Alpha entrainment is responsible for the attentional blink phenomenon.
ABSTRACT The attentional blink phenomenon is the reduced ability to report a second target (T2) after identifying a first target (T1) in a rapid serial visual presentation (RSVP) of stimuli (e.g., letters), which are presented at approximately 10 items per second. Several explanations have been proposed, which focus primarily on cognitive aspects, such as attentional filter-, capacity limitation- and retrieval failure-processes. Here, we focus on the hypothesis that an entrainment of alpha oscillations (with a frequency of about 10Hz) is a critical factor for the attentional blink phenomenon. Our hypothesis is based on the fact that item presentation rate in the RSVP typically lies in the alpha frequency range and is motivated by theories assuming an inhibitory function for alpha. We predict that entrainment - during the time window of T2 presentation - is larger for attentional blink (AB) items (when T2 cannot be reported) than for NoAB trials (when T2 cannot be reported). The results support our hypothesis and show that alpha entrainment as measured by the amplitude of the alpha evoked response and the extent of alpha phase concentration is larger for AB than for NoAB trials. Together with the lack of differences in alpha power these findings demonstrate that the differences between AB and NoAB trials - during presentation onset of T2 - are due to an entrainment of alpha phase and not due to an amplitude modulation. Thus, we conclude that alpha entrainment may be considered the critical factor underlying the attentional blink phenomenon.
- SourceAvailable from: Cornelia Kranczioch[Show abstract] [Hide abstract]
ABSTRACT: The attentional blink (AB) is a deficit in conscious perception of the second of two targets if it follows the first within 200-500 msec. The AB phenomenon has been linked to pretarget oscillatory alpha activity. However, this is based on paradigms that use a rapid serial visual presentation (RSVP) stimulus stream in which the targets are embedded. This distracter stream is usually presented at a frequency of 10 Hz and thus generates a steady-state visual-evoked potential (ssVEP) at the center of the alpha frequency band. This makes the interpretation of alpha findings in the AB difficult. To be able to relate these findings either to the presence of the ssVEP or to an effect of endogenously generated alpha activity, we compared AB paradigms with and without different pretarget distracter streams. The distracter stream was always presented at 12 Hz, and power and intertrial phase coherence were analyzed in the alpha range (8-12 Hz). Without a distracter stream alpha power dropped before target presentation, whereas coherence did not change. Presence of a distracter stream was linked to stronger pretarget power reduction and increased coherence, which were both modulated by distracter stream characteristics. With regard to the AB results indicated that, whereas ssVEP-related power tended to be higher when both targets were detected, endogenous alpha power tended to be lower. We argue that the pattern of results indicates that in the pretarget interval several processes act in parallel. The balance between these processes relates to the occurrence of an AB.Journal of Cognitive Neuroscience 01/2014; · 4.49 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Key to successfully negotiating our environment is our ability to adapt to current settings based on recent experiences and behaviour. Response conflict paradigms (e.g., the Stroop task) have provided evidence for increases in executive control after errors, leading to slowed responses that are more likely to be correct, and less susceptible to response congruency effects. Here we investigate whether failures of perceptual awareness, rather than failures at decisional or response stages of information processing, lead to similar adjustments in visual attention. We employed an attentional blink task in which subjects often fail to consciously register the second of two targets embedded in a rapid serial visual presentation stream of distractors, and examined how target errors influence performance on subsequent trials. Performance was inferior after Target 2 errors and these inter-trial effects were independent of the temporal lag between the targets and were not due to more global changes in attention across runs of trials. These results shed light on the nature of attentional calibration in response to failures of perceptual consciousness.PLoS ONE 01/2013; 8(4):e60623. · 3.53 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Alpha band (8-12Hz) phase dynamics in the visual cortex are thought to reflect fluctuations in cortical excitability that influences perceptual processing. As such, visual stimuli are better detected when their onset is concurrent with specific phases of the alpha cycle. However, it is unclear whether alpha phase differentially influences cognitive performance at specific times relative to stimulus onset (i.e., is the influence of phase maximal before, at, or after stimulus onset?). To address this, participants performed a delayed-recognition, working memory (WM) task for visual motion direction during two separate visits. The first visit utilized functional magnetic resonance (fMRI) imaging to identify neural regions associated with task performance. Replicating previous studies, fMRI data showed enagement of visual cortical area V5, as well as a prefrontal cortical region, the inferior frontal junction (IFJ). During the second visit, transcranial magnetic stimulation (TMS) was applied separately to both the right IFJ and right V5 (with the vertex as a control region) while electroencephalography (EEG) was simultaneously recorded. During each trial, a single pulse of TMS (spTMS) was applied at one of six time points (-200, -100, -50, 0, 80, 160ms) relative to the encoded stimulus onset. Results demonstrated a relationship between the phase of the posterior alpha signal prior to stimulus encoding and subsequent response times to the memory probe two seconds later. Specifically, spTMS to V5, and not the IFJ or vertex, yielded faster response times, indicating improved WM performance, when delivered during the peak, compared to the trough, of the alpha cycle, but only when spTMS was applied 100ms prior to stimulus onset. These faster responses to the probe correlated with decreased early event related potential (ERP) amplitudes (i.e., P1) to the probe stimuli. Moreover, participants that were least affected by spTMS exhibited greater functional connectivity between V5 and fronto-parietal regions. These results suggest that posterior alpha phase indexes a critical time period for motion processing in the context of WM encoding goals, which occurs in anticipation of stimulus onset.NeuroImage 07/2013; · 6.25 Impact Factor
Alpha entrainment is responsible for the attentional blink phenomenon
Andrea Zaunera, Robert Fellingera, Joachim Grossb, Simon Hanslmayrc, Kimron Shapirod, Walter Grubera,
Sebastian Müllera, Wolfgang Klimescha,⁎
aDepartment of Physiological Psychology, University of Salzburg, Austria
bCentre for Cognitive Neuroimaging, Institute of Neuroscience and Psychology, University of Glasgow, UK
cDepartment of Psychology, Zukunftskolleg, University of Konstanz, Germany
dBangor University, School of Psychology, UK
a b s t r a c ta r t i c l ei n f o
Accepted 29 June 2012
Available online 14 July 2012
The attentional blink phenomenon is the reduced ability to report a second target (T2) after identifying a first
target (T1) in a rapid serial visual presentation (RSVP) of stimuli (e.g., letters), which are presented at ap-
proximately 10 items per second. Several explanations have been proposed, which focus primarily on cogni-
tive aspects, such as attentional filter-, capacity limitation- and retrieval failure‐processes.
Here,we focus onthe hypothesis that anentrainmentofalpha oscillations (with a frequency of about 10 Hz) isa
critical factor for the attentional blink phenomenon. Our hypothesis is based on the fact that item presentation
rate in the RSVP typically lies in the alpha frequency range and is motivated by theories assuming an inhibitory
function for alpha. We predict that entrainment – during the time window of T2 presentation – is larger for at-
tentional blink (AB) items (when T2 cannot be reported) than for NoAB trials (when T2 cannot be reported).
The results support our hypothesis and show that alpha entrainment as measured by the amplitude of the
alpha evoked response and the extent of alpha phase concentration is larger for AB than for NoAB trials. To-
gether with the lack of differences in alpha power these findings demonstrate that the differences between
AB and NoAB trials – during presentation onset of T2 – are due to an entrainment of alpha phase and not
due to an amplitude modulation. Thus, we conclude that alpha entrainment may be considered the critical
factor underlying the attentional blink phenomenon.
