4.2.1. Time-frequency Analysis of Independent Clusters
The components decomposed from EEG signals were grouped into six clusters: mid-frontal, central, left and right sensorimotor, parietal and occipital.
The frontal ERSP revealed that the theta band power increased during the Planning stage, rose to its peak at the Searching stage, and then gradually returned to baseline during the Cruising stage (Figure 16). The other cluster of component ERSPs showed that power slightly (central) or visibly (left and right sensorimotor, parietal and occipital) suppressed around 10 Hz and 20 Hz (alpha blocking) between target onset and arriving at target then rebounded during resting.
48
Figure 16. This figure showed the scalp maps, dipoles locations and ERSPs of six grouped clusters: frontal (a), central (b), left (c) and right sensorimotor (d), parietal (e) and occipital (f). The areas between broken lines from left to right represent baseline (0~2 s), plan (2~4 s), search (4~7 s), cruise (7~11 s), and rest stages (11~13 s), respectively. The left curve is baseline spectra power and the center color map indicates frequency power difference related to baseline power.
The frontal theta band power increased during the Planning stage and rose to the peak at search stage then slowly returned to baseline power during the Cruising stage. And the other components ERSPs showed alpha blocking (suppression) between target onset and arriving at target. The color of dipoles indicated different conditions: male in regular (blue) and irregular (sky blue) mazes and female in regular (red) and irregular (yellow) mazes.
49
4.2.2. Power Spectra Analyses
Participants whose data contained cluster components from ICA decomposition (frontal: 6 males and 7 females; central: 3 males and 7 females;
left sensorimotor: 4 males and 5 females; right sensorimotor: 5 males and 7 females; parietal: 3 males and 5 females; occipital: 4 males and 7 females) were included in further statistical analyses. Therefore, forty two separate ANOVA were carried out on the combinations of frequency bands and clusters, respectively. These result were displayed on Table 5 (see Appendix 1 for detal interaction information) and grouped distributively in each condition as follows.
50
Table 5. Cluster frequency bands showing significant main effects are indicated with asterisks in the corresponding cell.
Main effect Description
51
*p<.05, **p<.001
o: There were interactions.
P: plan, S: search, C: cruise, G: global, L: local, F: female, M: male, R: regular and I: irregular.
These power differences were grouped into Stage (a1: frontal theta; a2: alpha and beta in the other clusters; a3: frontal beta; a4: delta in most clusters), Gender (b), Landmark (c) and Structure (d).
The capital letter meant the comparison was executed in this condition, and the lower case indicated the significant difference (p < .05) and relations between these conditions. Only significant differences were shown in the table.
4.2.3. Stage
The relative power of frontal in ERSP showed theta increase and its ANOVA results also revealed significant main effect of Stage in the frontal theta band (F(2, 22) = 22.05, p < .001, ηp
2 = .67) and no significant interactions between Stage and other factors (Table 5a1). The theta power (Figure 17a) of the Searching stage (3.24) was larger than that of the Cruising stage (2.22), which was in turn larger that the Planning stage (1.55).
The other cluster (except frontal) demonstrated the alpha blocking also revealed significant main effect of Stage in the alpha (F(2, 14) = 34.72, p < .001, ηp
2 = 0.83), beta1 (F(2, 14) = 25.44, p < .001, ηp
2 = .78) and beta2 band (F(2, 14)
= 43.74, p < .001, ηp
2 = .86) in left sensorimotor (counter-lateral of control hand) and no significant interactions between Stage and other factors (Table 5a2). The alpha, beta1 and beta2 power of the Planning stage (-2.63, -2.55 and -1.92) were larger than that of the Searching (-3.54, -3.73 and -2.97) and Cruising stages (-3.87, -3.85 and -3.04), Furthermore, most alpha and beta power of Planning stage were also significant larger than that of the Searching or Cruising stages in central, right sensorimotor, parietal and occipital (Table 5a2). Although the
52
power difference between Stages not reached significant level in few conditions (female central alpha, male right sensorimotor alpha and parietal beta2), this also had similar trend revealed the largest alpha and beta power of the Planning stage. Moreover, the parietal and occipital theta power had the similar difference of stages with alpha power and revealed significant main effects of Stage in the theta band (parietal: F(2, 12) = 8.60, p < .01, ηp
2 = .59; occipital: F(2, 18) = 34.83, p < .001, ηp
2 = .80) and no significant interactions between Stage and other factors (Table 5a2). The parietal and occipital theta power of the Searching stage (-0.32 and -1.01) were larger than that of the Cruising stage (-0.94 and -2.02).
In the other obvious effect, there was a significant main effect of Stage in the frontal beta1 band (F(2, 22) = 12.07, p < .001, ηp
2 = .52) and no significant interactions between Stage and other factors (Table 5a3). The beta1 power (Figure 17a) of the Planning (0.28) and Searching stages (-0.01) were larger than that of the Cruising stage (-0.36), Further, there was also a significant main effect of Stage in the frontal beta2 band (F(2, 22) = 7.66, p < .01, ηp
2 = .41) and a significant interaction between Stage and Landmark (F(2, 22) = 4.48, p < .05,
η
p2= .29). Post hoc analysis indicated the beta2 power of the Planning stage (-0.23) was significant larger than that of the Cruising stage (-0.61) in Global landmark and this power of the Planning stage (-0.47) was only showed a larger trend than that of Cruising stage (-0.60) in Local landmark.
