were either congruent (a match between brand-related images and brand logo) or incongruent (a mismatch between brand-related images and brand logo) while having their brain signals recorded. Event-related potential and EEG time-frequency domain features were extracted from the signals of the target image (brand logo). The results showed significantly larger N400 peak and relative theta power increase for incongruent compared to congruent logos, which could be attributed to an error-monitoring process. Thus, we argue that brands are encoded deeply in consumers' minds, and cognitive processing of mismatched (vs matched) brand logos is more difficult, leading to greater error monitoring. The results were mostly consistent with previous studies investigating semantic incongruences in the linguistic field. Therefore, the error-monitoring process could be extended beyond linguistic forms, for example to images and brands.

Figure 1

ERP obtained from average on trials of each condition and specific regions, and topo-map of N400 activity. Panels (a) to (e) show ERP activity for frontal, central, parietal, and occipital regions respectively. In all panels from (a) to (d), red curves show the ERP of congruent condition, and blue curve shows ERP of incongruent condition. The shaded area around each line indicates standard deviation of signals divided by square root of number of channels. In the same panels, the vertical red line in 0 ms indicates the start of the stimuli. The small topo-map at the bottom-left of each panel shows N400 activity and indicates the selected electrodes of each region. In panel (c), for the central region, the area highlighted in gray shows the significant difference between congruent and incongruent conditions (p = 0.04, effect size = 0.54), that occurs in N400. Panel (e) shows the difference of brain activity in two conditions (congruent-incongruent) in N400 (averaged from 400 to 600 ms). The hot colors show positive activity (i.e., congruent > incongruent) and the cold colors indicate negative activity (i.e., congruent < incongruent).

Figure 2

Time–frequency and topo-map (of theta band in central region) for congruent and incongruent conditions, calculated from brain activity of central region. Hot colors show positive relative power and cold colors show negative relative power. Panels (a) and (c) show time–frequency of congruent and incongruent conditions respectively. Both panels are plotted from − 0.75 to 2.75 s. The y-axis indicates frequency and the x-axis indicates time. The vertical black line in 0 s indicates the start of the stimuli. Panels (b) and (d) show topo-maps of brain activity in theta band of congruent and incongruent conditions respectively. The activity is averaged over theta (4 to 7 Hz) and time (0 to 2 s), and plotted as topo-maps.

Figure 3

Time–frequency differences (congruent-incongruent), topo-maps (of theta band), and power spectrums. Hot colors show positive relative power and cold colors show negative relative power. Panels (a) and (c) show the subtracted TF of incongruent condition from congruent condition (congruent-incongruent). The white rectangle shows the region of theta frequency (4 to 7 Hz) and time (700 to 1200 ms), where the difference is maximum. The statistics, topo-maps, and power spectrum of this figure are all focused on this region. Panel (a) shows the TF difference in mid-frontal region and panel (c) shows the TF difference in central region. Panels (b) and (d) show topo-maps for mid-frontal and central regions respectively. They show brain activity averaged on the theta frequency and 700 to 1200 ms (the area of white rectangular). Panels (e) and (f) show the power spectrum activity of mid-frontal and central regions respectively. They show power spectrum of all frequencies averaged over time. The gray regions in both panels indicate the theta band, and there is significant difference between congruent and incongruent conditions both in mid-frontal (p = 0.03, effect size = − 0.54) and central (p = 0.01, effect size = − 0.69) regions. In both panels, the yellow curve shows the power spectrum of congruent, the red curve shows the power spectrum of incongruent, and the black curve shows the average power spectrum of congruent and incongruent conditions.

Figure 4

Example for stimuli and the task procedure. Panel (a) upper part shows an example for congruent set, where the last image (brand logo) matches the brand of the first three images. Panel (a) bottom part is an example for incongruent set where the last image and first three images do not match. Panel (b) shows the task procedure, which started with 12 s of instruction followed by congruent and incongruent image sets (randomized through the experiment). After each image set, there was a 3-s fixation cross.

Figure 5

Shows the steps in our methodology. The plots related to congruent condition are indicated by red color, and incongruent condition by blue color. Panel (a) shows congruent and incongruent target brands (the last image of each set). Panel (b) shows an example of the EEG data followed by pre-processing steps. Panel (c) relates to pre-processed epochs corresponding to each condition. By averaging the epochs of each condition we reached panel (c-1), showing the ERP activity for each condition. Panel (d) shows the time–frequency activity of each epoch for two conditions separately (in each TF, hot colors indicate positive relative power and cold colors indicate negative relative power). By averaging of TF of each condition, panel (d-1) is obtained, showing the actual time–frequency activity of each condition. Panel (e) shows the statistical analysis using the permutation test—starting with actual difference calculation, then shuffling the extracted feature 10,000 times and calculating the difference in each iteration, and finally building up the random distribution and comparing it with the actual difference.