Table 2.
The relation between EEG response amplitudes and the degree of realness when controlling for the confounding factor eye size.
| Signal | Channel | Model (L/Q) |
AIC | BIC | Log-likelihood | LRT |
|---|---|---|---|---|---|---|
| SSVEP | Oz | L | 486.45 | 505.88 | − 238.22 | χ2 = 9.535, |
| Q | 478.91 | 502.23 | − 233.46 | p = 0.002* | ||
| Cluster | L | 311.60 | 331.03 | − 150.80 | χ2 = 12.303, | |
| Q | 301.30 | 324.62 | − 144.65 | p < 0.001* | ||
| SSD | L | 873.96 | 893.39 | − 431.98 | χ2 = 13.521, | |
| Q | 862.44 | 885.75 | − 425.22 | p < 0.001* | ||
| N170 | PO8 | L | 451.62 | 471.05 | − 220.81 | χ2 = 18.529, |
| Q | 435.09 | 458.40 | − 211.54 | p < 0.001* | ||
| Cluster | L | 537.15 | 556.58 | − 263.58 | χ2 = 20.634, | |
| Q | 518.52 | 541.84 | − 253.26 | p < 0.001* |
L and Q refer to the linear and quadratic models, respectively. LRT is the likelihood ratio test. Akaike information criterion (AIC), Bayesian information criterion (BIC) are indices of model fit. Likelihood is the log transformed likelihood. The regression models with lower AIC, lower BIC, and higher likelihood indicate a better model fit. SSD refers to the results after applying SSD filters (*Significance level: p < 0.05).