Furthermore, the delta band power demonstrated differences between stages across some clusters (Table 5a4). There were significant main effects of Stage in the frontal (F(2, 22) = 25.07, p < .001, ηp
53
stage (-0.09 and -1.71). Moreover, most delta power of the Planning stage were significant larger than that of the Cruising stage in right sensorimotor and parietal, and the other condition (male parietal) delta power of the Planning stage (-1.05) was also showed a larger trend than that of Cruising stage (-1.58).
4.2.4. Gender
In the left sensorimotor component, the main effect of Gender was significant (F(1, 7) = 6.62, p < .05, ηp
2 = .49) in the gamma1 band power (Figure 17c). No significant interactions were found for Gender and other factors (Table 5b). Males’ frequency power (-0.56 was larger than females’ (-1.24). In the right sensorimotor component, the main effect of Stage in the gamma1 band power was significant (F(2, 20) = 7.96, p < .01, ηp
2 = .44) and there was a significant interaction between Stage and Gender (F(2, 20) = 7.23, p < .01, ηp
2 = .42). This interaction was due to the diversity that there was significant Stage difference (the gamma1 power of the Planning stage (-0.60) were larger than that of the Searching (-1.27) and Cruising stages (-1.34) for female but not significant for male.
In the frontal component, although there was no significant difference (p
> .05) between theta power of Gender, there was still a trend indicated female’s theta power (2.88) was larger than male’s (1.79). If the frequency band narrowed to 6 and 8 Hz (Figure 17b), this power difference of Gender would reach significant level (F(1, 11) = 5.07, p < .05, ηp
2 = .32). And there were also no significant interactions between Gender and other factors for this frequency band. For this high-theta band, females’ frequency power (3.61) was larger than males’ (2.00).
4.2.5. Landmark
The relative power of Landmark differed for the delta band power across most clusters (Table 5c). The main effects of Landmark were significant for the
54
There were no significant interactions between Landmark and other factors. The frequency power of frontal (Figure 17e), central and occipital (Figure 17f) were larger in the Global landmark (0.46, 0.24 and -0.69) than in the Local landmark (-0.04, -0.15 and -1.26) condition. Further, there was a significant interaction between Stage and Landmark (F(2, 14) = 3.93, p < .05, ηp
2 = .36) in left sensorimotor. This interaction was due to the difference that the Global delta power (-0.05) was significant larger than Local power (-0.44) in Planning stage but similar with Local power in Searching (-0.47 vs. -0.44; p > .05) and Cruising stages (-0.41 vs. -0.39; p > .05). In the right sensorimotor, there were also significant interactions between Structure, Gender and Landmark (F(1, 10) = 15.05, p < .01, ηp
2 = .60), and between Structure and Landmark (F(1, 10) = 21.74, p < .001, ηp2
= .69). Post hoc analysis indicated the delta power difference of Landmark only reached significant level (-0.39 for Global; -1.22 for Local; p
< .05) for male in regular maze but not significant in the other conditions.
4.2.6. Structure
No significant main effect associated with Structure was found (all p > .05) for each cluster frequency band. There were only significant interactions between Stage, Gender and Structure (F(2, 16) = 3.73, p < .05, ηp
2 = .32), and between Stage and Structure (F(2, 16) = 6.43, p < .01, ηp
2 = .45) for the gamma2 band power in central, and there was also a significant interaction between Stage and Structure (F(2, 18) = 4.26, p < .05, ηp2
= .32) and a significant main effect of Stage (F(2, 18) = 8.97, p < .01, ηp
2 = .50) for the gamma1 band power in occipital (Table 5d). In the central cluster, these interaction were due to the difference that the gamma2 power was significant larger in the Regular structure (-0.08) than that in the Irregular structure (-0.46) for male in Planning stage but not significant in the other conditions. Because of three numbers of males in this
55
cluster, this difference was not discussed later. In the occipital cluster, post hoc analysis revealed that the gamma1 band power of Planning stage (-0.29) was larger than that of the Searching (-0.82) and Cruising stages (-0.88) in the Regular structure and this frequency band power of Planning (-0.38) and Cruising stages (-0.44) were larger than that of the Searching stage (-0.80) in the Irregular structure maze. Then the gamma1 band power of Irregular structure was also larger than that of Regular structure during the Cruising stage.
Figure 17. This figure showed the relative power spectra of conditions which included stage (a), gender (b and c), structure (d) and landmark (e and f). The gray areas indicated the power were significant difference (four-way
mixed-design ANOVA) during this frequency band. The scalp maps indicated these relative power spectra were from frontal (a, b, d and e), left sensorimotor (c) and occipital (f).
56
V Discussions
This study performed the wayfinding tasks of learning and retrieval and analyzed the behavior data of learning task as well as EEG signals of retrieval.
These results were discussed as follows: