Figure 1. A: Examples of the memory configurations in the grouped, partially grouped, and ungrouped conditions and for the placeholder configuration presented on the unattended display side. Note that each memory configuration presented exactly six different colors and orientations, such that the overall physical stimulation was identical in all three memory configurations. Moreover, the physical stimulation of the placeholder was in critical respects (including the size and positions) comparable to the memory configurations, without however providing relevant color and orientation information. B: Example trial sequence, depicting a memory array that presents a grouped configuration on the right side of the display (as indicated by the preceding arrow cue) and a to-be-ignored placeholder configuration of six gray circles on the left side. Following a retention interval, the test display is presented, which would reveal a probe item on the cued side (and a placeholder circle on the uncued side). The probe would either depict a color change (left), an orientation change (middle), or no change (right).
Trials were presented in randomized order such that all conditions, that is, the possible configurations (grouped, partially grouped, and ungrouped) and change types (no change, color, or orientation change), were presented randomly intermixed across trials. This ensured that observers were required to memorize both the color and orientation features in the memory displays. All participants performed 9 practice blocks of 64 trials each on the day before the experiment, to become familiar with the (rather demanding) task. The experiment itself then consisted of 18 blocks of 64 trials each, amounting to 1152 experimental trials. After each block, participants had the opportunity to take a short break.
EEG recording. The EEG data was continuously recorded and digitized at 1000 Hz using a 64 channel Ag/AgCl active electrode system connected to a polyester elastic head-cap (EasyCap64, Brain Products, Munich, Germany). The electrodes were positioned in accordance with the international 10-10 system. The horizontal electrooculogram (EOG) was recorded from electrodes placed at the outer canthi of the eyes (F9 and F10). The vertical EOG was recorded from an electrode beneath the left eye (VEOG; positioned at the same distance from the center of the eye as the Fp1 electrode), in order to detect blinks and vertical eye movements. The electrode signals were amplified using a wireless amplifier system (BrainAmp, Brain Products, Munich, Germany) with a 0.1- to 250-Hz bandpass filter. During data acquisition, all electrodes were referenced to FCz and re-referenced offline to averaged mastoids. All electrode impedances were kept below 5 kΩ.
Artifact Rejection. Offline signal processing was performed using the Brain Vision Analyzer software (BrainProducts, Munich, Germany). The raw data were inspected visually to manually remove nonstereotypical noise and they were then high-pass filtered using a Butterworth infinite impulse response filter at 0.5 Hz (24 dB per octave). Next, an infomax independent component analysis was carried out to identify components representing blinks and/or horizontal eye movements and to remove these artifacts before back-projection of the residual components (1% of all trials were removed because of eye-movement artifacts). Prestimulus baseline correction (−1000 ms to −800 ms before VWM array onset) was performed on the raw voltages. ERPs were calculated time-locked to the onset of the memory display, with segments extending from 200 ms before stimulus onset until 1300 ms afterwards. Only trials without artifacts [defined as any signal exceeding ± 60 μV, bursts of electromyographic activity (the maximum voltage step allowed per sampling point was 50 μV) and activity lower than 0.5 μV within intervals of 500 ms (indicating dead channels)] were considered for further analysis on an individual-channel basis before the ERP waveforms were averaged.
ERP data analysis. We included 6 parieto-occipital electrodes chosen a-priori and based on previous findings (e.g., Adam et al., 2018; Fukuda et al., 2015): PO3, PO4, PO7, PO8, O1, and O2. Specifically, we subtracted ERPs from parieto-occipital electrodes ipsilateral to the memory array’s location from contralateral ERPs. Based on predictions drawn from previous work (Wiegand et al. 2015; Diaz et al. 2021), we examined for an attentional modulation of modality-specific sensory responses in the visual PPC, N1pc, N2pc, and CDA components (130–160 ms, 160–200 ms, 260–330 ms, and 350–1300 ms post memory display, respectively; in these time-windows, the respective components were clearly present in all conditions) at lateral parieto-occipital sites.
Oscillatory Amplitude Analysis . To gain an overall picture of the frequency components, the pre-processed EEG time series data from each separate channel were Morlet-wavelet filtered into 31 frequency bands, fmin=3 Hz to fmax=120 Hz with the Morlet time–frequency compromise parameter m, being m = 5. For the time window between -200 and 1300 ms following stimulus onset, we extracted instantaneous amplitude values for the alpha band (8–12 Hz) and calculated the alpha amplitude for the same parieto-occipital electrodes as in the ERP analysis (PO3, PO4, PO7, PO8, O1, and O2). We calculated the mean amplitude over contralateral and ipsilateral posterior channels (PO3, PO4, PO7, PO8, O1, and O2) in the same way as for the ERP analysis (i.e., relative to the memory array). To compute the lateralization magnitude, we took the difference between the contralateral and ipsilateral alpha amplitude averaged over the specified time window (350–1300 ms) comparable to the procedure as described for the CDA.
Results
Behavioral data. To determine whether there were differences in accuracy across the different experimental conditions, we performed a repeated-measures analysis of variance (ANOVA) with the factors Object Configuration (grouped, partially grouped, ungrouped) and Change Type (color, orientation). Greenhouse-Geisser-corrected values are reported when Mauchley’s test of sphericity was significant (p < .05). We additionally report Bayes factors (BF10 ) for non-significant results to evaluate the evidence for the null hypothesis (see Jeffreys, 1961; Kass & Raftery, 1995). The Bayes factor provides the ratio with which the alternative hypothesis is favored over the null hypothesis (values below 1/3 may be taken to support the null hypothesis, whereas values greater than 3 would provide evidence in favor of the alternative hypothesis; see Jeffreys, 1961; Kass & Raftery, 1995). As we had a-priori hypotheses about the direction of effects (we predicted grouping to lead to increased memory performance), one-tailed paired samples t-tests (along with one-tailed Bayesian paired samples t-tests) were used for comparisons between the various object configurations.
Figure 2A presents the percentage of correct responses as a function of object configuration, separately for color and orientation changes. The Object-Configuration by Change-Type repeated-measures ANOVA yielded significant main effects of Object Configuration, F (2, 46) = 70.97, p < .001,ηp2 = .76, and Change Type,F (1, 23) = 4.63, p = .04,ηp2 = .17. There was a graded effect of Object Configuration, with the highest accuracy for grouped configurations (73%), followed by partially grouped (66%) and ungrouped (63%) configurations (all p ’s < .001,d zs > 0.88 for the pairwise comparisons between configurations). In addition, accuracy was higher for color changes than for orientation changes (68% vs. 66%). Finally, the Object-Configuration × Change-Type interaction was significant,F (2, 46) = 30.47, p < .001,ηp2 = .57: the enhancement of performance with increasing grouping strength was several times larger for orientation changes (grouped vs. ungrouped: 16%, p< .001, d z = 2.11; grouped vs. partially grouped: 12%, p < .001,d z = 1.70; partially grouped vs. ungrouped: 5%,p < .001, d z = 0.82) than for color changes (grouped vs. ungrouped: 4%, p < .001,d z = 0.87; grouped vs. partially grouped: 2%,p = .013, d z = 0.48; partially grouped vs. ungrouped: 2%, p = .038, d z = 0.38). It should be noted, however, that both types of change benefited significantly (albeit to a differential degree) from the increase in grouping strength. Overall, the mean performance was around 67%, while decreasing in some conditions to ~60% (e.g., in the orientation change condition with ungrouped configurations). Importantly, though, the mean accuracies were significantly above chance level in all conditions, t s(24)> 11.61, p s < .001, d s> 2.37.