Participants
Fifteen students (8 M 7 F, mean age = 25 ± 3.8 years) at the China Agricultural University voluntarily participated in the study. All were right-handed with normal or corrected-to-normal vision and had no previous neurological/psychiatric history. The participants received monetary compensation for their time and gave their informed consent in accordance with the IRB at the Institute of Psychology at the Chinese Academy of Sciences.
Stimuli and Procedure
Eighty-four digitized color pictures were selected from IAPS [40]: 42 were classified as aversive pictures (e.g., mutilations, pointing guns) and 42 were classified as neutral pictures (e.g., landscapes, household appliances) (see the Additional description). In accordance with the IAPS scoring, the aversive pictures were more negative in valence than the neutral pictures (2.15 ± 0.43 vs. 5.00 ± 0.35; t (41) = -36.534, p < 0.001), and also more exciting in arousal (6.34 ± 0.62 vs. 3.24 ± 0.59; t (41) = 23.324, p < 0.001).
When presented, all of the pictures occupied between 3 and 4.5° of the angle of vision on either side of the visual midline. Each target stimulus was drawn from a set of 12 capital letters from the Latin alphabet. During the presentation, the letters on a given trial would appear at 1 of 12 positions, each of which was on 1 of 6 equidistant radii of an imaginary circular array 2 or 6 cm from the screen's center.
The participants were seated in an electrically isolated, sound- and light-attenuated room and viewed a computer monitor from a distance of 75 cm.
All of the trials began with a picture, which remained on the screen for 1000 ms, followed by a short interval (which varied randomly from 400 to 600 ms). Three capital letters at random positions around the cross hairs were then presented for 300 ms. After a short delay (which varied randomly from 4000 to 4300 ms), a lowercase letter was presented as a probe stimulus for 300 ms, and the position of this lowercase letter was also picked from the possible 12 positions (see Figure 5). The participants were asked to decide whether or not the lowercase letter was congruent with one of the preceding capital letters, and were instructed to try to respond correctly. The stimulus sets of verbal and spatial WM tasks were equivalent, but different instructions were given to the participants before the beginning of each task. In the verbal task, they were instructed to judge whether the name of the letter in the probe phase was the same as or different from the letter in the target phase, and to ignore the letter's location. However, in the spatial task, only the location of the letter, not its name, was to be remembered and judged. Half of the participants were told to press the "V" key with their left index finger for similar stimuli and the "M" key with their right index finger for different stimuli. Similar and different stimuli were presented in equal proportions. For the other half of participants, the assignment of the response hand was reversed. If no response was given, then the next trial began after 1500 ms. The participants were informed that the pictures had nothing to do with their tasks, but were asked to look at the pictures when they were presented.
The experiment was divided into two sessions that took place on the same day. At one session, only aversive pictures were presented, and at the other, only neutral pictures were presented. The sequence of aversive and neutral sessions was balanced between the subjects. This approach was used because stable emotion and WM interaction require a relatively constant emotional mood. There was a twenty minutes interval between the two sessions to avoid emotional interference. Each session included three verbal task blocks and three spatial task blocks that were pseudorandom in each session. Each block involved 24 trials, which resulted in 72 trials in each experimental condition and meant that each picture was presented at least three times (negative-verbal WM, negative-spatial WM, neutral-verbal WM, neutral-spatial WM). All of the trials in each block were automatically generated and fully randomized by the E-prime program.
Before the beginning of the experiment, the participants were given a positive and negative affect schedule (PANAS) consisting of 10 examples of positive affect (interested, excited, strong, enthusiastic, proud, alert, inspired, determined, attentive, and active) and 10 examples of negative affect (distressed, upset, guilty, scared, hostile, irritable, ashamed, nervous, jittery, and afraid) [41]. They were asked to rate how accurately each of the terms on the scale described the way that they were feeling at that moment (1 = not at all, 5 = very much). After each session of the experiment, the participants filled out another PANAS scale.
ERP Recordings and Analysis
The electroencephalogram (EEG) was recorded from 64 scalp sites using Ag/AgCl electrodes mounted in an elastic cap (NeuroScan Inc.), with the reference on the left mastoid. Vertical electrooculogram (EOG) recording electrodes were positioned above and below the left eye, and horizontal EOG recording electrodes were positioned at the outer canthi of both eyes. All electrode impedances were kept below 5 kΩ. The EEG and EOG were bandpass filtered between DC and 100 Hz and were sampled at a digitization rate of 500 Hz. To allow off-line analysis, they were digitally filtered with a 16 Hz lowpass filter. Trials with various artifacts were rejected with a criterion of ± 100 μV. The ERPs were averaged for trials with correct responses. At least 55 trials were conducted per average per condition (61 trials for negative-verbal WM, 58 trials for negative-spatial WM, 55 trials for neutral-verbal WM, and 56 trials for neutral-spatial WM).
The ERP waveforms were time locked to the onset of the target and probe stimuli, respectively. The averaged epoch for the target stimulus-locked ERP was 4400 ms, ranging from 400 ms before the onset of the target stimuli to 4000 ms after the target stimuli. The averaged epoch for the probe stimulus-locked ERP was 1200 ms, which included a 200 ms pre-probe baseline.
The following 21 sites were chosen for statistical analysis for both the target stimulus-locked ERP and probe stimulus-locked ERP components: FPz, Fz, FCz, Cz, AF3, AF4, F3, F4, FC3, FC4, C3, and C4 (12 anterior sites); Pz, POz, Oz, P3, P4, PO3, PO4, O1, and O2 (9 posterior sites) (see Figure 6). For the target stimulus-locked ERP, the early and lately P3b potentials were measured in time windows of 280-450 ms and 330-770 ms separately over the 9 posterior and 12 anterior sites, as they had different scalp distributions. The negative slow wave (NSW) was broadly distributed over the whole scalp, and was measured by all 21 electrodes in a time window of 1000-4000 ms.
For the probe stimulus-locked ERP, the early P3b and late P3b components were mainly distributed over the nine posterior electrodes and measured in 230-460 ms and 460-800 ms time windows, respectively. To demonstrate the scalp distribution of the emotional state effects, topographic voltage maps of the difference waves (obtained by subtracting the averaged ERPs of the neutral trials from those of the negative trials) were generated at intervals of 300-600 ms.
An ANOVA of each ERP component was conducted with four factors: emotional state (negative/neutral), task type (spatial/verbal), laterality (left/midline/right), and anterior-posterior scalp location (FP/F/FC/C/P/PO/O). The Greenhouse-Geisser correction was used to compensate for sphericity violations.