© 2012 Elsevier Inc. All rights reserved.
The aim of the present study is to test the hypothesis that alpha en-
trainment is a critical factor for the occurrence of the attentional blink
phenomenon, which is the reduced ability to report a second target
after identifying a first target in a rapid serial visual presentation
(RSVP) of stimuli (e.g., letters), presented at approximately 10 items
per second. The structure of a single trial is illustrated in Fig. 1. Each
item is displayed briefly with or without an interstimulus interval
(ISI). In the case an ISI is used, item exposure time (of e.g., 25 ms) and
ISI (of e.g., 75 ms) add up to a SOA of about 100 ms. Subjects have to
search for two targets, T1 and T2. The critical result is that subjects fail
to report T2 in about 50% of the cases when T1 is identified and when
the two targets are separated by at least 1 item but not more than
about 7 intervening items. This failure to report T2 – after a successful
encoding of T1 – constitutes the attentional blink phenomenon. It
can be observed in a time window of about 100 to 500 ms after T1
(Raymond et al., 1992; Shapiro et al., 1997) provided at least one
non-target stimulus follows T1 and T2.
items are presented with a stimulus onset asynchrony (SOA) of about
100 ms (which represents a stimulation frequency of around 10 Hz)
and is motivated by theories which assume an inhibitory function for
alpha oscillations (Foxe and Snyder, 2011; Jensen and Mazaheri, 2010;
Klimesch et al., 2007; Mathewson et al., 2011). In addition, there is evi-
dence that alpha is actively and causally involved in shaping visual per-
ception and responds with a phase specific entrainment as recent
rhythmic transcranial magnetic stimulation studies have shown (Romei
et al., 2010; Thut et al., 2011b). Here we focus on a specific aspect of the
assumed inhibitory function of alpha, which was termed ‘P1 inhibition
timing hypothesis’ (Klimesch, 2011). It states that the P1 of the visual
rization (Klimesch, 2011, Klimesch et al., 2011).
according to Hanslmayr et al. (2011) – in three groups: ‘filter-’, ‘capacity
limitation-’, and ‘retrieval failure’ theories. The central idea of filter
theories is that the attentional focus (‘filter’) on T1 acts to suppress the
processing of subsequent stimuli (cf. Olivers and Meeter, 2008 for a
more recent model; Raymond et al., 1992). Related theories by Di Lollo
NeuroImage 63 (2012) 674–686
⁎ Corresponding author at: University of Salzburg, Department of Physiological
Psychology, Institute of Psychology, Hellbrunnerstr. 34, A-5020 Salzburg, Austria.
Fax: +43 662 8044 5126.
E-mail address: firstname.lastname@example.org (W. Klimesch).
1053-8119/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/ynimg
et al. (2005) or Kawahara et al. (2006) assume that a ‘re-orientation’ of
the attentional filter towards T2 is the critical factor underlying the
attentional blink phenomenon. Capacity limitation theories proceed
from the assumption that the establishment of an episodic code for T2 –
which is a necessary precondition to report that target at the end of a
trial – interferes with the episodic encoding of T1 (cf. Chun and Potter,
1995; Jolicoeur, 1999; Jolicoeur and Dell'Acqua, 1998). Finally, the third
group of theories assumes that the failure to report T2 is due to the
inability of retrieving T2 from working memory (cf. Duncan et al., 1994;
Shapiro et al., 1994; Ward et al., 1996).
Only a few studies have focused on a relationship between alpha and
the attentional blink. As an example, MacLean and colleagues have
nitude. Large resting alpha power and a large extent of event-related
alpha desynchronization (i.e. ERD, reflecting suppression of alpha
power) during task performance is associated with a large magnitude
of attentional blink (MacLean and Arnell, 2011; MacLean et al., 2012).
These findings are important because they suggest a direct link between
er, there is – to our knowledge – no research that has yet considered
alpha entrainment (due to the stimulation frequency of the RSVP) as
the cause for the failure to report T2, although it is a well established
finding that flickering stimuli (such as the RSVP) evoke a steady state vi-
sually evoked potential (SSVEP) with a dominant frequency in the
alpha band. The properties of driven EEG activity are well investigated
(e.g. Herrmann, 2001; Kawaguchi et al., 1993; Lakie and Combes, 1999;
Lazarev et al., 2001; Mast and Victor, 1991; Sakamoto et al., 1993; Shils
et al., 1996; for a recent review see Thut et al., 2011a) and show in gen-
eral that the alpha frequency range is most responsive. An important as-
pect thereby is that a stimulation in the broad alpha frequency range
leads to a partial or full entrainment of the alpha rhythm. We proceed
here from the assumption that alpha entrainment with minimal phase
lag relative to stimulus onset of T2 is responsible for the attentional
blink. In order to motivate our hypothesis, we give a brief review of the
few studies that have investigated the relationship between the phase
lag of the SSVEP and cognitive performance.
lag’ or ‘SSVEP phase’; cf. Silberstein, 1995) between the flickering stim-
ulus and the oscillatory EEG response. Silberstein and colleagues were
able to demonstrate that with increasing task demands SSVEP phase
lag (latency) tends to increase. As an example, in a visual search task,
designed to test the influence of increased attentional demands, sub-
jects had to view shapes of squares and circles either passively or
under the instruction to detect a modified circle (Silberstein et al.,
1990). Compared to passive viewing, the requirement to detect a mod-
ified circle was associated with a transient reduction of SSVEP ampli-
tude at occipital sites and an increased phase lag. In another study
cognitive processes during performance of the Wisconsin Card Sorting
Test (WCST) were analyzed with the SSPT technique (Silberstein et
al., 1995). This test consists of many cards containing objects that vary
in color, number and shape. Subjects are presented several cards and
asked to find some rule that is common for the presented cards. After
subjects have found the sorting criterion, a new criterion is introduced
bythe experimenter inthe nexttrial. Thisprocedure isrepeatedseveral
times. From neuropsychological evidence it is well known that particu-
larly the shift to a new criterion is a specific task demand that is closely
associated with the prefrontal cortex (cf. Gazzaniga, 1995). Most inter-
estingly, it was found that during the introduction of a new sorting cri-
terion a significant reduction in SSVEP amplitudes and an increased
phase lag was observed at anterior recording sites (cf. Silberstein,
1995 for a review).
These findings suggest that a demanding cognitive task is associat-
ed with a pronounced phase lag, whereas minimally demanding tasks
are associated with a small phase lag. With respect to our hypothesis,
we assume that a large phase lag enables good performance whereas
Fig. 1. Illustration of a trial of the attentional blink paradigm. A) First a fixation cross was presented for 1500 ms, followed by a stream of 20 letters, which consisted of consonants
and vowels. Each letter was presented for 100 ms. Target 1 (T1) always appeared at the 7th position and was characterized by a green letter. A second target (white X=T2)
appeared after 300 ms (Lag 3, 10th position of stimulus stream) in 50 % of the trials. Instead of T2, distracters (consonants) were presented in control trials. After the stream of
20 letters, subjects had to decide whether T1 and T2 were presented. A time window of 1500 ms was given for each decision. A blank screen appeared for 1500 ms at the end
of the trial. B) In the varied block stimulus presentation was jittered between 80 to 120 ms with +/− 10, 20 ms steps, whereby T2 always appeared 300 ms after T1 (Lag 3).
A. Zauner et al. / NeuroImage 63 (2012) 674–686
a small lag may reduce processing capacity. We assume here that
phase locking of the SSVEP – in the alpha frequency range – with min-
imal phase lag is an important factor underlying the attentional blink.
In addition, we expect that the absolute phase angle will also play an
important role. The reason for this latter assumption is that the ap-
pearance of the P1 – which has a peak latency of about 100 ms –
will be generated at exactly that time point where the next following
stimulus is presented. If alpha is completely entrained with minimal
phase lag, the interference between those processes that underlie
the generation of the P1 and those that enable encoding of the next
following stimulus will be particularly large. Because there is good ev-
idence that the P1 not only has a frequency characteristic in the alpha
band but also may be generated at least in part by alpha oscillations it
is obvious to assume that the cognitive processes associated with the
P1 will interfere with the encoding of the next following stimulus.
Research by Klimesch et al. (2011) suggests that the P1 reflects
early stimulus categorization which precedes stimulus identification
(a process associated with the appearance of the N1 component).
Our hypothesis is that attentional blink may stem from an interfer-
ence between early stimulus identification of the preceding target
(T1) and the encoding of the subsequent target (T2). If the onset of
these two processes overlap (i.e., are time locked with minimal delay
or ‘lag’), they will interfere and result in a failure to encode T2. We,
thus, predict a larger entrainment (as measured by alpha phase locking
and phase concentration) during the stimulus onset of T2 in trials with
attentional blink (AB trials) as compared to trials without attentional
blink (NoAB trials). In order to investigate, whether entrainment de-
creases when the RSVP stream is not presented with an SOA of exactly
the same length for all presentations but instead with a jittered SOA,
we perform two experimental blocks, one with fixed and one with var-
ied presentation times.
We want to emphasize that for our hypothesis alpha phase locking
in relation to stimulation onset, reflecting entrainment, is the critical
factor. This should not be confused with alpha phase coherence,
whichreflects the normalized degree ofphase variabilitybetweenelec-
trode pairs. This coherence measure which sometimes is simply re-
ferred to as ‘alpha phase synchronization’ was already investigated in
attentional blink tasks. As an example, Gross et al. (2004) made the im-
portant observation that beta phase coherence is reduced prior to T2 in
NoAB trials only. This finding agrees with studies that have focused on
the influence of prestimulus alpha phase coherence and target identifi-
cation performance. Hanslmayr et al. (2007) have shown that reduced
long-range synchrony in the prestimulus interval predicts successful
stimulus identification (cf. also Kranczioch et al., 2007). These findings
are consistent with but are indifferent for testing our hypothesis.
Specific evidence for our hypothesis comes from findings reported
by Mathewson and colleagues. As an example, in a visual target de-
tection task, Mathewson et al. (2009) found that for undetected trials,
the phase at stimulus onset was different from that of detected trials
(cf. Busch et al., 2009 for similar findings). Most importantly, when
the target was not detected a prominent negative peak at stimulus
onset was associated with significantly reduced P1 amplitude. This
may suggest that in trials where the phase of alpha at stimulus
onset interferes with the generation of the P1, the stimulus will not
be detected. The idea is that an ongoing alpha with 10 Hz and a peri-
od of 100 ms which exhibits a negative peak at stimulus onset will
develop a negative peak at 100 ms poststimulus which would lead
to a suppression of the (positive) P1 amplitude.
The logic of our analyzing approach is the following. We use a stan-
dard attentional blink paradigm and in a first step, will measure the
peak amplitude of the peristimulus ERP component in response to T2.
negative peakaround the onset ofT2for AB ascomparedto NoABtrials.
In a next step we will analyze the alpha filtered ERP in order to test,
whether the expected findings are prominent for this frequency
range. Then we will analyze phase locking of alpha (relative to T2
onset), in order to test, whether AB trials exhibit a larger extent of
phase locking (entrainment) than NoAB trials. Finally, we will analyze
absolute phase, in order todeterminedifferencesin thephase angle be-
tween AB and NoAB trials. We expect that only for AB trials, (absolute)
alpha phase will coincide with the negative peak of the peristimulus
ERP component (which is termed CT2 later in the text).
An original sample of 24 subjects was obtained on the basis of
course requirements that subjects had to fulfill to finish their diploma
study. All subjects participated in the experiment after giving in-
formed consent. They filled out a short response sheet where they
were asked for their age, psychoactive drug uses, and neurological
disorders. All of the recordings were carefully supervised by the first
author (A.Z.) and the EEG was checked for artifacts after the experi-
ment was performed. 10 subjects could not be used for data analysis.
This large rejection rate is partly due to the fact that the student ex-
perimenters were not that skilled as professional EEG assistants are
and that subjects participated to receive their course credits. Out of
the 10 rejected subjects, 4 subjects were lost due to electrodes losing
the proper impedance during the experiment, 2 subjects reporting
their intake of drugs to cure depression, and 4 subjects showing ex-
cessive movement artifacts. The final sample consisted of 14 subjects
(7 males and 7 females). Mean age was 25.6 years (SD=3.7 years).
All subjects of the final sample reported no neurological disorders
or psychological pathologies.
Stimuli and task
Weused atraditionalAB-paradigm,ase.g., described inthestudyby
Kranczioch et al. (2003). Each trial consisted of a rapid serial visual pre-
sentation (RSVP) of 20 letters. Two types of trials were constructed, ex-
perimental trials containing two targets and control trials containing
only one target. The two targets were a green letter (vowel or conso-
nant) and the (white) letter X. The green letter was the first target
(T1) which always appeared at the 7th position, whereas the letter X
wasthe secondtarget(T2)whichalways appearedatthe 10thposition.
A set of 160 trials wasused, comprising 80 experimental and 80 control
trials. Inhalf of the experimental trials, T1 was a vowel, in the other half
of thetrials a consonant. Likewiseincontroltrials, inhalf of the casesT1
was a vowel, in the other half of the trials a consonant.
For the present study we used a subset of 21 letters of the alpha-
bet, which was obtained by excluding vowel I and the consonants F,
K, Q, and Z. This subset contains 4 vowels and 17 consonants. The con-
struction of individual trials was done with the following restrictions.
After the selection of target items (with 2 positions for experimental
and 1 for control trials), there were 18 or 19 item positions for
distracter items. Only consonants were used as distracters. Thus, a
remaining set of 15 or 16 consonants (in experimental items T1 can
also be a consonant, whereas T2 always consists of the consonant
‘x’) had to be used to fill 18 or 19 item positions. This means that
item repetitions were necessary. Consonants were presented in ran-
dom sequence but with the restriction that the same consonant
would never occupy two adjacent positions.
Each trial started with the presentation of a fixation cross that
appeared 1500 ms before (and remained on the screen until) the
onset of the first letter of the entire 20-letter stream. Half of all trials
were experimental trials, the other half control trials. They were
presented in a random sequence. In control trials T2 was not
presented. All letters were presented in white on black background
(with the exception of T1 which was presented in green).
Two different experimental blocks were performed. In block A, each
letter of the RSVP was exposed for 100 ms (‘fixed’ presentation block).
A. Zauner et al. / NeuroImage 63 (2012) 674–686
Inblock B,the ‘varied’ presentationblock, presentationtime was ‘jittered’
pseudo-randomly between 80 and 120 ms with +/−10, 20 ms steps,
whereby T2 always appeared 300 ms after T1. As for block A, 160 trials
(80 experimental, 80 control) were constructed for block B.
The subject's task was to report both targets. They were informed
that T2 would be missing in only half of all trials. At the end of each
trial (immediately after the presentation of the 20 letters) a visual re-
had to report whether the item was a consonant or vowel. During the re-
sponse cue for T2, subjects had to report whether they have seen an ‘X’.
Responses were given by pressing a respective response key. The
interstimulus interval between trials (between the offset of the response
and the structure of a single trial is illustrated in Fig. 1.
EEG data acquisition
A Brain Vision Recorder (1000 Hz, 64 channels; BrainProducts, Inc.)
was used for EEG recording. EEG-signals were referenced to a nose elec-
trode and subsequently (off-line) re-referenced to digitally linked
((A1+A2)/2) ear lobe references. Band-pass filters were set from 0.5
to100 Hz and a notchfilterat 50 Hz. Signals were digitizedata sampling
rate of 500 Hz. 60 Ag–AgCl-electrodes were mounted using an EasyCap
on the following positions: Fp1, Fp2, Af7, Af3, Afz, Af4, Af8, F7, F5, F3, F1,
Fz, F2, F4, F6, F8, Ft7, Fc5, Fc3, Fc1, Fcz, Fc2, Fc4, Fc6, Ft8, T7, C5, C3, C1,
Cz, C2, C4, C6, T8, Tp7, Cp5, Cp3, Cp1, Cpz, Cp2, Cp4, Cp6, Tp8, P7, P5, P3,
P1, Pz, P2, P4, P6, P8, Po7, Po3, Poz, Po4, Po8, O1, Oz, O2. Impedances
were kept below 8 kΩ. To control for vertical and horizontal eye move-
ments two bipolar EOG-channels were mounted. After re-referencing
epochs containing eye or muscle artifacts were rejected. Data were seg-
mented from 1000 to 1500 ms relative to T1. Data analyses were
performed using BrainVision Analyzer (BrainProducts, Inc.) and Matlab®
7.9 (The MathWorks, Inc.).
Each session started with the recording of the resting EEG for
1 min. Subjects were asked to close their eyes.
EEG data analysis
All basic steps were done with Brain Vision Analyzer. Single-trial
phase-angle analyses were done with custom-made Matlab-scripts.
At first the data were re-referenced as mentioned above and then
broadly filtered between 0.5 and 70 Hz. Then data were manually
checked and corrected for muscle- and eye-blinked artifacts. Subse-
quently data were separately filtered between 1–30 Hz and 8–12 Hz
and segmented according to target detection performance, resulting
in NoAB trials (both targets detected), in AB trials (T1 detected, T2
missed) and control trials (only T1 presented and detected); the lat-
ter allowing a false alarm rate to be calculated.
Calculation of ERPs
Due to the short SOA (and repeated rhythmic stimulus presentation)
which prevents the EEG to return to baseline, the obtained ERPs actually
represent SSVEPs. Our analysis aims to analyze the dominant ERP re-
sponse component to T1 and T2. Because the SSVEP cannot contain the
well known P1 and N1—components which can be observed for long
SOAs (of about 500 ms and longer), we term the dominant components
to T1 and T2, component (C) elicited by T1 and T2 respectively (CT1
For the calculation of ERPs, single trial data were filtered between
1–30 Hz and 8–12 Hz. Visual inspection revealed a negative compo-
nent around stimulation onset of T1 and T2. The mean peak ampli-
tude and latency of CT2 were semi-automatically computed in a
time window of 250 to 350 ms around T2 appearance. The peak
detection was carried out over the ERP for each subject and each con-
dition (experimental and control) separately for each ROI (see
Calculation of alpha power, phase locking and phase distribution
Whole power was calculated employing wavelet transformation
for 1 to 20 Hz with the use of a 7-cycle complex Morlet wavelet to ob-
tain an adequate time–frequency resolution. Time–frequency analy-
ses were carried out for all trials in each condition. The frequency of
interest was the 8–12 Hz alpha band.
Phase locking index
for a given frequency across time (Schack and Klimesch, 2002; Tallon-
Baudry et al., 1996). The PLI is a normalized value that ranges between
0 (reflecting a lack of phase locking) and 1 (reflecting perfect phase
locking). For PLI analyses complex (single trial) wavelet coefficients
were calculated for every time and frequency bin within a frequency
range of 1–20 Hz. Statistical analyses focus on the broad alpha frequency
ual alpha frequency (IAF) we used a narrow band around 10 Hz (9.5–
In order to investigate differences in the preferred phase angle be-
tween AB and NoAB and control trials, data were Hilbert transformed.
The Kuipers uniform test was run to test for deviations of uniform dis-
tribution. Paired sample t-test controlled for equal set size of single tri-
als (AB, NoAB) for fixed SOA (t(13)=−.359; p=.725) and varied SOA
(t(13)=−.837; p=.418). For all cases with significant deviations from
mean phase angle, was computed.
ROI analysis, fixed and varied presentations
Alpha usually is the largest at posterior (occipital and parietal) re-
cording sites. In order to determine, whether the expected effects are
indeed restricted to these sites, we performed paired sample t-test for
all sites to test for significant differences between conditions in a time
window around the presentation of T2. Because our central hypothe-
sis is focusing on alpha phase locking relative to T2, we focus on a
time interval preceding and following the onset of the second target
by 50 ms each (i.e. an interval of 250 to 350 ms following the onset
of T1). The mean ERP activity (in terms of the difference-area of the
respective ERP-segments) between (1) AB and NoAB, (2) AB and con-
trol trials, (3) NoAB and control trials was calculated and statistically
compared. Only those electrodes were selected at which all three
comparisons yielded significant effects (pb. 05).
For the evaluationof statisticaldifferences betweenconditions of the
different EEG parameters (e.g. CT2 amplitude and latency, power and
PLI), one-way ANOVAs for repeated measurements with 3 conditions
(AB, NoAB and control) were run separately for each ROI. Greenhouse–
Geisser correction was applied where necessary and the significance
level was set to pb.05. Phase distribution was evaluated by the Kuipers
For the examination of the influence of individual alpha frequency
(IAF) on behavior (AB magnitude) and phase locking at stimulation fre-
quency (as measured by the PLI at 10 Hz) weperformed 2-wayANOVAs
with factor IAF (subjects with IAF around 10 Hz versus below and above
10 Hz) and factor task (fixed vs. varied). Dependent measures were ei-
ther AB magnitude (percentage of trials with AB) or PLI. IAF was deter-
mined for the resting EEG at electrodes O1, Oz, and O2. IAF varied
between 8.8 and 11.7 Hz. Factor IAF was obtained by grouping subjects
A. Zauner et al. / NeuroImage 63 (2012) 674–686
in a group with an IAF around 10 Hz (9.0–10.7 Hz) and a group deviant
from 10 Hz (below 9.0 and above 10.7 Hz).
ROI analysis, fixed and varied presentations
For the fixed presentation block, the analysis yielded 5 ROIs, which
were termed (1) central ROI (z, C2, C4), (2) centro-parietal ROI (CP5,
CP3, CP1, CPz, CP2, CP4), (3) parietal ROI (P5, P3, P1, Pz, P2, P4),
(4) parieto-occipital ROI (Po7, Po3, Poz, Po4), and an (5) occipital
ROI (O1, Oz, O2). An overview of the selected ROIs is given in Fig. 2.
For the varied presentation block no significant results were
obtained. This shows that the ERP-segment around T2 does not differ
significantly between conditions. Despite this lack of significant find-
ings, the same ROIs – as obtained for the fixed representation block –
were used to assess differences between conditions on selected pa-
rameters as reported below.
The direction of statistical differences is consistent for all ROIs and
shows that mean polarity (in the time window of 250 to 350 ms rel-
ative to the onset of T1) is most negative for AB trials and least nega-
tive (or already positive) for NoAB, with control trials exhibiting an
intermediate position (for an illustration of this finding see Fig. 3a).
Behavioral data, fixed and varied presentations
An AB trial is defined by a correct report of T1 but a failure to re-
port T2. NoAB trials consist of a correct report of both targets, T1
and T2. Thus, both AB and NoAB trials represent experimental trials,
in which the first target (T1) was correctly reported. For the fixed
presentation task, this set consisted of 76.7 trials (SD=3.8) which
is 95.9% of all experimental trials. This means that in 4.1% from the
80 experimental trials, T1 was not reported. From the remaining
76.7 trials in which T1 was reported, 50.2% were NoAB and 49.8%
were AB trials. For control trials, the target (T1) could be reported
in 90.3% of all cases. T1 was missed in 2.3% and T2 was erroneously
reported, i.e., false alarms, in 7.4% of the cases (cf. a summary of the
behavioral data is in Table 1).
In the varied presentation task, T1 was not reported in 6.4% of all
trials. From the remaining 70.0 trials in which T1 was reported,
52.2% were NoAB and 47.8% were AB trials. For control trials, the T1
could be reported in 90.9% of all cases. T1 was missed in 2.4% and
T2 was erroneously reported in 6.7% of the cases.
The results of a 2-way ANOVA with factor TASK (fixed, varied)
and TARGET TYPE (T1, T2) and the percentage of correct responses
showed – as expected – a significant effect for TARGET TYPE
(F(1,13)=136.8; pb.001). Most importantly, neither significant ef-
fects for factor TASK nor the interaction TASK×TARGET TYPE was
IAF and performance
The results of a 2-way ANOVA with factor IAF (subjects with IAF
around10 Hzvs.IAFdeviantfrom10 Hz),factorTASK(fixedvs.varied),
and AB magnitude as dependent variable showed that individual alpha
frequency had noimpact on performance. Neither factor IAF nor the in-
teraction reached or exceeded the 5%-level of significance. IAF varied
between 8.8 and 11.7 Hz with a mean of 10.4 Hz (SD=1.1 Hz).
Steady state responses, fixed presentation
An example for an ERP of the entire stream of 20 item presenta-
tions for the parietal ROI is shown in Fig. 3a. Visual inspection reveals
a negative component that appears around the onset of almost each
stimulus. Most interestingly, immediately before the presentation of
T2 (i.e., at stimulus onset of the letter preceding T2), this negative
component reaches a maximum (relative to all 20 letter presenta-
tions). It is important to note that here the negative component is
larger for NoAB and control trials as compared to AB trials. During
the presentation of T2 – as the ROI analysis has already shown –
this relationship reverses, with AB trials showing now the largest
negative component relative to the other two trial types.
The band pass filtered ERPs (in the 8–12 Hz frequency range) – as
depicted in Fig. 3b for the parietal ROI – show that the prestimulus
negative component has a strong frequency characteristic in the
alpha band. It is also interesting to note that after the appearance of
T1, the evoked alpha response increases until the onset of T2 and
then decreases for the next 2–3 item presentations.
Steady state responses, varied presentation
The ERPs of the varied presentation block are depicted in Fig. 4a.
They show that the pronounced ERP differences between conditions
(i.e., between AB, NoAB and control trials) during – and following –
the onset of T2, which can be observed in the fixed task are absent
in the varied task. Most interestingly, the alpha filtered ERPs are
quite similar in the fixed and varied presentation blocks.
CT2 peak amplitude and latency, fixed presentation
The results of one-way ANOVAs for the CT2 peak amplitude show
significant main effects between AB, NoAB and control trials for cen-
tral (F2/26=10,6; pb.001), centro-parietal (F2/26=15,6; pb.001), pa-
rietal (F2/26=12,2; pb.001), parieto-occipital (F2/26=6,5; pb.01) but
not for occipital ROIs. Pairwise comparisons showed significant
(pb.05) larger amplitudes for AB as compared to NoAB or control tri-
als. No significant differences were obtained between NoAB and con-
trol trials. No significant effects were observed for CT2 peak latencies.
CT2 peak amplitude and latency, varied presentation
The one-way ANOVAs for the data of the varied task did not yield
significant effects for any of the ROIs. Thus, the amplitude of the CT2
component does not differ between conditions.
For peak latency of the CT2 component, a significant main effect
(F2/26=3.9; pb.05) was observed at a parieto-occipital ROI.
Fig. 2. Illustration of the results of the ROI analysis. Only these electrodes were selected
for ROI at which AB significantly differed from both NOAB and control trials and in ad-
dition NOAB from control trials (pb.05).
A. Zauner et al. / NeuroImage 63 (2012) 674–686
Inspection of the respective means indicates that the latency for NoAB
trials is about 20 ms shorter than for AB and control trials.
Alpha band (8–12 Hz)
Whole power, fixed and varied presentations
None of the ANOVAs for whole power in the fixed and varied tasks
reached significance. This indicates that there is no significant power
differences between AB, NoAB and AB trials that are not time or phase
Evoked alpha: CT2 peak amplitude and latency of the filtered ERPs, fixed
The filtered ERPs' significant main effects between AB, NoAB and
control trials were obtained for centro-parietal (F2/26=5,3; pb.05),
parietal (F2/26=6.4; pb.01), parieto-occipital (F2/26=4,1; pb.05)
but not for central and occipital ROIs. Pairwise comparisons yielded
similar effects as for the unfiltered data. Again, the direction of the
differences is that the CT2 amplitude is largest for AB as compared
to NoAB and AB trials. No significant effects were observed for CT2
Evoked alpha: CT2 peak amplitude and latency of the filtered ERPs, varied
For the peak component, a similar pattern of results could be ob-
served (cf. Figs. 3a and 4a), but the differences between AB, NoAB
and control trials did not reach significance. For the centro-parietal
ROI, the respective F-value (F2/26=5.3; p=.056) closely missed
-1000 -900 -800 -700 -600 -500 -400 -300 -200 -1000 100200300
400500 600700 800900 1000 1100 1200 1300
-1000 -900 -800 -700 -600 -500 -400 -300 -200 -100 0100 200 300 400 500 600 700 800 900 1000 1100 1200 1300
Fig. 3. SSVEP's fixed SOA. A) Unfiltered ERP for parietal ROI. Bold vertical dashed lines indicate the onset of T1 and T2 (for experimental trials). Vertical dashed lines reflect the onset
of distracters. Note the strong negative component around T2 appearance for AB (red) compared to NoAB (green) and control (black). B) 8–12Hz filtered ERP for parietal ROI.
