Abstract

Neuroscience explanations of conscious access focus on neural events elicited by stimuli. In contrast, here, we used the attentional blink paradigm in combination with event-related brain potentials to examine whether the ongoing state of the brain before a stimulus can determine both conscious access and the poststimulus neural events associated with consciousness. Participants were required to detect 2 target letters from digit distractors while their brain activity was being recorded. Trials were classified based on whether the secondcritical target (T2) was detected. We found that T2-detection was predetermined by brain activity prior to the onset of the stimulation stream. Specifically, T2-detected trials were predicated by a frontocentral positive going deflection that started more than 200 ms before the stream began. Accurate T2 detection was also accompanied by enhanced poststimulus neural activity, as reflected by a larger P3b component. Furthermore, prestimulus and poststimulus markers of T2-detection were highly correlated with one another. We therefore argue that conscious experiences are shaped by potentially random fluctuations in neural activity. Overall, the results reveal that conscious access is underpinned by an important relationship involving predictive prestimulus neural activity and responsive poststimulus brain activity.

Introduction

The pathway to consciousness is gated by various filtering mechanisms that prevent the cognitive system from becoming overloaded by sensory stimulation. For example, attentional mechanisms filter stimuli based on features such as location, task relevance, and salience (Luck et al. 2000). However, fluctuations in accurate target report often occur even when physical characteristics and task demands remain constant. Given the importance of prestimulus brain activity in determining perception (van Dijk et al. 2008; Romei et al. 2010), memory recall (Otten et al. 2006), and decision making (Soon et al. 2008), it is feasible that fluctuations in prestimulus activity also predict how well individuals identify and report stimuli that have been perceived. In the current study, we used the attentional blink (AB) paradigm in combination with the excellent temporal resolution of electroencephalography (EEG) to investigate whether consciousness, and the evoked neural responses associated with consciousness, are determined by ongoing brain states. The AB describes a deficit in accurately identifying the second of 2 masked targets (T1 and T2) when T2 appears 200–500 ms after T1 (Raymond et al. 1992). Although T1 and T2 both undergo perceptual and semantic processing (Luck et al. 1996; Vogel et al. 1998), for reasons that we seek to clarify, T2 is reportable only on a fraction of trials.

The poststimulus neural determinants of target detection in the AB are reasonably well established. For example, the P3b event-related potential (ERP) has been frequently employed within AB research as an index of working memory consolidation and resource allocation (Vogel et al. 1998; Kranczioch et al. 2007; Craston et al. 2009). Kranczioch et al. (2007) demonstrate a robust interaction between T1-P3b- and T2-P3b-evoked potentials. Specifically, the T1-P3b is reduced and the T2-P3b is enhanced when T2 is correctly reported, suggesting that T2 detection requires resources to be diverted from T1. Another pertinent study has revealed that brain activity across T2-detected and T2-undetected trials is similar until approximately 270 ms after T2 appears (Sergent et al. 2005). After this time point, T2-detected trials are associated with several neural components that are absent from T2-undetected trials. Sergent et al. (2005) therefore argue that conscious access is related to a late neural component (likely the P3b) that initiates activity across a distributed cortical network. Kranczioch et al. (2007) further assert that T2-detection is related to activation of various distributed neural processes. In their study, T2-detected trials were associated with increased phase coherence in the low beta range and decreased coherence in the alpha range during T2 presentation and before T1 onset. Because prestimulus variations in neural activity were not the focus of the study of Kranczioch et al. (2007), no statistical connection between prestimulus and poststimulus brain activity was investigated. By contrast, the major aim of our study is to link predictive, prestimulus brain activity with reactive, poststimulus activity. Furthermore, the prestimulus effects found by Kranczioch et al. (2007) occurred while distractors were present on the monitor. It is therefore possible that those findings were associated with deviations in the distractor stimuli themselves. The current study examined brain activity in a 1000-ms period between the fixation cross and the onset of the rapid serial visual presentation (RSVP) stream, while no visual stimulation was present.

Here, we investigated the contributions of baseline brain activity to target detection in the AB. By “baseline activity” we mean neural activity occurring prior to any target or distractor stimuli, when no visual stimuli are presented on the monitor. The current study specifically asked whether this baseline activity can determine both conscious access and the established poststimulus neural events associated with target detection. Participants were required to identify 2 target letters among digit distractors presented at 90 ms/item. The crucial test was to examine the relationship between prestimulus brain activity, conscious access to T2, and the neural marker of successful target report—the T2-evoked P3b component.

Materials and Methods

Participants

Initially, 24 young adults took part in the study. One participant was removed due to an inability to achieve 50% accuracy for T1. Data from 23 participants were therefore analyzed (15 females, 21–30 years old). Participants provided informed, written consent, had normal or corrected-to-normal vision, and were fluent in English. The study was approved by the Psychology Research Ethics Committee at the University of Cambridge, UK.

Stimuli and Procedure

Stimuli were presented on a cathode ray tube monitor with a 100-Hz refresh rate. Alphanumeric stimuli were generated using Presentation (Neurobehavioural Systems, Version 11). Targets were the uppercase letters B, C, E, F, J, N, P, and Y. Distractors were single digits excluding 0 and 1. Alphanumeric stimuli were presented in black, on a white screen. Stimuli subtended visual angles of 3.8° vertically and 2.9° horizontally, assuming a viewing distance of 57 cm. On each trial, a fixation cue (a cross shape subtending 2° × 2°) was presented in the center of the monitor for 200 ms. The RSVP stream began 1000 ms after the onset of the fixation cross. Each RSVP stream contained 19 items (2 targets and 17 distractors) that were presented one after the other in the center of the monitor (see Fig. 1). At the end of each stream, participants reported the target letters and were required to guess if they were unsure. Participants then indicated their level of confidence in the accuracy of their answers. An ordered, one-dimensional Likert scale was used to measure confidence. Participants were asked “How confident are you in the accuracy of your answers?” and 4 possible response options were presented in a vertical list on the screen. The options were: 1 = Not at all confident, 2 = A little bit confident, 3 = Quite confident, and 4 = Extremely confident. Participants used the keyboard number keys to select the option best aligned with their subjective confidence.

Figure 1.

The AB paradigm with target letters and digit distractors. Stimuli replace one another in the center of the monitor at a rate of 90 ms/item. In the current EEG study, participants were required to identify the 2 letters after viewing a 19-item stream. T2 appeared at lag 3 on 90% of trials. Electroencephalographic activity was recorded before and during each stream.

Figure 1.

The AB paradigm with target letters and digit distractors. Stimuli replace one another in the center of the monitor at a rate of 90 ms/item. In the current EEG study, participants were required to identify the 2 letters after viewing a 19-item stream. T2 appeared at lag 3 on 90% of trials. Electroencephalographic activity was recorded before and during each stream.

Distractors were presented for 90 ms with no interstimulus interval. In order to realize a 50% distribution of T2-detected versus T2-undetected outcomes, target letters were presented for 50 ms and followed by a 40-ms hash (#) mask with no interstimulus interval. In this manner, the target-mask duration was equivalent to the distractor duration. The identities of the target letters and the digit distractors were randomly assigned on each trial with the restriction that successive items were not the same. In order to prevent the predictable occurrence of the first target, T1 randomly appeared as the fourth, fifth, or sixth item in the stream. Therefore, T1 appeared 1360, 1450, or 1540 ms after the fixation cross.

There were 2 experiments that participants completed one after the other in a single session. First, a behavioral experiment assessed the efficiency of the AB manipulation. In this experiment, T2 appeared at lags 1, 3, or 8 with equal frequency. An EEG experiment followed the behavioral experiment. In the EEG experiment, T2 appeared at lag 3 on 90% of trials and at lag 1 and lag 8 on 5% of trials each. Focusing trials on lag 3 allowed us to maximize EEG signal strength at lag 3 where we expected to elicit the AB deficit.

The behavioral experiment contained 3 blocks of 80 trials, totaling 240 experimental trials. Participants completed this section in under half an hour. The EEG experiment contained 5 blocks of 80 trials, totaling 400 trials. This experiment took approximately 1 h to complete. The order of the trials within each block was randomized. Participants could have short breaks between blocks. Testing occurred individually in an acoustically and electrically shielded booth.

EEG Acquisition and Preprocessing

EEG was recorded using the Electrical Geodesics system and a 129-channel Geodesic sensor net (Electrical Geodesics, Oregon). The sampling rate was 500 Hz. An anti-aliasing low-pass filter of 70 Hz was applied during data acquisition. Offline, the data were band-pass filtered between 0.01 and 30 Hz and recomputed to an average reference. The continuous EEG was segmented into epochs between −200 and 3000 ms relative to the presentation of the fixation cross. Spline interpolation was carried out on individual channels if required. The mean percentage and range of interpolated channels was 2.3% (range: 0–9.4%).

Epochs were excluded from analysis if they met any of the following artifact rejection criteria: voltage deviations exceeded ±100 μV relative to baseline, the maximum gradient exceeded 50 μV, or activity was lower than 0.5 μV. Across participants, 75.72% of trials were retained after filtering and artifact rejection. These remaining epochs were used for 2 kinds of analysis. In the first analysis, ERPs were time locked to the onset of the fixation cross. In the second analysis, exactly the same trials were used but this time ERPs were time locked to the onset of T1. In both analyses, epochs were baseline corrected relative to the epoch average.

Data Analysis

Participants’ responses were scored as correct if a target’s identity was correctly reported, regardless of whether T1 and T2 were reported in the correct order. For both behavioral and EEG analyses, T2 accuracy was always conditionalized on T1 (T2|T1) so that trials were only analyzed if T1 was correctly identified. Trials were then categorized as T2-detected or T2-undetected according to whether the second target was correctly identified. In both the behavioral and EEG experiments, behavioral accuracy scores were analyzed using a two-way analysis of variance (ANOVA) with lag (1 vs. 3 vs. 8) and target (T1 vs. T2|T1) as within-subjects factors. Two Tukey post hoc contrasts were used to examine the interaction effect. The first contrast compared accuracy at lag 1 versus lag 3 and the second compared accuracy at lag 3 versus lag 8. Each of these contrasts was applied to the T1 accuracy data and then to the T2|T1 accuracy data.

To examine confidence ratings, we used a 2 × 4 ANOVA with T2-detection (T2-detected vs. T2-undetected) and confidence as factors. Confidence ratings were only analyzed on lag 3 trials, where T2 falls within the AB period. Three Bonferroni-corrected post hoc contrasts were used to examine the interaction effect. The first contrast compared the frequency of confidence ratings at the 2 extreme ratings versus the 2 middle ratings. The second compared the frequency of low confidence ratings (the lower 2 confidence options) on T2-detected trials versus T2-undetected trials. The third contrast compared the frequency of high confidence ratings (the higher 2 confidence options) on T2-detected trials versus T2-undetected trials. We also investigated whether T2 accuracy on a given trial was related to T2 accuracy on the previous trial. To achieve this, we calculated the percentage of T2-detected trials and T2-undetected trials that were preceded by a T2-detected trial or a T2-undetected trial. These data were entered into a 2 × 2 ANOVA with current trial (T2-detected vs. T2-undetected) and previous trial (T2-detected vs. T2-undetected) as factors. This analysis was only used for the EEG experiment, where T2 was typically presented within the AB. The previous trial analysis would not be meaningful for the behavioral experiment because previous trials would have included lag 1 and lag 8 trials, where T2 is usually detected. Furthermore, trials on which T1 was not detected were excluded from this analysis.

EEG trials were sorted into 2 conditions based on whether T2 was consciously detected or not. Trials were then averaged for each individual, and individuals’ data were averaged to form group ERP means for the T2-detected and T2-undetected conditions. Prestimulus differences in ERPs across the 2 conditions (i.e., differences between T2-detected and T2-undetected trials in the period between the fixation cross and the onset of the RSVP stream) were identified by means of permutational matched t-tests run 10 000 times for each sampling point at each electrode with a critical alpha level of P < 0.05 using the bootstrap statistics toolbox of Zoubir and Iskander (http://www.csp.curtin.edu.au/downloads/bootstrap_toolbox.html; see Zoubir and Boashash 1998). Prestimulus effects were identified at 7 frontal electrode sites with similar polarity and with a clear topography in the period 230–100 ms before the RSVP stream began (this equated to 770–900 ms after the fixation cross appeared). These 7 electrodes formed our frontal pool (electrodes labeled 5, 11, 12, 16, 18, 23, and 24 in the Electrical Geodesics HGSN electrode net; see Supplementary Fig. 1) and are shown circled in Figure 3. Prestimulus effects were further confirmed by permutation statistics run 10 000 times for the grand mean of the electrodes in the frontal pool by taking the mean amplitude for each electrode during the period 230–100 ms before the RSVP stream (see Maris and Oostenveld 2007). Additional prestimulus effects were identified with opposite polarity surrounding the 7 electrodes in our frontal pool. We consider these more sporadic effects to be the phase reversal of the frontal effect and therefore did not analyze them separately. In order to probe the prestimulus effect, we analyzed activity in the circumscribed frontal electrode pool time locked to the onset of the fixation cross. In these epochs, the zero time point represents the onset of the fixation cross, and 1000 ms represents the onset of the RSVP stream. Mean amplitude was calculated for T2-detected and T2-undetected trials in the 230–100 ms period before the RSVP stream (i.e., 770–900 ms after the fixation cross). In addition to the permutation tests that corrected for multiple comparisons, we directly compared mean amplitudes between T2-detected and T2-undetected trials in the period 230–100 ms before the RSVP stream using a repeated-measures t-test with T2-detection (T2-detected vs. T2-undetected) as the repeated-measures factor.

The exact same epochs used in the prestimulus analysis were then time locked to the onset of T1 (so that the zero time point now represents the onset of T1, rather than the onset of the fixation cross). This enabled analysis of the expected poststimulus T1-locked amplitude effects in the centro-parietal region where the P3b typically shows its maximum amplitude (electrodes 61, 62, 67, 72, 77, 78). P3b mean amplitude, peak amplitude, and peak latency were calculated in the period 300–600 ms after T1 (T1-P3b) and 650–950 ms after T1 (T2-P3b). The 300-ms P3b time windows were based on existing P3b literature and visual inspection of the data. The T2-P3b window was positioned slightly later than the T1-P3b window due to significant overlap in activity between T1 and T2. Amplitude and latency values were analyzed using a target (T1 vs. T2) by T2-detection (T2-detected vs. T2-undetected) ANOVA. The interaction effect was probed using two Tukey post hoc comparison contrasts. First, T1-P3b mean amplitude was contrasted across T2-detected and T2-undetected trials. Second, T2-P3b mean amplitude was compared across T2-detected and T2-undetected trials. In order to confirm that correct report of T2 actually reflected conscious detection of that stimulus, we reran this analysis, using only T2-detected trials that were associated with the 2 high confidence ratings. To increase power, we included both low and high confidence ratings for T2-undetected trials because—irrespective of participants’ confidence ratings—we were certain that T2 did not achieve consciousness on T2-undetected trials.

We also examined whether participants’ confidence ratings contributed to the ERP findings, over and above the effects of T2 accuracy. To achieve this, we classified trials according to both accuracy and confidence. Confidence was either “high” (which included the 2 upper confidence ratings) or “low’” (including the 2 lower confidence ratings). To examine the prestimulus effect, mean amplitudes in the frontal pool in the period 230–100 ms before the RSVP stream (i.e., 770–900 ms after the fixation cross) were compared using a repeated-measures ANOVA with T2-detection (T2-detected vs. T2-undetected) and confidence (high vs. low) as factors. To examine the P3b effects, mean amplitudes in the parietal pool were calculated for the T1-P3b and T2-P3b periods. These data were subjected a repeated-measures ANOVA with target (T1 vs. T2), T2-detection (T2-detected vs. T2-undetected), and confidence (high vs. low) as factors.

A correlational analysis was used to determine whether prestimulus EEG effects were statistically related to poststimulus ERP components. To achieve this, the mean amplitude of the prestimulus effect (T2-undetected trials minus T2-detected trials) was determined for each individual at the frontal electrode pool for the period 230–100 ms prior to the onset of the RSVP stream. We then derived a comparable poststimulus effect by calculating the difference between T2-P3b amplitude on T2-undetected trials versus T2-detected trials at each time point in the T2-P3b period (650–950 ms after T1). Correlations were identified by means of permutation tests run 10 000 times for each relevant sampling point across the T2-P3b period at each electrode site with a critical alpha level of P < 0.05 using the bootstrap statistics toolbox of Zoubir and Boashash (1998). The outcome of the point-by-point tests was confirmed by an additional single permutation test run 10 000 times on the grand mean, as described above for the prestimulus effect (see Maris and Oostenveld 2007). Another correlational analysis was also run to examine the relationship between the prestimulus effect and the T1-P3b period (300–600 ms after T1).

Results

Behavioral Data from the Behavioral Experiment

Results replicated the expected AB effect (see Fig. 2A). The target × lag ANOVA revealed that T1 was detected more accurately than T2 (F1,22 = 45.243, P < 0.001, η2 = 0.673). However, target detection performance was moderated by lag (lag: F2,44 = 38.460, P < 0.001, η2 = 0.636; interaction: F2,44 = 105.738, P < 0.001, η2 = 0.828). Tukey pairwise comparisons confirmed the AB effect: T2 accuracy was significantly reduced at lag 3 compared with lag 1 (P < 0.001). Furthermore, T2 accuracy had recovered by lag 8 because accuracy was significantly improved at lag 8 compared with lag 3 (P < 0.001). The T1 accuracy data showed the typical finding of reduced accuracy at lag 1 relative to lag 3 (P = 0.004) but equivalently high accuracy at lags 3 and 8 (P = 0.999).

Figure 2.

Behavioral data from Experiments 1 and 2. All error bars shown standard errors of the mean. (A) Accuracy for T1 and T2|T1 across lags 1, 3, and 8 in Experiment 1. Tukey post hoc contrasts confirmed that T1 accuracy was reduced at lag 1 versus lag 3 but did not differ between lags 3 and 8. Additionally, Tukey comparisons confirmed that T2|T1 accuracy was significantly higher at lag 1 than lag 3 and was higher at lag 8 versus lag 3. (B) Confidence ratings on T2-detected and T2-undetected trials in Experiment 1. (C) Accuracy for T1 and T2|T1 across lags 1, 3, and 8 in Experiment 1. Tukey post hoc contrasts revealed that T1 accuracy was reduced at lag 1 versus lag 3 but did not differ between lags 3 and 8. T2|T1 accuracy was significantly higher at lag 1 than lag 3 and did not differ between lags 3 and 8. (D) Confidence ratings on T2-detected and T2-undetected trials in Experiment 2. Across both Experiments 1 and 2, post hoc contrasts revealed that participants were more likely to use the middle 2 confidence ratings than the 2 extreme confidence ratings. Furthermore, the 2 low confidence ratings were more likely to be employed on T2-undetected trials versus T2-detected trials. Conversely, the 2 high confidence ratings were more likely to be employed on T2-detected trials versus T2-undetected trials.

Figure 2.

Behavioral data from Experiments 1 and 2. All error bars shown standard errors of the mean. (A) Accuracy for T1 and T2|T1 across lags 1, 3, and 8 in Experiment 1. Tukey post hoc contrasts confirmed that T1 accuracy was reduced at lag 1 versus lag 3 but did not differ between lags 3 and 8. Additionally, Tukey comparisons confirmed that T2|T1 accuracy was significantly higher at lag 1 than lag 3 and was higher at lag 8 versus lag 3. (B) Confidence ratings on T2-detected and T2-undetected trials in Experiment 1. (C) Accuracy for T1 and T2|T1 across lags 1, 3, and 8 in Experiment 1. Tukey post hoc contrasts revealed that T1 accuracy was reduced at lag 1 versus lag 3 but did not differ between lags 3 and 8. T2|T1 accuracy was significantly higher at lag 1 than lag 3 and did not differ between lags 3 and 8. (D) Confidence ratings on T2-detected and T2-undetected trials in Experiment 2. Across both Experiments 1 and 2, post hoc contrasts revealed that participants were more likely to use the middle 2 confidence ratings than the 2 extreme confidence ratings. Furthermore, the 2 low confidence ratings were more likely to be employed on T2-undetected trials versus T2-detected trials. Conversely, the 2 high confidence ratings were more likely to be employed on T2-detected trials versus T2-undetected trials.

Participants’ confidence ratings across T2-detected and T2-undetected trials are shown in Figure 2B. In the confidence × T2-detection ANOVA, the confidence main effect and interaction effect returned significant results (confidence: F3,66 = 33.857, P < 0.001, η2 = 0.606; interaction: F3,66 = 5.738, P = 0.001, η2 = 0.207). Post hoc comparisons indicated that participants were more likely to use the middle 2 confidence ratings compared with the 2 extremes (P < 0.001). Additionally, participants were more likely to employ the 2 low confidence ratings on T2-undetected trials compared with T2-detected trials (P = 0.006) and more likely to employ the 2 high confidence ratings on T2-detected trials compared with T2-undetected trials (P = 0.006).

Behavioral Data from the EEG Experiment

Even though the distribution of trials across lags was unequal in the EEG experiment, the behavioral data in this EEG experiment paralleled results from the behavioral experiment (see Fig. 2C). The target × lag ANOVA returned significant main effects of target and lag and a significant interaction effect (target: F1,22 = 54.276, P < 0.001, η2 = 0.712; lag: F2,44 = 27.334, P < 0.001, η2 = 0.554; interaction: F2,44 = 99.999, P < 0.001, η2 = 0.820). Tukey pairwise comparisons revealed that T2 accuracy was significantly reduced at lag 3 compared with lag 1 (P < 0.001). T2 accuracy did not differ between lags 3 and 8 (P = 0.152). The T1 accuracy data showed reduced accuracy at lag 1 relative to lag 3 (P = 0.012), but equivalently high accuracy at lags 3 and 8 (P = 0.064). The confidence × T2-detection ANOVA also returned a significant confidence main effect and an interaction (confidence: F3,66 = 17.018, P < 0.001, η2 = 0.436; interaction: F3,66 = 12.216, P < 0.001, η2 = 0.357). These data are shown in Figure. 2D. Post hoc comparisons confirmed that participants employed the upper and lower extreme confidence ratings with reduced frequency compared with the middle 2 ratings (P < 0.001). Furthermore, participants were more likely to employ the 2 low confidence ratings on trials where they missed T2 than on T2-detected trials (P < 0.001) and more likely to employ the 2 high confidence ratings when they detected T2 compared with T2-undetected trials (P < 0.001).

With regards to the current versus previous trial analysis, a nonsignificant current trial × previous trial interaction suggested that the distribution of T2-detected and T2-undetected outcomes was random or at least unsystematic with regards to the previous trial (F < 1). Post hoc contrasts confirmed that T2-detected trials were not significantly more likely to have been preceded by T2-detected trials (19.623% of trials) versus T2-undetected trials (17.498% of trials) (P = 0.845). Similarly, T2-undetected trials were equally likely to have been preceded by T2-detected trials (17.469%) versus T2-undetected trials (18.162%) (P = 0.964). In other words, there was no difference in the likelihood that a given trial was preceded by a trial with the same outcome or one with a different outcome.

Prestimulus ERP Effect

The EEG results indicated that this experiment was successful in generating approximately equivalent numbers of T2-detected and T2-undetected trials. Of the accepted epochs, 40% were T2-detected trials, 45% were T2-undetected trials, and 15% were other trial types (where T1 or both targets were undetected). As shown in Figure 3A, neural activity across trials that would later be classified as T2-detected or T2-undetected trials was visually identical immediately after the fixation cross (at time 0). Neural activity then diverged between the T2-detected and T2-undetected trials prior to the onset of the RSVP stream, before reuniting after the onset of the RSVP stream. The prestimulus effect appeared as a positive going deflection in T2-detected trials relative to T2-undetected trials. The statistical analyses confirmed that mean amplitude was significantly larger on T2-detected trials than on T2-undetected trials during the period 230–100 ms before the onset of the RSVP stream, which corresponded to 770–900 ms after the fixation cross (t22 = 2.398, P = 0.025). Figure 3B displays the topography of the prestimulus effect. Importantly, this difference occurred at least 600 ms before T1 was presented and therefore more than 2000 ms before the participants’ behavioral response to T2 (and therefore well before the trials could be categorized as T2-detected or T2-undetected trials).

Figure 3.

The prestimulus ERP effect locked to the onset of the fixation cross. (A) Group mean ERPs across T2-detected and T2-undetected trials in the frontal electrode pool. Mean amplitude was contrasted between T2-detected and T2-undetected trials in the period 230–100 ms before the onset of the RSVP stream (i.e., 770–900 ms after the onset of the fixation cross). The significant prestimulus difference between T2-detected and T2-undetected trials is marked by the gray box. (B) Difference topography for the mean amplitude of T2-detected minus T2-undetected trials in the period 230–100 ms before the RSVP stream. Highlighted electrodes indicate statistically significant differences between T2-detected and T2-undetected trials in the period 230–100 ms before the RSVP stream. The 7 circled electrodes indicate the location of the frontal electrode pool used for statistical analyses.

Figure 3.

The prestimulus ERP effect locked to the onset of the fixation cross. (A) Group mean ERPs across T2-detected and T2-undetected trials in the frontal electrode pool. Mean amplitude was contrasted between T2-detected and T2-undetected trials in the period 230–100 ms before the onset of the RSVP stream (i.e., 770–900 ms after the onset of the fixation cross). The significant prestimulus difference between T2-detected and T2-undetected trials is marked by the gray box. (B) Difference topography for the mean amplitude of T2-detected minus T2-undetected trials in the period 230–100 ms before the RSVP stream. Highlighted electrodes indicate statistically significant differences between T2-detected and T2-undetected trials in the period 230–100 ms before the RSVP stream. The 7 circled electrodes indicate the location of the frontal electrode pool used for statistical analyses.

The T2-detection × Confidence ANOVA did not reveal any significant main or interaction effects pertaining to confidence (Fs < 1). As expected, the T2-detection main effect confirmed the significant difference between T2-detected and T2-undetected trials in the period 230–100 ms before the RSVP stream (F1,22 = 5.422, P = 0.029, η2 = 0.198).

ERP Effects following T1 presentation

Figure 4A displays the ERPs for T2-detected and T2-undetected trials, time locked to T1 onset. Figure 4B displays the mean amplitudes for the T1-P3b and the T2-P3b across T2-detected and T2-undetected trials. Mean T1-P3b amplitude was significantly larger than T2-P3b amplitude (F1,22 = 7.809, P = 0.011, η2 = 0.262). However, a significant interaction revealed that this effect was modulated by T2-detection (F1,22 = 25.331, P < 0.001, η2 = 0.535). Tukey post hoc comparisons revealed that the T1-P3b mean amplitude was significantly increased on T2-undetected trials compared with T2-detected trials (P = 0.006). Conversely, the T2-P3b was significantly increased on T2-detected trials compared with T2-undetected trials (P = 0.014). The same results emerged when only T2-detected trials with high confidence ratings were included in the analysis (interaction effect: F1,22 = 21.994, P < 0.001, η2 = 0.500). The results were further replicated when peak rather than mean amplitude was measured (interaction effect: F1,22 = 7.256, P = 0.013, η2 = 0.248). Peak latency did not differ across T2-detected and T2-undetected trials (F1,22 = 1.476, P = 0.237, η2 = 0.063). Like the prestimulus ERP confidence analysis, the Target × T2-detection × Confidence analysis did not return any significant effects pertaining to confidence (Fs < 1). Therefore, although T2 accuracy clearly influenced the EEG data, subjective feelings of confidence did not.

Figure 4.

Poststimulus neural activity locked to T1 onset. (A) Group mean ERPs in T2-detected and T2-undetected trials in the centro-parietal electrode pool. The T1-P3b and the T2-P3b ERP components are as indicated. Tukey post hoc contrasts confirmed that T1-P3b mean amplitude was significantly higher on T2-undetected trials versus T2-detected trials and that T2-P3b mean amplitude was significantly higher on T2-detected trials versus T2-undetected trials. (B) Mean amplitude of the T1-P3b and T2-P3b ERP components in the periods 300–600 ms after T1 (T1-P3b) and 650–950 ms after T1 (T2-P3b).

Figure 4.

Poststimulus neural activity locked to T1 onset. (A) Group mean ERPs in T2-detected and T2-undetected trials in the centro-parietal electrode pool. The T1-P3b and the T2-P3b ERP components are as indicated. Tukey post hoc contrasts confirmed that T1-P3b mean amplitude was significantly higher on T2-undetected trials versus T2-detected trials and that T2-P3b mean amplitude was significantly higher on T2-detected trials versus T2-undetected trials. (B) Mean amplitude of the T1-P3b and T2-P3b ERP components in the periods 300–600 ms after T1 (T1-P3b) and 650–950 ms after T1 (T2-P3b).

Relationship between Prestimulus and Poststimulus Effects

In order to test for a relationship between prestimulus and poststimulus neural effects, a correlation analysis was employed to compare the prestimulus effect with the poststimulus effect across the duration of the T2-P3b window (650–950 ms after T1). This effect was strongest in the middle of the T2-P3b period, between 750 and 850 ms after T1 onset (which corresponded to 480–580 ms after T2). As shown in Figure 5A, the correlation between the prestimulus effect in the frontal region (230–100 ms before the onset of the stream) and the T2-P3b effect in the parietal region (480–580 ms after T2) was statistically robust (r = −0.5411, P = 0.008). Correlations appeared at a subgroup of electrodes showing the T2-P3b effect (electrodes circled in Fig. 5B). Unlike the relationship between prestimulus activity and the T2-P3b, no statistically robust relationship was found between prestimulus activity and poststimulus activity during the T1-P3b period.

Figure 5.

Relationship between correct report, prestimulus brain activity, and poststimulus brain activity. (A) Correlation between the T2-P3b poststimulus effect (mean amplitude difference for T2-undetected trials minus T2-detected trials) in the parietal group of electrodes in the period 750–850 ms after T1 (which corresponded to 480–580 ms after T2) and the prestimulus effect (mean amplitude difference for T2-undetected trials minus T2-detected trials) in the frontal group of electrodes in the period 230–100 ms before the onset of the RSVP stream (which corresponded to 770–900 ms after the fixation cross). The linear regression equation is displayed. (B) Difference topography (T2-undetected trials minus T2-detected trials) for the poststimulus effect in the middle of the T2-P3b period (750–850 ms after T1 or 480–580 ms after T2). The coloring of the map shows amplitude. The bold electrodes indicate the location of electrodes with statistically significant prestimulus/poststimulus effect correlations. The circled electrodes indicate the parietal group used to calculate the T2-P3b effect.

Figure 5.

Relationship between correct report, prestimulus brain activity, and poststimulus brain activity. (A) Correlation between the T2-P3b poststimulus effect (mean amplitude difference for T2-undetected trials minus T2-detected trials) in the parietal group of electrodes in the period 750–850 ms after T1 (which corresponded to 480–580 ms after T2) and the prestimulus effect (mean amplitude difference for T2-undetected trials minus T2-detected trials) in the frontal group of electrodes in the period 230–100 ms before the onset of the RSVP stream (which corresponded to 770–900 ms after the fixation cross). The linear regression equation is displayed. (B) Difference topography (T2-undetected trials minus T2-detected trials) for the poststimulus effect in the middle of the T2-P3b period (750–850 ms after T1 or 480–580 ms after T2). The coloring of the map shows amplitude. The bold electrodes indicate the location of electrodes with statistically significant prestimulus/poststimulus effect correlations. The circled electrodes indicate the parietal group used to calculate the T2-P3b effect.

Discussion

The present study reveals the interrelationships between prestimulus brain activity, poststimulus brain activity, consciousness, and subjective confidence. In a novel contribution to the literature, our results indicate that the successful report of T2 and the neural activity related to successful detection were strongly associated with participants’ brain states prior to the onset of the RSVP stream. Specifically, trials on which T2 was correctly identified were predicated by a positive going deflection in frontocentral regions prior to the onset of the RSVP stream, relative to trials when T2 was not detected. Besides identifying a prestimulus determinant of T2-detection, we also provide evidence for a tight relationship between novel prestimulus effects and traditional poststimulus (P3b) markers of conscious access.

As discussed in the Introduction, prestimulus effects in the ongoing EEG can predict the perception of a given stimulus (Hanslmayr et al. 2007; Hesselmann et al. 2008; van Dijk et al. 2008; Mathewson et al. 2009). However, our prestimulus findings are not relevant to the mere visual perception of T2 because the AB is a postperceptual deficit. Electrophysiological evidence confirms the postperceptual nature of the AB: undetected T2s evoke the N1 and P1 ERP components that are indicative of perceptual processing (Vogel et al. 1998). Furthermore, the ERP component associated with semantic analysis (N400) is present even if there is no conscious report of T2 (Vogel et al. 1998; Rolke et al. 2001; Pesciarelli et al. 2007). Previous and current findings therefore suggest that our prestimulus results should be considered markers of conscious access and not visual perception. The current study consequently extends existing work by revealing that a higher order cognitive process—conscious access (or at least successful target report)—is predetermined by neural activity well before this process actually happens.

Similarly to our prestimulus findings, the poststimulus data help to reveal the neural mechanisms involved in successful target detection. Supporting resource-sharing hypotheses (see Shapiro et al. 2006; Dux et al. 2008, 2009), we report a significant interaction between the amplitude of the T1-P3b and the T2-P3b across T2-detected and T2-undetected trials. Additionally, post hoc comparisons confirmed that the T1-P3b was enhanced and the T2-P3b was reduced on T2-undetected trials hence suggesting that fewer resources are available to process T2 when excessive resources are devoted to T1. Craston et al. (2009) suggest that both a significant interaction and significant post hoc tests are necessary to provide evidence in favor of resource sharing in the AB. To the best of our knowledge, our paper is the first to report such results and therefore provides robust support for resource sharing.

The poststimulus P3b effects all showed the parieto-central topographies that are typically associated with the P3b component. However, it is important to note that we do not support the notion espoused by Fell et al. (2002) and Sergent et al. (2005) that neural activity evoked by T1 “fully” determines the resources allocated to T2, thereby determining whether or not an AB will occur. The robust relationship between prestimulus activity and T2-detection suggests that T2 accuracy is not exclusively determined by the size of the T1-P3b. We therefore contend that even though resource sharing likely plays a contributory role in the AB deficit, other mechanisms (as indexed by prestimulus activity) must also be involved. On a related note, it is important to recognize that our data suggest that poststimulus EEG events do not exclusively depend on evoked EEG activity but are also influenced by properties of the ongoing EEG signal (Makeig 2002; Makeig et al. 2004).

Besides independent prestimulus and poststimulus ERP results, the interrelationship between consciousness, prestimulus brain activity, and poststimulus brain activity represents an important aspect of the current study. Prestimulus brain activity measured in the frontocentral region correlated with the peak of the T2-P3b component measured at different, centro-parietal electrodes (750–850 ms after T1, which corresponded to 480–580 ms after T2). The negative direction of the correlation indicated that larger differences between T2-detected and T2-undetected activity in the prestimulus period corresponded to smaller differences between the T2-detected T2-P3b and the T2-undetected T2-P3b. Importantly, this correlation is unlikely to be a data processing (e.g., filtering) artifact because it appeared across different electrode sites and because we employed a zero phase-shift filter. Despite the correlational nature of this relationship, the current data identify a powerful link between the neural activity that predicts T2-detection and the poststimulus activity that traditionally indexes T2-detection.

Of further interest, the topographical location of the electrodes with prestimulus/poststimulus correlations partially overlapped with the T2-P3b difference topography (T2-undetected trials—T2-detected trials) and the typical P3b topography, which has a clear centro-parietal peak (Donchin 1981). It is well known that the P3b ERP wave is not a unitary index of a single mental event but consists of multiple overlapping ERP waves (Verleger 1997; Dien et al. 2004). We therefore hypothesize that the partial overlap between the correlation map and the T2-P3b suggests that some (yet undefined) constituent of the P3b wave complex is related to the cognitive mechanism indexed by the prestimulus effect. This result suggests that a careful investigation of the topography of similar partial P3b effects in future studies is necessary to determine the functional role of the P3b-related mechanisms in prestimulus/poststimulus correlations. It is important to note that the prestimulus effect we identified was exclusively related to poststimulus activity generated by T2 and not the T1-P3b. This is not particularly surprising given that the AB reflects a deficit in T2 processing, and T1 report is usually at ceiling. Furthermore, numerous examinations of individual differences in the AB find effects on T2-detection but not T1 report, thereby supporting the absence of a relationship between T1 and prestimulus activity in the current study (Martens et al. 2006; Colzato, Bajo, et al. 2008; Colzato, Slagter, et al. 2008; Martens and Valchev 2009; Dale and Arnell 2010).

The behavioral data also contribute meaning to the current findings because participants’ subjective reports of confidence were meaningfully linked to objective accuracy. That is, participants were more likely to feel confident when they correctly reported T2 and more likely to feel unconfident when they incorrectly reported T2. In addition, participants’ confidence ratings were maximal for the central 2 confidence ratings and minimal for the extreme levels of high and low confidence. Because confidence largely mapped onto T2-accuracy, it is not surprising that the confidence ratings did not contribute to the EEG data over and above accuracy. That is, T2 accuracy was able to largely account for the variability in both confidence ratings and the EEG data.

The confidence findings we report are in contrast to the bimodal distribution of subjective visibility previously reported by Sergent et al. (2005), which caused Sergent et al. (2005) to argue that consciousness is not gradual but an all or none phenomenon. However, there are noticeable differences between that study and our own. Sergent et al. (2005) asked participants to indicate how visible T2 was (a perceptual rating), whereas we asked participants to rate their feelings of confidence (a cognitive rating). It may be possible that confidence emerges as a gradual phenomenon but visibility is an all or none phenomenon. More recent data suggests that consciousness emerges gradually (Nieuwenhuis and de Kleijn 2011). When an alternative measure of consciousness (postdecision wagering) was employed, participants’ responses were continuously distributed, hence implying a gradual transition between conscious and nonconscious processing (Nieuwenhuis and de Kleijn 2011). Furthermore, Sergent et al. (2005) employed a visual analog scale, whereas we utilized a Likert scale. A Likert scale was preferable for our study because the use of 4 response options (rather than an excessively large number of options) would have enhanced the reliability of responding across trials (Lozano et al. 2008). Additionally, the even-numbered Likert scale we employed was useful in that it prevented participants from employing a neutral response option. Neutral responses can be problematic because they are not statistically meaningful unless a large number of response options are employed, and they also present an attractive “escape option” for participants unwilling or uninterested in expressing a directional view (Riker 1944; Maxell and Jacoby 1972). We do, however, acknowledge the limitations of the Likert scale employed here. The restricted number of response options may not have encompassed the range of participants’ subjective feelings. Furthermore, participants may not have felt comfortable making extreme choices, thereby biasing them to use the more moderate, central responses. It would be interesting for future research to examine whether confidence ratings are similarly mapped across both Likert and analog scales. Yet, what is clear from the current study is that consciousness is not an all or none phenomenon—at least when confidence is being measured via a Likert scale.

A number of different mechanisms may be put forward to explain the functional significance of the prestimulus effect reported here. One possibility to be considered is that the prestimulus effect represents the amount of controlled effort or vigilance that the participant invests on each trial. In other words, prestimulus activity might reflect the way in which attention is recruited prior to the onset of a task. According to this possibility, different levels of attentional investments would result in different behavioral outcomes and participants would therefore be able to manipulate (perhaps unknowingly) the size of their AB deficit by altering their investment in the task. However, there are some caveats to this explanation. First, this generic task-investment argument can be applied to any attentional task and yet attentional tasks do not all demonstrate a robust AB-like deficit. Second, accuracy on AB tasks does not improve with training (Braun 1994), hence implying that participants do not possess control over their performance.

An alternative explanation of prestimulus effects related to conscious access is that the effects represent random, trial-by-trial fluctuations in neural activity (see Hesselmann et al. 2008; Lakatos et al. 2008; Wyart and Tallon-Baudry 2009). Our data appear to support this view because T2-detected and T2-undetected trials were equally likely to have been preceded by trials on which T2 was correctly or incorrectly detected. Furthermore, it has been suggested that random fluctuations in resource allocation generates variability in T1-P3b amplitude (Shapiro et al. 2006; Kranczioch et al. 2007). We agree that fluctuations in neural activity may impact T1 processing under certain conditions. However, we suggest that T2-P3b activity is more variable than activity elicited by T1, such that the T2-P3b is more likely to be related to prestimulus activity. Supporting this view, our head-wide analysis revealed that the prestimulus effect was significantly related to T2-P3b activity but not T1-P3b activity. According to the “random fluctuation” hypothesis, neural variability may influence consciousness directly (prestimulus activity → consciousness) or indirectly, by manipulating another cognitive process (prestimulus activity → moderating cognitive mechanism → consciousness). Although the directness of the relationship is uncertain, 2 potential candidates for the moderating cognitive mechanism in the indirect route are clear. These candidates are derived from the facts that the prestimulus effect correlates with the P3b component and the P3b component indexes resource allocation or consolidation in working memory (Donchin 1981; Donchin and Coles 1988; Vogel and Luck 2002). If the indirect route is correct, prestimulus activity likely determines the degree of resources available for stimulus processing, indirectly determining whether T2 will be correctly reported. Alternatively, the prestimulus effect may influence how well the targets are consolidated in working memory (see Otten et al. 2010).

The present study therefore establishes the importance of prestimulus neural activity in T2-detection, suggesting that our experiences may be shaped by random fluctuations in neural activity that generates optimal or suboptimal neural performance. These findings challenge existing views of consciousness and motivate the need to incorporate prestimulus effects into contemporary cognitive theories. Our data additionally suggest a number of interesting possibilities for follow-up studies. Further research should aim to investigate the generalizability of the prestimulus effect, examining whether the effect occurs when T2 is presented outside the blink period, for example, at lag 1 or lag 8. It would also be of interest to investigate whether the effect emerges in other related but distinct phenomena such as inattentional and repetition blindness. Finally, uncovering how (or even if) another cognitive mechanisms mediates the relationship between prestimulus activity and consciousness would offer important theoretical insight into this area. For example, one could investigate whether individuals with larger prestimulus effects also show enhanced or reduced working memory spans. To conclude, the current study, in combination with future research, will be instrumental in revealing the neural underpinnings of consciousness, as well as the functional significance and operating mechanisms of the prestimulus effect.

Supplementary Material

Supplementary material can be found at: http://www.cercor.oxfordjournals.org/

Funding

Medical Research Council (G0900643 to D.S.); Gates Cambridge Trust Studentship (to H.L.P.).

We would like to thank Fruzsina Soltesz and Richard Samworth for their assistance with statistical analysis. Conflict of Interest : None declared.

References

Braun
J
Vision and attention: the role of training
Nature
 , 
1994
, vol. 
393
 (pg. 
424
-
425
)
Colzato
L
Bajo
MT
van den Wildenberg
W
Paolieri
D
How does bilingualism improve executive control? A comparison of active and reactive inhibition mechanisms
J Exp Psychol Learn Mem Cogn
 , 
2008
, vol. 
34
 (pg. 
302
-
312
)
Colzato
L
Slagter
H
Spape
M
Hommel
B
Blinks of the eye predict blinks of the mind
Neuropsychologia
 , 
2008
, vol. 
46
 (pg. 
3179
-
3183
)
Craston
P
Wyble
B
Chennu
S
Bowman
H
The attentional blink reveals serial working memory encoding: evidence from virtual and human event-related potentials
J Cogn Neurosci
 , 
2009
, vol. 
21
 (pg. 
550
-
566
)
Dale
G
Arnell
K
Individual differences in dispositional focus of attention predict attentional blink magnitude
Atten Percept Psychophys
 , 
2010
, vol. 
72
 (pg. 
602
-
606
)
Dien
J
Spencer
KM
Donchin
E
Parsing the late positive complex: mental chronometry and the ERP components that inhabit the neighborhood of the P300
Psychophysiology
 , 
2004
, vol. 
41
 (pg. 
665
-
678
)
Donchin
E
Surprise!… Surprise?
Psychophysiology
 , 
1981
, vol. 
18
 (pg. 
493
-
513
)
Donchin
E
Coles
MGH
Is the P300 component a manifestation of context updating?
Behav Brain Sci
 , 
1988
, vol. 
11
 (pg. 
357
-
374
)
Dux
P
Asplund
C
Marois
R
An attentional blink for sequentially presented targets: evidence in favor of resource depletion accounts
Psychon Bull Rev
 , 
2008
, vol. 
15
 pg. 
809
 
