Electroencephalography and clinical Neurophysiology, 84 (1992) 101- 109
101
Elsevier Scientific Publishers Ireland, Ltd.
EVOPOT 91115
Steady-state visual evoked responses in high and low alpha subjects Ross A. Pigeau and Andrea M. Frame Defence and Civil Institute of Environmental Medicine, North York, Ont. (Canada)
(Accepted for publication: 19 October 1991)
Summary Reactivityto photic stimulation was studied in 16 subjects who were divided into 2 groups based on the relative amounts of spontaneous EEG alpha produced during a 4 min eyes-closed baseline condition. During the experimental session the subjects were presented with thirteen 1 min periods of sinusoidally modulated light stimulation at frequencies ranging from 2 Hz below their spontaneous peak alpha frequency (SPAF) to 2 Hz above in 0.33 Hz intervals. Using FFTs, steady-state visual evoked responses (VERs) were extracted from each subject's EEG for each condition. VER magnitude for high alpha subjects varied with the proximityof the stimulus frequency to the SPAF, the VER for low alpha subjects did not. Conversely, low alpha subjects showed a similar effect in side-band alpha activity(once the VER had been extracted) whereas high alpha subjects did not. The results are explained in terms of a possible difference in the coupling strength between thalamic and cortical alpha sources. Key words: Alpha generators; Steady-state VER; EEG; Alpha
Alpha is the prominent electroencephalographic wave form for normal, relaxed adults with eyes closed. It is characterized by 8-13 Hz activity manifesting over posterior regions of the brain with the largest voltages centred in the occipital cortex (IFSECN 1974). A dominant alpha frequency varying about 10 Hz (Peters6n and Eeg-Olofsson 1971; Osaka 1984) is often displayed with a generalized response of side-band activation within approximately + 1 Hz (e.g., Schwibbe et al. 1981). There is wide variation among individuals in their alpha response. It can be produced as continuous trains of high or low voltage activity and can appear quite sinusoidal (i.e., spindles) or irregular (Lindsley and Wicke 1974). Within individuals these alpha traits are quite consistent over time, however, results from studies correlating endogenous levels of alpha to personality characteristics (e.g., extroversion and introversion) or intelligence are equivocal (Knott 1976; Niedermeyer 1982). Often the effects were minuscule or not reproducible. Nevertheless, individual differences in alpha production can affect experimental results when alpha is used as the dependent variable. Pigeau et al. (1987) showed that high, middle and low alpha subjects differ with respect to the degree of hemispheric specialization shown for cognitive tasks chosen to induce asymmetries. These results suggest that conflicts in the
Correspondence to: Ross A. Pigeau, DCIEM, P.O. Box 2000, North York, Ont. M3M 3B9 (Canada).
literature concerning E E G and asymmetry, specifically Galin and Ornstein vs. Gevins (see Gevins et al. 1979a,b; Ornstein et al. 1980; Galin et al. 1982) may be due to differences in the level of endogenous alpha activity produced by the experimental subjects. The present p a p e r investigates further electrophysiological differences between high and low alpha subjects by assessing their reactivity to photic stimulation at, or near, each subject's spontaneous peak alpha frequency (SPAF). Visual stimulation at rates too fast to allow brain mechanisms to return to their resting levels (between flashes) produces steady-state visual evoked responses ( V E R ) reaching maximum amplitudes in the occipital cortex (Regan 1966). A V E R results from entrainment or photic driving of the E E G at the frequency of the stimulus. With an accurate and stable stimulus, the V E R can be concentrated into very narrow frequency bands especially if a sinusoidally modulated light (SML) source is used (Regan 1989). The advantages of SML stimulation over square wave flashes for this purpose have been well documented (see Van der Tweel and Verduyn Lunel 1965; Regan 1968; Townsend et al. 1975). Substantial differences exist among individuals in their ability to produce steady-state V E R s (Klemm et al. 1980; Yolton et al. 1983). There is growing evidence that this ability is not due to differences in attention, arousal, electrode placement, eye movements or accommodation but rather to differences in the amount of E E G noise (i.e., E E G at frequencies other than those under investigation) present during stimulation
102 (Fagan et al. 1984, 1985; Klemm et al. 1982). Often this noise manifests itself in the alpha frequency range leading to the possibility that endogenous differences in alpha production may affect VERs. The present experiment examines whether high and low alpha subjects differ in their VERs at SML stimulation frequencies spanning a 4 Hz range (in 1/3 Hz intervals) about each subject's SPAF. If the groups demonstrate disparate response characteristics to photic stimulation, then basic electrophysiological differences would be indicated.
Methods
Subjects Sixteen voluteers from the Defence and Civil Institute of Environmental Medicine participated in the study. There were 9 females and 7 males ranging in age from 20 to 35 years. All had normal or corrected vision and reported no history of epilepsy or seizures.
Apparatus The EEG signals were amplified and recorded (for backup purposes) by an 8-channel Medilog 9000 ambulatory recorder (model 90GS). The amplified E E G was coupled to an optical isolator (Bio-Data Pack) for subject protection, band-pass filtered (0.5-40 Hz) and digitized (128 samples/sec) in real-time by an 8 channel A / D converter (National Instruments NB-MIO16H with 12-bit resolution) installed in a Macintosh IIx microcomputer. The raw digitized data were stored on a resident 80 megabyte hard disk for off-line analysis. Labview software (version 1.2 by National Instruments) was used to generate the sinusoidally modulated light (SML) stimuli, digitize and store the E E G and perform FFTs on the data for analysis. During the experiment the required SML stimulus frequency was generated by a Labview algorithm that calculated a single sinusoidal wave form (at the appropriate frequency) comprising of 500 points which then was repeatedly converted from digitial to analog using the D / A channel of the NB-MIO-16H. The SML signal was amplified and connected to 4 light emiting diodes mounted on the inside top, bottom, left and right of each lens of a pair of goggles with opaque lens covers. During stimulation a modulation depth of 30% was used, with the goggles emitting an average luminance of 400 c d / m 2.
Procedure Ag-AgC1 electrodes were applied to the left and right occipital (O1, 02) and the vertex parietal (Pz) locations. Pz was referenced to the left ear while O1 and 0 2 were referenced to linked ears. Two electrodes
R.A. PIGEAU,A.M. FRAME on the neck served as grounds. Inter-electrode impedances were below 5 kO. The subjects were run separately in a quiet room with reduced illumination while seated in a comfortable chair. All baseline and experimental conditions were performed with the subjects' eyes closed to allow maximal alpha production. Although this reduced the amount of light entering the eyes during SML stimulation, pilot studies showed that strong VERs were still produced using this technique. Subjects were instructed to relax but to remain awake during the experiment. Four minutes of spontaneous EEG was recorded as baseline for determining each subject's spontaneous peak alpha frequency (SPAF). For each of thirty 8 sec epochs FFTs were calculated for the right occipital lead. From these spectra the frequencies associated with the largest power values lying in the 8-13 Hz alpha range were selected and averaged to produce a mean SPAF. The LED goggles were then placed on the subjects and 2 min of SML stimulation was given at their SPAF. This was followed by a protocol of 13 SML sessions each lasting 64 sec, in which the SML frequency ranged from 2 Hz below each subject's SPAF to 2 Hz above in 1/3 Hz intervals. The order of presentation was randomized for each subject. A rest period lasting 30-60 sec was given between each session, during which the subjects opened their eyes and verbally interacted with the experimenter. This break ensured that the subjects were awake and alert for each treatment condition. Finally, 4 min of eyes closed spontaneous activity was taken as a post-baseline.
Results
Four channels of data were digitized and stored (at 128 samples/sec): the 3 E E G signals (O1, 0 2 and Pz) and the SML signal driving the LEDs in the goggles. Preliminary testing indicated there was no cross-talk between the channels. Due to an intermittent amplifier malfunction, data from the left occipital electrode (O1) could not be analysed. A square root transformation was applied to all EEG FFT power scores for statistical analyses (Pollock et al. 1981, 1986). To classify the subjects into high and low alpha subjects the following procedure was used. A mean total alpha (8-13 Hz) power measure was generated for each subject by averaging the sum of the FFT power values lying between 8 and 13 Hz for the thirty 8 sec epochs of right occipital EEG from the baseline condition. These means were rank ordered from highest to lowest and split into 2 groups at the median. As can be seen from Table I, high alpha subjects (HASs) exhibited a lower mean SPAF than low alpha subjects
103
VERs IN HIGH AND LOW ALPHA SUBJECTS
a)
1.8x107.~_ 1.6x1071.4X1071.2x107_ 1.0x107" "~ 8.0x1066.0x106~O 4"0x106" O. 2.0x106° 0.0xl00"
.~
la la ~HA-~8
Hz
12 13
~s
b)
2.0x106-
•~ 1"8x1061.6x10B~ 1.4x106_ 1.2x1061.0x106" 8.0x105. 6.0x105"
~ 4,0x105. a. 2.0x1050.0xl0 °"
Fig. 1. FFT power spectra for each subject in the high (a) and low (b) alpha groups. Each spectrum is the (bin by bin) sum of 30 (8 sec) spectra calculated during the 4 min baseline condition. The subjects are ordered on the basis of the amount of alpha produced.
(LASs) - - 9.5 vs. 10.3 Hz respectively (statistically significant, t (14)= -2.6, P < 0.023). High alpha subjects also demonstrated a smaller mean standard deviation about their SPAF than did LASs - - +0.34 vs. +0.72 Hz (t (14)= -3.7, P < 0.002). The robustness of these individual differences is evident because the values remained constant between the baseline and post-baseline conditions. Note that roughly equal numbers of males and females make up each group. Fig. la and b illustrate power spectra for the subjects in each group during the baseline condition. These were qalculated by individually summing the 30 ( 8 se¢).FFF TABLE I
:
spectra lying between 6 and 15 Hz and then ordering them with respect to the total amount of alpha the subjects produced. As can be seen, the majority of LASs demonstrate peak power values within the alpha bandwidth similar to those observed in HASs (although smaller by a factor of = 7). In order to detect the largest possible VERs resulting from SML stimulation, a single FFT was calculated for the entire 64 sec of each of the 13 treatment conditions, yielding power spectra with a frequency resolution of 0.015625 Hz (i.e., 64 frequency bins/Hz). A portion of the spectra spanning 4 Hz, centred at each subject's SPAF, was then extracted to be averaged (bin by bin) across subjects for each SML condition. In Fig. 2a and b VERs are evident for both groups at 1/3 Hz increments about the SPAF. However, HASs show maximal VERs at the SPAF with declining responses with stimulation frequencies further away from the SPAF. LASs do not not show this trend except perhaps at higher stimulation frequencies. A more detailed analysis of the data presented in Fig. 2a and b is needed to isolate the contribution of the VER from background alpha activity. (Note: first harmonics were also isolated and analysed; however, the VERs were so small (an advantage of SML stimulation) that no effects were found.) It is often the case that the VER occupies more than one frequency bin of the power spectrum - - due to practical limits on resolution (in the present case, 64 bins/Hz). To correctly extract the VER effect from the E E G power spectrum a technique was developed which made use of the pure sine wave channel (of the stimulus) digitized concurrently with the EEG. The resulting power spectrum of the SML stimulus (from the same 64 sec of data) contained a very sharp peak at the stimulation frequency with 99% of its power occupying no more than 3 frequency bins (i.e., within 0.046875 Hz). To extract the VER effect from the EEG power spectrum, the SML power spectrum was converted to percentages and then multiplied, bin by bin within the alpha frequency range, with the EEG power spectrum. Since most of the SML spectrum (converted to percentages) contained zeroes or very small values in bins not at the stimulus frequency, multiplying together the two arrays produced a new spectrum which isolated the
~"
Comparisons of high alpha subjects (HASs) and low alpha" subjects (LASs) for the pre- and post-baseline conditions at the right occipital site. Mean alpha power values are in arbitrary units. Baseline
Mean alphapower Mean of SPAF S.D. of SPAF
Post-baseline
High alpha subjects (HASs)
Low alpha subjects (LASs)
High alpha subjects (HASs)
Low alpha subjects (LASs)
232.4 5:64.2 9.53 5:0.54 Hz 0.345:0.16 Hz
93.3 +33.0 10.37-+ 0.76 Hz 0.72+ 0.25 Hz
219.4 +59.7 9.57 5:0.46 Hz 0.345:0.16 Hz
85.9 5:37.5 10.33 -+ 0.40 Hz 0.865:0.46 Hz
104
R.A. PIGEAU, A.M. FRAME
a)
High Alpha Subjects (Right Occipital)
1 *IN
IL Condition
F F T frequency about S P A F
"~" -
-" ~"
Low Alpha Subjects
b)
(Right Occipital)
t.
+1.66 Hz O
B ~
i
_
~I ~[ Il lRmmi lmml 6~ ~m ~
r~
FFT
.
_
+1. "m33m +.66 Hz
+.33 Hz SPAF
~ r ~n~ ~3_
SML Condition
frequency about SPAF
Fig. 2. Averaged FF-T spectra for HASs (a) and LASs (b) for the 13 SML conditions. Spectra range from - 2 Hz to + 2 Hz centred at the SPAF with a frequency resolution of 1/64 Hz. Although VERs are evident in both figures, HASs show maximal VERs at stimulation frequencies proximal to the SPAF.
VERs 1N H I G H A N D LOW A L P H A SUBJECTS
105
a~ 140-
b) Vertex Parietal Electrode
Right Occipital Electrode 140-
120~ 1201
~100S02
80
60g.
40~
40 2
r~ 20i i i i i i i i i i i i i i
i
i
i
i
i
i
i
i
i
i
i
i
t
i
SML Stimulus Conditions
SML Stimulus Conditions
~
HAS
~--
LAS
]
Fig. 3. V E R power ( + S.E.M.) of high and low alpha subjects for the 13 SML conditions at the right occipital (a) and vertex parietal (b) electrodes. All values are squared root transformations of the FFT spectra. Only HASs at the right occipital site showed a statistically significant effect across conditions.
VER effect in the EEG. The power values from this new spectrum were then summed to produce a VER power measure. Fig. 3a and b illustrate the (square root) VER power values across all SML conditions for high and low alpha subjects at the occipital and parietal sites. The abscissas are arranged in order from - 2 to + 2 Hz about the SPAF. Separate repeated measures ANOVAs were performed on the data with HuynhFeldt epsilon corrections for heterogeneity of variance. From Fig. 3a it can be seen that I-lASs produce larger VERs when the stimulus frequency is within = + 0.66 Hz of the SPAF ( F (12, 72) = 3.23, H-F, = 0.294, P
101
Elsevier Scientific Publishers Ireland, Ltd.
EVOPOT 91115
Steady-state visual evoked responses in high and low alpha subjects Ross A. Pigeau and Andrea M. Frame Defence and Civil Institute of Environmental Medicine, North York, Ont. (Canada)
(Accepted for publication: 19 October 1991)
Summary Reactivityto photic stimulation was studied in 16 subjects who were divided into 2 groups based on the relative amounts of spontaneous EEG alpha produced during a 4 min eyes-closed baseline condition. During the experimental session the subjects were presented with thirteen 1 min periods of sinusoidally modulated light stimulation at frequencies ranging from 2 Hz below their spontaneous peak alpha frequency (SPAF) to 2 Hz above in 0.33 Hz intervals. Using FFTs, steady-state visual evoked responses (VERs) were extracted from each subject's EEG for each condition. VER magnitude for high alpha subjects varied with the proximityof the stimulus frequency to the SPAF, the VER for low alpha subjects did not. Conversely, low alpha subjects showed a similar effect in side-band alpha activity(once the VER had been extracted) whereas high alpha subjects did not. The results are explained in terms of a possible difference in the coupling strength between thalamic and cortical alpha sources. Key words: Alpha generators; Steady-state VER; EEG; Alpha
Alpha is the prominent electroencephalographic wave form for normal, relaxed adults with eyes closed. It is characterized by 8-13 Hz activity manifesting over posterior regions of the brain with the largest voltages centred in the occipital cortex (IFSECN 1974). A dominant alpha frequency varying about 10 Hz (Peters6n and Eeg-Olofsson 1971; Osaka 1984) is often displayed with a generalized response of side-band activation within approximately + 1 Hz (e.g., Schwibbe et al. 1981). There is wide variation among individuals in their alpha response. It can be produced as continuous trains of high or low voltage activity and can appear quite sinusoidal (i.e., spindles) or irregular (Lindsley and Wicke 1974). Within individuals these alpha traits are quite consistent over time, however, results from studies correlating endogenous levels of alpha to personality characteristics (e.g., extroversion and introversion) or intelligence are equivocal (Knott 1976; Niedermeyer 1982). Often the effects were minuscule or not reproducible. Nevertheless, individual differences in alpha production can affect experimental results when alpha is used as the dependent variable. Pigeau et al. (1987) showed that high, middle and low alpha subjects differ with respect to the degree of hemispheric specialization shown for cognitive tasks chosen to induce asymmetries. These results suggest that conflicts in the
Correspondence to: Ross A. Pigeau, DCIEM, P.O. Box 2000, North York, Ont. M3M 3B9 (Canada).
literature concerning E E G and asymmetry, specifically Galin and Ornstein vs. Gevins (see Gevins et al. 1979a,b; Ornstein et al. 1980; Galin et al. 1982) may be due to differences in the level of endogenous alpha activity produced by the experimental subjects. The present p a p e r investigates further electrophysiological differences between high and low alpha subjects by assessing their reactivity to photic stimulation at, or near, each subject's spontaneous peak alpha frequency (SPAF). Visual stimulation at rates too fast to allow brain mechanisms to return to their resting levels (between flashes) produces steady-state visual evoked responses ( V E R ) reaching maximum amplitudes in the occipital cortex (Regan 1966). A V E R results from entrainment or photic driving of the E E G at the frequency of the stimulus. With an accurate and stable stimulus, the V E R can be concentrated into very narrow frequency bands especially if a sinusoidally modulated light (SML) source is used (Regan 1989). The advantages of SML stimulation over square wave flashes for this purpose have been well documented (see Van der Tweel and Verduyn Lunel 1965; Regan 1968; Townsend et al. 1975). Substantial differences exist among individuals in their ability to produce steady-state V E R s (Klemm et al. 1980; Yolton et al. 1983). There is growing evidence that this ability is not due to differences in attention, arousal, electrode placement, eye movements or accommodation but rather to differences in the amount of E E G noise (i.e., E E G at frequencies other than those under investigation) present during stimulation
102 (Fagan et al. 1984, 1985; Klemm et al. 1982). Often this noise manifests itself in the alpha frequency range leading to the possibility that endogenous differences in alpha production may affect VERs. The present experiment examines whether high and low alpha subjects differ in their VERs at SML stimulation frequencies spanning a 4 Hz range (in 1/3 Hz intervals) about each subject's SPAF. If the groups demonstrate disparate response characteristics to photic stimulation, then basic electrophysiological differences would be indicated.
Methods
Subjects Sixteen voluteers from the Defence and Civil Institute of Environmental Medicine participated in the study. There were 9 females and 7 males ranging in age from 20 to 35 years. All had normal or corrected vision and reported no history of epilepsy or seizures.
Apparatus The EEG signals were amplified and recorded (for backup purposes) by an 8-channel Medilog 9000 ambulatory recorder (model 90GS). The amplified E E G was coupled to an optical isolator (Bio-Data Pack) for subject protection, band-pass filtered (0.5-40 Hz) and digitized (128 samples/sec) in real-time by an 8 channel A / D converter (National Instruments NB-MIO16H with 12-bit resolution) installed in a Macintosh IIx microcomputer. The raw digitized data were stored on a resident 80 megabyte hard disk for off-line analysis. Labview software (version 1.2 by National Instruments) was used to generate the sinusoidally modulated light (SML) stimuli, digitize and store the E E G and perform FFTs on the data for analysis. During the experiment the required SML stimulus frequency was generated by a Labview algorithm that calculated a single sinusoidal wave form (at the appropriate frequency) comprising of 500 points which then was repeatedly converted from digitial to analog using the D / A channel of the NB-MIO-16H. The SML signal was amplified and connected to 4 light emiting diodes mounted on the inside top, bottom, left and right of each lens of a pair of goggles with opaque lens covers. During stimulation a modulation depth of 30% was used, with the goggles emitting an average luminance of 400 c d / m 2.
Procedure Ag-AgC1 electrodes were applied to the left and right occipital (O1, 02) and the vertex parietal (Pz) locations. Pz was referenced to the left ear while O1 and 0 2 were referenced to linked ears. Two electrodes
R.A. PIGEAU,A.M. FRAME on the neck served as grounds. Inter-electrode impedances were below 5 kO. The subjects were run separately in a quiet room with reduced illumination while seated in a comfortable chair. All baseline and experimental conditions were performed with the subjects' eyes closed to allow maximal alpha production. Although this reduced the amount of light entering the eyes during SML stimulation, pilot studies showed that strong VERs were still produced using this technique. Subjects were instructed to relax but to remain awake during the experiment. Four minutes of spontaneous EEG was recorded as baseline for determining each subject's spontaneous peak alpha frequency (SPAF). For each of thirty 8 sec epochs FFTs were calculated for the right occipital lead. From these spectra the frequencies associated with the largest power values lying in the 8-13 Hz alpha range were selected and averaged to produce a mean SPAF. The LED goggles were then placed on the subjects and 2 min of SML stimulation was given at their SPAF. This was followed by a protocol of 13 SML sessions each lasting 64 sec, in which the SML frequency ranged from 2 Hz below each subject's SPAF to 2 Hz above in 1/3 Hz intervals. The order of presentation was randomized for each subject. A rest period lasting 30-60 sec was given between each session, during which the subjects opened their eyes and verbally interacted with the experimenter. This break ensured that the subjects were awake and alert for each treatment condition. Finally, 4 min of eyes closed spontaneous activity was taken as a post-baseline.
Results
Four channels of data were digitized and stored (at 128 samples/sec): the 3 E E G signals (O1, 0 2 and Pz) and the SML signal driving the LEDs in the goggles. Preliminary testing indicated there was no cross-talk between the channels. Due to an intermittent amplifier malfunction, data from the left occipital electrode (O1) could not be analysed. A square root transformation was applied to all EEG FFT power scores for statistical analyses (Pollock et al. 1981, 1986). To classify the subjects into high and low alpha subjects the following procedure was used. A mean total alpha (8-13 Hz) power measure was generated for each subject by averaging the sum of the FFT power values lying between 8 and 13 Hz for the thirty 8 sec epochs of right occipital EEG from the baseline condition. These means were rank ordered from highest to lowest and split into 2 groups at the median. As can be seen from Table I, high alpha subjects (HASs) exhibited a lower mean SPAF than low alpha subjects
103
VERs IN HIGH AND LOW ALPHA SUBJECTS
a)
1.8x107.~_ 1.6x1071.4X1071.2x107_ 1.0x107" "~ 8.0x1066.0x106~O 4"0x106" O. 2.0x106° 0.0xl00"
.~
la la ~HA-~8
Hz
12 13
~s
b)
2.0x106-
•~ 1"8x1061.6x10B~ 1.4x106_ 1.2x1061.0x106" 8.0x105. 6.0x105"
~ 4,0x105. a. 2.0x1050.0xl0 °"
Fig. 1. FFT power spectra for each subject in the high (a) and low (b) alpha groups. Each spectrum is the (bin by bin) sum of 30 (8 sec) spectra calculated during the 4 min baseline condition. The subjects are ordered on the basis of the amount of alpha produced.
(LASs) - - 9.5 vs. 10.3 Hz respectively (statistically significant, t (14)= -2.6, P < 0.023). High alpha subjects also demonstrated a smaller mean standard deviation about their SPAF than did LASs - - +0.34 vs. +0.72 Hz (t (14)= -3.7, P < 0.002). The robustness of these individual differences is evident because the values remained constant between the baseline and post-baseline conditions. Note that roughly equal numbers of males and females make up each group. Fig. la and b illustrate power spectra for the subjects in each group during the baseline condition. These were qalculated by individually summing the 30 ( 8 se¢).FFF TABLE I
:
spectra lying between 6 and 15 Hz and then ordering them with respect to the total amount of alpha the subjects produced. As can be seen, the majority of LASs demonstrate peak power values within the alpha bandwidth similar to those observed in HASs (although smaller by a factor of = 7). In order to detect the largest possible VERs resulting from SML stimulation, a single FFT was calculated for the entire 64 sec of each of the 13 treatment conditions, yielding power spectra with a frequency resolution of 0.015625 Hz (i.e., 64 frequency bins/Hz). A portion of the spectra spanning 4 Hz, centred at each subject's SPAF, was then extracted to be averaged (bin by bin) across subjects for each SML condition. In Fig. 2a and b VERs are evident for both groups at 1/3 Hz increments about the SPAF. However, HASs show maximal VERs at the SPAF with declining responses with stimulation frequencies further away from the SPAF. LASs do not not show this trend except perhaps at higher stimulation frequencies. A more detailed analysis of the data presented in Fig. 2a and b is needed to isolate the contribution of the VER from background alpha activity. (Note: first harmonics were also isolated and analysed; however, the VERs were so small (an advantage of SML stimulation) that no effects were found.) It is often the case that the VER occupies more than one frequency bin of the power spectrum - - due to practical limits on resolution (in the present case, 64 bins/Hz). To correctly extract the VER effect from the E E G power spectrum a technique was developed which made use of the pure sine wave channel (of the stimulus) digitized concurrently with the EEG. The resulting power spectrum of the SML stimulus (from the same 64 sec of data) contained a very sharp peak at the stimulation frequency with 99% of its power occupying no more than 3 frequency bins (i.e., within 0.046875 Hz). To extract the VER effect from the EEG power spectrum, the SML power spectrum was converted to percentages and then multiplied, bin by bin within the alpha frequency range, with the EEG power spectrum. Since most of the SML spectrum (converted to percentages) contained zeroes or very small values in bins not at the stimulus frequency, multiplying together the two arrays produced a new spectrum which isolated the
~"
Comparisons of high alpha subjects (HASs) and low alpha" subjects (LASs) for the pre- and post-baseline conditions at the right occipital site. Mean alpha power values are in arbitrary units. Baseline
Mean alphapower Mean of SPAF S.D. of SPAF
Post-baseline
High alpha subjects (HASs)
Low alpha subjects (LASs)
High alpha subjects (HASs)
Low alpha subjects (LASs)
232.4 5:64.2 9.53 5:0.54 Hz 0.345:0.16 Hz
93.3 +33.0 10.37-+ 0.76 Hz 0.72+ 0.25 Hz
219.4 +59.7 9.57 5:0.46 Hz 0.345:0.16 Hz
85.9 5:37.5 10.33 -+ 0.40 Hz 0.865:0.46 Hz
104
R.A. PIGEAU, A.M. FRAME
a)
High Alpha Subjects (Right Occipital)
1 *IN
IL Condition
F F T frequency about S P A F
"~" -
-" ~"
Low Alpha Subjects
b)
(Right Occipital)
t.
+1.66 Hz O
B ~
i
_
~I ~[ Il lRmmi lmml 6~ ~m ~
r~
FFT
.
_
+1. "m33m +.66 Hz
+.33 Hz SPAF
~ r ~n~ ~3_
SML Condition
frequency about SPAF
Fig. 2. Averaged FF-T spectra for HASs (a) and LASs (b) for the 13 SML conditions. Spectra range from - 2 Hz to + 2 Hz centred at the SPAF with a frequency resolution of 1/64 Hz. Although VERs are evident in both figures, HASs show maximal VERs at stimulation frequencies proximal to the SPAF.
VERs 1N H I G H A N D LOW A L P H A SUBJECTS
105
a~ 140-
b) Vertex Parietal Electrode
Right Occipital Electrode 140-
120~ 1201
~100S02
80
60g.
40~
40 2
r~ 20i i i i i i i i i i i i i i
i
i
i
i
i
i
i
i
i
i
i
i
t
i
SML Stimulus Conditions
SML Stimulus Conditions
~
HAS
~--
LAS
]
Fig. 3. V E R power ( + S.E.M.) of high and low alpha subjects for the 13 SML conditions at the right occipital (a) and vertex parietal (b) electrodes. All values are squared root transformations of the FFT spectra. Only HASs at the right occipital site showed a statistically significant effect across conditions.
VER effect in the EEG. The power values from this new spectrum were then summed to produce a VER power measure. Fig. 3a and b illustrate the (square root) VER power values across all SML conditions for high and low alpha subjects at the occipital and parietal sites. The abscissas are arranged in order from - 2 to + 2 Hz about the SPAF. Separate repeated measures ANOVAs were performed on the data with HuynhFeldt epsilon corrections for heterogeneity of variance. From Fig. 3a it can be seen that I-lASs produce larger VERs when the stimulus frequency is within = + 0.66 Hz of the SPAF ( F (12, 72) = 3.23, H-F, = 0.294, P
