國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.2 ERP Data
The N400 peak amplitude was defined as the most negative voltage within the interval of 300-500 ms. EEG data were only assessed at mid-line electrodes (Fz, Cz, and Pz) for the reason that N400 is the largest over central-prietal sites (Kutas &
Federmeier, 2011). Furthermore, two ERP peaks within the 60–210 ms (N100 and P200) time window were also analyzed. The N100 and P200 components were
quantified within 60–150 ms and 150–210 ms time windows, respectively, at Cz sites.
These three components were automatically detected and manually corrected where necessary. Three–way ANOVAs (2×2×3) was applied to compare the measures among two groups (Sleep vs. Wakefulness), two test sessions (pre- vs. post- test), and three pair types (OI, OR, and New). The peak amplitudes and latencies of N400 were compared among different conditions by ANOVA. An effect was considered as significant if p-value is less than 0.05. For post-hoc comparisons, the Scheffé tests were applied. In these analyses, the degrees of freedom associated with the F ratios of all ANOVAs were corrected by applying the Geisser-Greenhouse procedure to
compensate for the violation of non-sphericity.
3.3 Correlations
In order to explore the role of different sleep stages on memory formation, correlations between changes in performance of paired-associates learning and sleep stage variables were carried out using Pearson’s correlation analysis. Moreover, Correlations between changes of ERP components at distinct electrode sites and sleep stages were also obtained. A p-value less than 0.05 was considered significant.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER FOUR: RESUTS
1. Behavioral Data
1.1 Percentage of Correct Judgment (CJ%)
In CJ% analysis, a three-way ANOVA revealed a significant main effect of groups [F(1,28)=5.07, P<.05] and test sessions [F(1,28)=7.36, P=.01]. However, there was no significant main effect of pair type [F(2,56)=2.39, P=.10]. There were reliable groups × test sessions [F(1,28)=15.96, P<.01] and test sessions × pair types [F(2,56)=15.96, P<.01] interactions. Neither groups × pair types [F(2,56)=0.03, P=.97] nor groups × sessions × pair types interactions reached significance
[F(2,56)=0.73, P=.49].
There was no significance between pre-test and post-test sessions in the
Wakefulness group [F(1,44)=0.77, P=.39], but the Sleep group improved significantly [F(1,44)=34.68, P<.001]. The CJ% was not different between groups in pre-test session [F(1,89)=0.007, P=.93], but CJ% of Sleep groups was significantly higher than that of Wakefulness group in the post-test session [F(1,89)=21.45, P<.001]
(Figure 6).
1.2 Reaction Time (RT)
In RT analysis, there were main effects of groups [F(1,28)=12.35, P<.01], test sessions [F(1,28)=10.92, P<.01], and pair types [F(2,56)=64.30, P<.001]. In pairwise comparison, RTs of either New or OI pairs was significantly faster than those of OR pairs, whereas RTs of between OI and New pairs were not significantly different.
Groups × test sessions interaction reached significance [F(1,28)=29.46, P<.01].
Neither test sessions × pair types [F(2,56)=1.76, P=.18], groups × pair types [F(2,56)=0.85, P=.43] nor groups × sessions × pair types interactions reached
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
significance [F(2,56)=2.19, P=.12].
RT of Sleep group was faster after a period of sleep [F(1,44)=58.96, P<.01], whereas RT of Wakefulness group was not significantly different between pre- and post-test sessions [F(1,44)=4.24, P<.05]. Between groups, there was no significance difference at baseline [F(1,89)=1.92, P=.17], but RT of Sleep group was faster than that of Wakefulness group after sleep manipulation [F(1,89)=54.64, P<.01] (Figure 7).
2. ERP Data
Figure 8 presents the grand averaged waveforms elicited by OI, OR and New pairs for all 30 recording sites in Sleep group. Figure 9 shows the grand averaged
waveforms in Wakefulness group. The N400 appeared in central mid-line (Fz, Cz and Pz). Figure 10 demonstrates the grand averaged waveforms evoked by OI pairs for both Sleep and Wakefulness groups.
2.1 N1-P2 Components
An ANOVA analysis of the N1 peak amplitude at Cz site showed significant main effect of phases [F(1,28)=6.20, P<0.05] and phases × groups interaction
[F(1,28)=6.82, P<0.05]. There were no main effects of types [F(2, 56)=1.20, P=0.31]
and groups [F(1,28)=0.07, P=0.79]. Interactions of types × groups [F(2, 56)=0.33, P=0.72], phases × types [F(2, 56)=0.02, P=0.98], and phases × types × groups [F(2,
56)=0.17, P=0.85] did not reach significance. In Wakefulness group, the simple main effect of phases revealed significant difference [F(1,14)=19.07, P=.001], which demonstrated N1 in post-test session was attenuated than that in pre-test session.
For the analysis of P2 peak amplitude, only main effect of types showed
significant [F(2,56)=12.64, P<0.001]. Post hoc analysis showed that P2 elicit by new pairs was significantly smaller than that induced by OI pairs. There were no other
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
main effects of group [F(1,28)=0.05, P=0.82] or phases [F(1,28)=0.02, P=0.89].
There are no two way interactions of phases × groups [F(1,28)=1.48, P=0.23], types × groups [F(2,56)=1.20, P=0.31], or phases × types [F(2,56)=0.96, P=0.39]. Types × groups × test sessions interaction was not significant [F(2,56)=0.78, P=0.47].
2.2 N400 Peak Induced by OI Pairs at Fz, Cz and Pz Sites
An ANOVA showed main effects of groups [F(1,28)=82.20, P<.001] and test sessions [F(1,28)=6.32, P<.05], whereas ANOVA did not demonstrate main effect of electrode sites [F(2,56)=0.96, P=.39]. Both electrode sites × test sessions
[F(2,56)=7.11, P<.01] and group × test sessions [F(1,28)=19.33, P<.001] interactions were significant. Neither electrode sites × groups [F(2,56)=0.39, P=.68] nor electrode sites × groups × test sessions [F(2,56)=0.72, P=.49] interactions demonstrated any significance.
In Sleep group, the peak amplitude of N400 in pre-test session was more attenuated than that in post-test session [F(1,14)=15.27, P<.01]. Moreover, the attenuation of N400 after sleep significantly appeared at Cz [F(1,29)=7.52, P=.011]
and Pz [F(1,29)=4.96, P<.05] electrode sites, but not at Fz site [F(1,29)=0.41, P=.53].
In Wakefulness group, there was no significance between two test sessions
[F(1,14)=4.06, P=.06]. There was no significance of N400 changes between electrode sites [F(2,28)=2.67, P=.09]. Additionally, the N400s between groups were not
significantly different at baseline [F(1,89)=1.81, P=.18], whereas the N400 of
Wakefulness group was significantly smaller than that of Sleep group [F(1,89)=12.96, P=.001].
2.3 N400 Latency Induced by OI Pairs at Fz, Cz and Pz Sites
ANOVA showed only main effect of electrode sites [F(2,56)=3.54, P<.05]. Post
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
hoc comparisons revealed the N400 latency of electrode Cz (364±6.18 msec) was shorter than that of electrode Fz (386.80±10.63 msec). No other main effect was observed for groups [F(1,28)=1.05, P=.31] and test sessions [F(1,28)=0.21, P=.06].
There was no significant two way interaction, including electrode sites × groups [F(2,56)=2.67, P=.09], electrode sites × sessions [F(2,56)=0.10, P=.91] and groups × sessions [F(1,28)=0.005, P=.94], or three way electrode sites × group × test sessions interaction [F(2,56)=0.62, P=.54].
2.4 The Peak Amplitude of N400 Evoked by Different Pairs Types at Cz site The N400 at Cz sites was more obvious than that at other sites, so the N400s induced by different pair types were analyzed by ANOVA. An ANOVA showed the main effect of pair types [F(2,56)=36.27, P<.001], test sessions [F(1,28)=5.85, P<.05]
and groups [F(1,28)=136.33, P<.001]. Figure 12 showed the N400s induced by different types in both test sessions.
All two-way interactions of groups × pair types [F(2,56)=4.86, P<.05], test sessions × pair types [F(2,56)=12.46, P<.001] and test sessions × groups
[F(1,28)=8.38, P<.05] were significant. There was a three way interaction of group × test sessions × pair types [F(2,56)=15.44, P<.001]. The N400s between groups in three distinct pair types were not significantly different at baseline [F(1,29)=0.76, 0.69 and 0.77 for OI, OR and New respectively, all Ps>.05]. After sleep manipulation, only N400 induced by OI pairs in Sleep group was significantly smaller than that in Wakefulness group [F(1,29)=33.44, P<.05] (Figure 11). The N400s of OR pairs between Sleep and Wakefulness groups were marginally significant [F(1,29)=3,99, P=.056].
In ERP analysis of Sleep group, the simple main effect of test session reached a significance [F(1,14)=10.07, P<.01]. Post-hoc comparison demonstrated that N400
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
was significantly attenuated after sleep (P<.01). Furthermore, the simple main effect of pair types also revealed a significance [F(2,28)=21.97, P<.001]. Post-hoc
comparison demonstrated that the peak amplitude of N400 induced by New pairs was significantly obvious, whereas N400 induced by OI pairs was smaller than that
induced by other types (all Ps<.01). Furthermore, N400 of both OI and OR pairs were attenuated after sleep [F(1,29)=7.52 and 5.38, respectively; both P<.05], whereas N400 of New pairs did not change significantly [F(1,29)=0.44, P=.55] (Figure 12A).
In Wakefulness group, the simple main effect of pair types revealed significant difference [F(1,14)=24.23, P<.001]. Post-hoc comparison showed that the peak amplitude of N400 was decreased gradually in order of which induced by New, OR and OI pairs (all Ps<.01). Either simple main effect of test session [F(1,14)=0.19, P=.67] or interaction of test session × types [F(1,14)=0.08, P=.78] was significant
(Figure 12B).
2.5 The Peak Latency of N400 Induced by Different Word Pairs at Cz site Analysis of N400 latency induced by different pairs revealed no main effects for two variables: main effect of test sessions [F(1,28)=0.48, P=.49], main effect of groups [F(1,28)=0.12, P=.73], and main effect of pair types [F(2,56)=1.35, P=.28].
However, main effect of pair types reached a significance [F(2,56)=5.14, P<.01].
There were no significant interactions of neither groups × pair types [F(2,56)=0.18, P=.85], groups × pair types [F(1,28)=0.24, P=.63], nor test sessions × pair types
[F(2,56)=1.35, P=.27]. Three-way interaction of group × test sessions × pair types was not significant [F(2,56)=1.04, P=.36]. Post hoc comparison showed that the N400 latency of new pairs was significantly longer than that of either OI or OR pairs.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3. Correlations Between Sleep Stages and Parameters of Memory Measures