Correct (%) Incorrect (%)Correct (%)Incorrect (%)
T2 false alarm rate
A. Zauner et al. / NeuroImage 63 (2012) 674–686
exceeding the 5%-level of significance. Again, no significant effects
were observed for CT2 peak latencies.
Phase locking index (PLI), fixed presentation
Significant main effects were observed for central (F2/18,289=4.8;
pb.05), for centro-parietal (F2/26=6.2; pb.01), parietal (F2/26=6.0;
pb.01) and parieto-occipital (F2/26=4.2; pb.05), but not for occipital
ROIs. For these ROIs with significant main effects, pairwise compari-
sons yielded significant differences between AB and control trials
only. Inspection of the respective means shows that phase locking is
largest for AB and smallest for control trials with NoAB trials exhibiting
slightly larger values than control trials. The largest effects were ob-
served for electrode P4.
Phase locking index (PLI), varied presentation
In a similar way as for the fixed presentation block, significant
main effects were observed for central (F2/26=3.7; pb.05), and for
centro-parietal (F2/26=3.4; pb.05) ROIs. No other ROIs reached
Entrainment of phase, fixed presentation
The findings for the phase locking index show that intertrial phase
stability is larger for AB trials. This finding, however, gives no infor-
mation about the preferred phase angle at which phase locking oc-
curs. Thus, we calculated the phase distribution during the onset of
T1 and T2 for electrode P4 which exhibited the largest PLI. Then, we
used the Kuipers test, to examine, whether the observed phase distri-
butions exhibit a significant phase concentration that differs from a
random distribution. Finally, for cases with significant phase concen-
trations, we calculated the mean phase angle. The results, as depicted
in Fig. 5a—show that a significant phase distribution could be ob-
served for T1 and T2 (tαcrit=.01 in both cases) but only for AB and
not for NoAB trials. For T1 the mean phase angle is at 245°, which is
in the positive going slope, near the 0-crossing. For T2, however the
mean angle is 161° which is close at the negative peak of the alpha
When the Kuipers test is calculated for any sample point of the en-
tire series of 20 letters, the following two important findings emerge.
As illustrated in Fig. 6a, a significant phase concentration is the
-1000 -900 -800 -700 -600 -500 -400 -300 -200 -1000100 200300
400 500 600700800 900 1000 1100 1200 1300
-1000 -900 -800 -700 -600 -500 -400 -300 -200 -100
400 500 600 700 800900 1000 1100 1200 1300
Fig. 4. SSVEP's varied SOA. Data presentation is analogous to Fig. 3.
A. Zauner et al. / NeuroImage 63 (2012) 674–686
dominant characteristic for almost the entire period of item presenta-
tions. However, in stark contrast to AB trials, a lack of phase concen-
tration – reflecting a ‘release’ from entrainment – can be observed for
Entrainment of phase, varied presentation
The general pattern of results is similar for the varied presenta-
tion. A significant phase distribution could be observed for T1 and
T2 (tαcrit=.01 in both cases) for AB trials. Phase angles also show
similar values (227° for T1 and 191° for T2). However, a significant
phase concentration with a mean phase angle of 244° can also be
seen for NoAB trials during T1.
Comparabletothe fixedpresentation,a lackofphaseconcentration–
reflecting a ‘release’ from entrainment – can also be observed for NoAB
trials inthe varied condition. However, a significant phase concentration
can be seen also for T1.
Influence of IAF on entrainment
The results of a 2-way ANOVA with factor IAF (subjects with IAF
around 10 Hz vs. IAF deviant from 10 Hz), factor TASK (fixed vs. var-
ied), and PLI at 10 Hz on electrode P4 as dependent measure showed
that individual alpha frequency had no significant influence on phase
locking at stimulation frequency. Neither factor IAF nor the interac-
tion reached or exceeded the 5%-level of significance.
Fig. 5. Mean phase angle for the 8–12 Hz filtered data. A) For fixed SOA, AB (left column) single trials showed significant deviation from uniform distribution with a mean phase
angle of 245° for T1 (top) and 161° for T2 (bottom) onset. NoAB trials were uniformly distributed (right column). B) For varied SOA, AB (left column) single trials showed a
non-uniform distribution with a mean phase angle of 227° for T1 (top) and 191° for T2 (bottom) onset. NoAB (right column) single trials were non-uniformly distributed at T1
onset (top) with a mean phase angle of 244°.
A. Zauner et al. / NeuroImage 63 (2012) 674–686
Frequency range of phase-locking
Our frequency analysis has focused on the broad alpha range of
8–12 Hz. In order to illustrate to what extent specific frequencies
are involved in the attentional blink phenomenon, we refer to the
time–frequency plots of the PLI as depicted in Fig. 8. The plots nicely
demonstrate that phase locking around T2 is largest in the broad
alpha band and also comprises the theta frequency range around
5 Hz but is not concentrated at stimulation frequency of 10 Hz
(loweralpha8–10 Hzvs.upperalpha10–12 Hz)andPLI(peristimulusto
the main effects nor for the interaction. This suggests that phase locking
around T2 is a broad-band phenomenon.
A comparison of phase locking between fixed and varied presentation
An interesting observation is that phase locking shows very simi-
lar results for the fixed and varied presentation blocks. For the alpha
band the peristimulus PLI (for the time interval of +/−50 ms preced-
ing and following the onset of T2) does not differ significantly
between the two blocks, as the results of a paired t-test (calculated
separately for each of the 5 ROIs) indicate. This lack of differences
most likely is due to the fact that T2 appears exactly 300 ms after
T1 in both blocks.
Phase locking in control trials
als. First of all, there is a methodological reason for this finding because
we have about twice as much control trials than AB or NoAB trials and
because the magnitude of the PLI is known to decrease with increasing
sample size (Kutil, 2011; Vinck et al., 2010). But there is possibly, in ad-
dition, yet another reason. Let us proceed from the fact that subjects
could not know whether T2 will be presented and let us assume that
half of the control trials exhibits large phase locking whereas the
other half shows only minor phase locking. In experimental trials the
half of trials with large phase locking constitute the AB trials, the
other half the NoAB trials. Exactly the same may hold true for control
averaging for– AB and NoAB trials is not possible because T2is missing.
As phase locking is calculated over all of the control trials, the variance
between all trials increases and the PLI decreases.
Fig. 6. Results of Kuiper's test for fixed SOA (8–12 Hz filtered data). A) AB (red), control (black) and all trials (grey) showed a non-uniform distribution over the time course
(continuous lines). Interestingly, NoAB trials (green) were uniformly distributed around T1 and T2 onset (broken lines). B) The phase-angle is illustrated for each condition
over the whole time course. C) Extract of phase-angle distribution for the time window around T1 and T2 appearance.