Dux
P
Asplund
C
Marois
R
Both exogenous and endogenous target salience manipulations support resource depletion accounts of the attentional blink: a reply to Olivers et al
Psychon Bull Rev
 , 
2009
, vol. 
16
 (pg. 
219
-
224
)
Fell
J
Klaver
P
Elger
C
Fernandez
G
Suppression of EEG gamma activity may cause the attentional blink
Conscious Cogn
 , 
2002
, vol. 
11
 (pg. 
114
-
122
)
Hanslmayr
S
Aslan
A
Staudigl
T
Klimesch
W
Herrmann
CS
Bauml
KH
Prestimulus oscillations predict visual perception performance between and within subjects
Neuroimage
 , 
2007
, vol. 
37
 (pg. 
1465
-
1473
)
Hesselmann
G
Kell
CA
Eger
E
Kleinschmidt
A
Spontaneous local variations in ongoing neural activity bias perceptual decisions
Proc Natl Acad Sci U S A
 , 
2008
, vol. 
105
 (pg. 
10984
-
10989
)
Kranczioch
C
Debener
S
Maye
A
Engel
A
Temporal dynamics of access to consciousness in the attentional blink
Neuroimage
 , 
2007
, vol. 
37
 (pg. 
947
-
955
)
Lakatos
P
Karmos
G
Mehta
AD
Ulbert
I
Schroeder
CE
Entrainment of neuronal oscillations as a mechanism of attentional selection
Science
 , 
2008
, vol. 
320
 (pg. 
110
-
113
)
Lozano
LM
Garcia-Cueto
E
Muniz
J
Effect of the number of response categories on the reliability and validity of rating scales
Methodology
 , 
2008
, vol. 
4
 (pg. 
73
-
77
)
Luck
SJ
Vogel
EK
Shapiro
K
Word meanings are accessed but cannot be reported during the attentional blink
Nature
 , 
1996
, vol. 
383
 (pg. 
616
-
618
)
Luck
SJ
Woodman
G
Vogel
EK
Event-related potential studies of attention
Trends Cogn Sci
 , 
2000
, vol. 
4
 (pg. 
432
-
440
)
Makeig
S
Unifying brain electrophysiology: work in progress
Trends Neurosci
 , 
2002
, vol. 
25
 pg. 
390
 
Makeig
S
Debener
S
Onton
J
Delorme
A
Mining event-related brain dynamics
Trends Cogn Sci
 , 
2004
, vol. 
8
 (pg. 
204
-
210
)
Maris
E
Oostenveld
R
Nonparametric statistical testing of EEG- and MEG-data
J Neurosci Methods
 , 
2007
, vol. 
164
 (pg. 
177
-
190
)
Martens
S
Munneke
J
Smid
H
Johnson
A
Quick minds don’t blink: electrophysiological correlates of individual differences in attentional selection
J Cogn Neurosci
 , 
2006
, vol. 
18
 (pg. 
1423
-
1438
)
Martens
S
Valchev
N
Individual differences in the attentional blink. The important role of irrelevant information
Exp Psychol
 , 
2009
, vol. 
56
 (pg. 
18
-
26
)
Mathewson
KE
Gratton
G
Fabiani
M
Beck
D
To see or not to see: prestimulus alpha phase predicts visual awareness
J Neurosci
 , 
2009
, vol. 
29
 (pg. 
2725
-
2732
)
Maxell
MS
Jacoby
J
Is there an optimal number of alternatives for Likert-scale items? Effects of testing time and scale properties
J Appl Psychol
 , 
1972
, vol. 
56
 (pg. 
506
-
509
)
Nieuwenhuis
S
de Kleijn
R
Consciousness of targets during the attentional blink: a gradual or all-or-none dimension?
Atten Percept Psychophys
 , 
2011
, vol. 
73
 (pg. 
364
-
373
)
Otten
L
Quayle
A
Akram
S
Ditewig
T
Rugg
M
Brain activity before an event predicts later recollection
Nat Neurosci
 , 
2006
, vol. 
9
 (pg. 
489
-
491
)
Otten
L
Quayle
A
Puvaneswaran
B
Prestimulus subsequent memory effects for auditory and visual events
J Cogn Neurosci
 , 
2010
, vol. 
22
 (pg. 
1212
-
1223
)
Pesciarelli
F
Kutas
M
Dell'Acqua
R
Peressotti
F
Job
R
Urbach
T
Semantic and repetition priming within the attentional blink: an event-related brain potential (ERP) investigation study
Biol Psychol
 , 
2007
, vol. 
76
 (pg. 
21
-
30
)
Raymond
JE
Shapiro
KL
Arnell
KM
Temporary suppression of visual processing in an RSVP task: an attentional blink?
J Exp Psychol Hum Percept Perform
 , 
1992
, vol. 
18
 (pg. 
849
-
860
)
Riker
BL
A comparison of methods in attitude research
J Abnorm Soc Psychol
 , 
1944
, vol. 
39
 (pg. 
24
-
42
)
Rolke
B
Heil
M
Streb
J
Hennighausen
E
Missed prime words within the attentional blink evoke an N400 semantic priming effect
Psychophysiology
 , 
2001
, vol. 
38
 (pg. 
165
-
174
)
Romei
V
Gross
J
Thut
G
On the role of pre-stimulus alpha-rhythms over occipito-parietal areas in visual input regulation: correlation or causation?
J Neurosci
 , 
2010
, vol. 
30
 (pg. 
8692
-
8697
)
Sergent
C
Baillet
S
Dehaene
S
Timing of the brain events underlying access to consciousness during the attentional blink
Nat Neurosci
 , 
2005
, vol. 
8
 (pg. 
1391
-
1400
)
Shapiro
K
Schmitz
F
Martens
S
Hommel
B
Schnitzler
A
Resource sharing in the attentional blink
Neuroreport
 , 
2006
, vol. 
17
 (pg. 
163
-
166
)
Soon
CS
Brass
M
Heinze
H-J
Haynes
J-D
Unconscious determinants of free decisions in the human brain
Nat Neurosci
 , 
2008
, vol. 
11
 (pg. 
543
-
545
)
van Dijk
H
Schoffelem
JM
Oostenveld
R
Jensen
O
Prestimulus oscillatory activity in the alpha band predicts visual discrimination ability
J Neurosci
 , 
2008
, vol. 
28
 (pg. 
1816
-
1823
)
Verleger
R
On the utility of P3 latency as an index of mental chronometry
Psychophysiology
 , 
1997
, vol. 
34
 (pg. 
131
-
156
)
Vogel
EK
Luck
SJ
Delayed working memory consolidation during the attentional blink
Psychon Bull Rev
 , 
2002
, vol. 
9
 (pg. 
739
-
743
)
Vogel
EK
Luck
SJ
Shapiro
KL
Electrophysiological evidence for a postperceptual locus of suppression during the attentional blink
J Exp Psychol Hum Percept Perform
 , 
1998
, vol. 
24
 (pg. 
1636
-
1674
)
Wyart
V
Tallon-Baudry
C
How ongoing fluctuations in human visual cortex predict perceptual awareness: baseline shift versus decision bias
J Neurosci
 , 
2009
, vol. 
29
 (pg. 
8715
-
8725
)
Zoubir
AM
Boashash
B
The Bootstrap and its applications in signal processing
IEEE Signal Proc Mag
 , 
1998
, vol. 
15
 (pg. 
56
-
76
)