A. Zauner et al. / NeuroImage 63 (2012) 674–686
The central hypothesis for this study is that alpha entrainment
with minimal phase lag is a critical factor for the attentional blink
phenomenon. Thus, we have predicted that entrainment – during
the time window of T2 presentation – is larger for AB than NoAB tri-
als. The extent and kind of alpha entrainment can be judged by at
least three EEG parameters, the amplitude of the ERP (=SSVEP) in re-
sponse to stimulus onset – particularly to T1 and T2 – the extent of
alpha phase concentration (e.g., as measured by the PLI) during stim-
ulus onset, and the absolute phase angle of alpha during stimulus
As depicted in Fig. 3a the SSVEP consists of components with posi-
tive and negative peaks that have a pronounced frequency characteris-
tic in the alpha range (cf. Fig. 3b). The interesting observation here is
that in a time window of 100 ms preceding T2 a large negative peak
can be observed. At the time point of stimulus onset of T2, this negative
peak component (termed CT2) is much larger for AB as compared to
NoAB and control trials. Statistical analyses clearly show that these dif-
ferences are statistically significant for the CT2 peak amplitude of the
ERP, for evoked alpha power, for alpha PLI but not for alpha power
(whole power). These findings demonstrate that the differences in the
CT2 amplitude of the SSVEP are due to the phase locking of the
ferences in alpha (whole) power indicates. Thus, we can conclude that
at least part of the increase in the CT2 peak amplitude – which charac-
sized that the increased CT2 amplitude in AB trials is unlikely to be
caused by a high proportion of unconsciously seen items that did not
reach/exceed the ‘seen’‐threshold. The reason is that unconsciously
seen items generally tend to lower ERP amplitude (e.g., Hillyard et al.,
Another important aspect is the entrainment of alpha phase with
the negative peak during stimulus onset of T2. Phase analysis re-
vealed a significant phase concentration of alpha at the negative
peak during presentation onset of T2 for AB trials only. This is an im-
portant observation that is central for the evaluation of the hypothe-
sis that alpha phase entrainment – with minimal phase lag – leads to
an interference of those processes that underlie the generation of the
P1 with those underlying the encoding of the next following stimulus.
The reasoning is the following: if the alpha rhythm is entrained with
zero phase lag, as we observed for AB trials, the negative alpha peak
(reflected by an phase angle of 180°; cf. Figs. 5–7) coincides exactly
with the presentation onset of T2 and, as a consequence, the next
Fig. 7. Results of Kuiper's test for varied SOA (8–12 Hz filtered data). A) AB (red), control (black) and all trials (grey) showed a non-uniform distribution over the time course (continuous
distribution for the time window around T1 and T2 appearance.
A. Zauner et al. / NeuroImage 63 (2012) 674–686
negative peak coincides exactly with the ‘processing peak’ of early cat-
egorization of T2. Under normal conditions (i.e., in tasks with an SOA
longer than about 500 ms) the P1 – reflecting early categorization
(cf. Klimesch, 2011 for an extensive review) – would be generated in
exactly that time window, where in AB trials the negative alpha peak
is generated. The lack of a significant alpha entrainment in NoAB trials
may provide the possibility to avoid this interference between early
categorization of T2 and encoding of the next following, because these
two processes would not be completely time locked to each other.
We conclude that alpha entrainment alone is not sufficient to gen-
erate the AB phenomenon. It is rather the combination between en-
trainment and phase concentration at the negative peak that is
important. As illustrated in Figs. 6 and 7 entrainment can be observed
in AB trials over almost the entire trial, including the presentation of
Fig. 8. Time-frequency plots for the phase locking index (PLI). Note the strong phase-locking in the broad alpha range (8–12) for AB fixed (left, top) and varied (right, top) SOA
compared to NoAB (middle) and control (bottom).
A. Zauner et al. / NeuroImage 63 (2012) 674–686
T1. But despite of entrainment during the onset of T1, this target can
be reported. The difference between T1 and T2 lies in the phase angle.
During presentation onset of T1, alpha phase angle is close to 240°,
which represents the positive going slope close to the zero crossing
of the alpha oscillation, whereas during T2 the phase angle is 180°.
Thus, the next negative peak of the alpha oscillation (relative to the
presentation onset of T1) appears earlier at about 80 ms and, thus,
does not overlap with the P1 time window as in the case for T2. The
calculation is the following: the phase angle at T1 is 240°. Relative
to this time point, the next following negative peak appears at 180°
after the cycle is completed going from 240° to 360° (=120°). Thus,
the peak latency is 300° (180°+120°=300°) which represents 83 ms
(considering that 1°=0.278 ms when assuming a frequency of exactly
We have to emphasize that the varied presentation block showed
very similar results, behaviorally and electrophysiologically. Exactly
the same pattern of results can be observed, also showing that en-
trainment with zero phase lag is the critical factor for the AB phenom-
enon. This is surprising, because we have assumed that the jittered
presentation would lead to both, a reduction in the frequency of at-
tentional blink trials and a reduction of phase entrainment. The key
for understanding this finding may be seen in another surprising find-
ing: phase locking in the fixed presentation block is not concentrated
at stimulation frequency (of 10 Hz) but shows up in a broad frequen-
cy range as Fig. 8 demonstrates. This is unexpected because flicker
studies usually show a specific response at the stimulation frequency
(cf. e.g. Herrmann, 2001). However, studies examining the relation-
ship between stimulation frequency and individual alpha frequency
(IAF) have revealed that response frequency in the EEG may shift to-
ward IAF. As an example, Gebber et al. (1999) reported that for a sub-
ject with IAF=10.8 Hz and a stimulus frequency of 4.9 Hz, the
response frequency was largest at 9.8 Hz. In addition to IAF, alpha en-
trainment also depends on the individual power of alpha (Sakamoto
et al., 1993).
This phenomenon of ‘IAF-entrainment’, which means that the
largest EEG response is not exactly at stimulation frequency (10 Hz
in our case) but close to IAF, could be responsible for the observed
broad response frequency in the alpha band, as it is well known
that IAF exhibits a large interindividual variation. This interpretation,
however, is not very likely, because we found that IAF neither had a
significant influence on AB magnitude, nor on phase locking. We
also have to emphasize that the variation in IAF was not very large.
Thus, an evaluation of the ‘IAF-entrainment’ hypothesis is difficult
on the basis of our data. An alternative interpretation is based on
the idea that alpha is not a single rhythm with a frequency at IAF
but consists of a population of rhythms that vary (possibly in a task de-
pendent manner) around IAF. Accordingly, the responsiveness of alpha
with respect to entrainment may be understood as a broad-band phe-
nomenon. This interpretation could easily explain why alpha phase
locking is not diminished in the varied presentation block. The jittered
presentation reflects a variation in stimulation frequency, but due to
‘broad band’ or ‘population’ entrainment, the entrained frequencies be-
IAF-entrainment has primarily been studied in tasks where subjects
were passively exposed to flicker stimulation without any additional
task demands. The present attentional blink paradigm, however, com-
bines rhythmic stimulation with an increase in cognitive demands.
‘Population’ entrainment may not be diminished as long as the
jittered SOAs lie in the alpha frequency range, which is the case in
our study. The jittered SOA varied between 80 and 120 ms. The
short SOA with 80 ms represents a period of fast alpha at 12.5 Hz,
whereas 120 ms represents a period of slow alpha at 8.3 Hz. Behav-
ioral results with jittered SOAs show that the attentional blink can in-
deed be reduced (Martin et al., 2011) if the jitter is large, ranging
from 34 ms to 170 ms (reflecting a frequency of about 30 Hz and
5 Hz respectively). Thus, it appears plausible to assume that jittered
presentations reduce the attentional blink, but probably only if the
corresponding stimulation frequencies are well outside the alpha fre-
quency range, thereby avoiding alpha entrainment.
There may even be a third reason which may be responsible for
the lack of behavioral and electrophysiological differences between
the fixed and varied presentation. In both blocks, T2 appeared exactly
at 300 ms poststimulus to T1. This enables subjects to use temporal
expectation to improve their performance (cf. Mathewson et al.,
2010), particularly if we would assume that temporal expectation
has a preferential time span in the range of several hundred ms,
which could be characteristic for a sensory buffer.
Finally, let us address the question whether other findings linking
alpha and AB magnitude are consistent with our observations regard-
ing entrainment. Of particular interest here is the work by MacLean
and colleagues. In one experiment they observed that resting alpha
power is positively associated with AB magnitude (MacLean et al.,
2012). In another, they observed a complex interaction between
(alpha) ERD, AB magnitude and lag (MacLean and Arnell, 2011).
With regard to the earlier experiment, it should be noted that ERD
was measured in a foreperiod of 2 s, i.e., before the RSVP started. At
short lag (351 ms after T1) large ERD (reflecting large alpha power
suppression) was associated with low T2 accuracy (high AB magni-
tude), but at long lag (936 ms after T1) was associated with high T2
accuracy. The authors interpreted this finding in terms of an atten-
tional (over-)investment leading to an increase in ERD in a foreperiod
preceding the RSVP which in turn leads to an increased AB magnitude
at short lag. These findings are well in line with the hypothesis that
alpha reflects an inhibitory filter (Klimesch et al., 2007; Klimesch,
It is worth emphasizing that in our study suppression of T2 is as-
sociated with a specific phase response but not with a change in
power. This is well in line with the findings of other target detection
tasks which are performed under difficult perceptual conditions. As
an example, Hanslmayr et al. (2005) observed that good performers
showed smaller power during a foreperiod (immediately preceding
target detection) than bad performers. In addition, it was found that
good performers showed a significantly larger phase response (in
terms of phase locking) during task performance. This finding also sug-
gests that attentional investment leads to a decrease in power in a
foreperiod and to an increased phase response during target detection.
In summarizing, the present findings provide clear and strong
support for the hypothesis that alpha entrainment is a critical factor
for the attentional blink phenomenon. Further studies will be neces-
sary to determine the exact role played by alpha and how oscillation
in this particular frequency band relates to existing theories of the AB,
which emphasize a strong role for the process of transferring percep-
tual input into short-term visual memory. For example, alpha may re-
flect a top-down process that controls access to a memory trace of an
expected item (cf. Klimesch, 2011). This process may increase the
likelihood of entrainment and the resulting interference on the pro-
cessing of T2.
This research was supported by the Austrian Science Foundation
(FWF Project P21503-B18). The first author, Andrea Zauner, of this
article was financially supported by the Doctoral College "Imaging
the Mind" of the Austrian Science Fund (FWF-W1233).
Busch, N.A., Dubois, J., VanRullen, R., 2009. The phase of ongoing EEG oscillations pre-
dicts visual perception. J. Neurosci. 29, 7869–7876.
Chun, M.M., Potter, M.C., 1995. A two-stage model for multiple target detection in rapid
serial visual presentation. J. Exp. Psychol. Hum. Percept. Perform. 21, 109–127.
Di Lollo, V., Kawahara, J., Shahab Ghorashi, S.M., Enns, J.T., 2005. The attentional blink:
resource depletion or temporary loss of control? Psychol. Res. 69, 191–200.
A. Zauner et al. / NeuroImage 63 (2012) 674–686
Duncan, J., Ward, R., Shapiro, K., 1994. Direct measurement of attentional dwell time in
human vision. Nature 369, 313–315.
Foxe, J.J., Snyder, A.C., 2011. The role of alpha-band brain oscillations as a sensory sup-
pression mechanism during selective attention. Front. Psychol. 2, 154.
Gazzaniga, M.S., 1995. The Cognitive Neurosciences. MIT Press, Cambridge, MA.
Gebber, G.L., Zhong, S., Lewis, C., Barman, S.M., 1999. Human brain alpha rhythm:
nonlinear oscillation or filtered noise? Brain Res. 818, 556–560.
Gross, J., Schmitz, F., Schnitzler, I., Kessler, K., Shapiro, K., Hommel, B., Schnitzler, A.,
2004. Modulation of long-range neural synchrony reflects temporal limitations of
visual attention in humans. Proc. Natl. Acad. Sci. U. S. A. (PNAS) 101, 13050–13055.
Hanslmayr, S., Klimesch, W., Sauseng, P., Gruber, W., Doppelmayr, M., Freunberger, R.,
Pecherstorfer, T., 2005. Visual discrimination performance is related to decreased
alpha amplitude but increased phase locking. Neurosci. Lett. 375, 64–68.
Hanslmayr, S., Aslan, A., Staudigl, T., Klimesch, W., Herrmann, C.S., Bäuml, K.H., 2007.
Prestimulus oscillations predict visual perception performance between and with-
in subjects. NeuroImage 37, 1465–1473.
Hanslmayr, S., Gross, J., Klimesch, W., Shapiro, K.L., 2011. The role of alpha oscillations
in temporal attention. Brain Res. Rev. 67, 331–343.
Herrmann, C.S., 2001. Human EEG responses to 1–100 Hz flicker: resonance phenom-
ena in visual cortex and their potential correlation to cognitive phenomena. Exp.
Brain Res. 137, 346–353.
Hillyard, S.A., Squires, K.C., Bauer, J.W., Lindsay, P.H., 1971. Evoked potential correlates
of auditory signal detection. Science 172, 1357–1360.
Jensen, O., Mazaheri, A., 2010. Shaping functional architecture by oscillatory alpha
activity: gating by inhibition. Front. Hum. Neurosci. 4, 12.
Jolicoeur, P., 1999. Concurrent response-selection demands modulate the attentional
blink. J. Exp. Psychol. Hum. Percept. Perform. 25, 1097–1113.
Jolicoeur, P., Dell'Acqua, R., 1998. The demonstration of short-term consolidation.
Cogn. Psychol. 36, 138–202.
Kawaguchi, T., Jijiwa, H., Watanabe, S., 1993. The dynamicsof phase relationshipsof alpha
waves during photic driving. Electroencephalogr. Clin. Neurophysiol. 87, 88–96.
Kawahara, J., Enns, J.T., Di Lollo, V., 2006. The attentional blink is not a unitary phenom-
enon. Psychol. Res. 70, 405–413.
Klimesch, W., 2011. Evoked alpha and early access to the knowledge system: the P1 in-
hibition timing hypothesis. Brain Res. 1408, 52–71.
Klimesch, W., Sauseng, P., Hanslmayr, S., 2007. EEG alpha oscillations: the inhibition-
timing hypothesis. Brain Res. Rev. 53, 63–88.
Klimesch, W., Fellinger, R., Freunberger, R., 2011. Alpha oscillations and early stages of
visual encoding. Front. Psychol. 2.
Kranczioch, C., Debener, S., Engel, A.K., 2003. Event-related potential correlates of the
attentional blink phenomenon. Cogn. Brain Res. 17, 177–187.
Kranczioch, C., Debener, S., Maye, A., Engel, A.K., 2007. Temporal dynamics of access to
consciousness in the attentional blink. NeuroImage 37, 947–955.
Kutil, R., 2011. Biased and unbiased estimation of the circular mean resultant length
and its variance. Statistics 1–13.
Lakie, M., Combes, N., 1999. The phase of postural hand tremor is not influenced by re-
petitive photic brain stimulation. Clin. Neurophysiol. 110, 2020–2025.
Lazarev, V.V., Simpson, D.M., Schubsky, B.M., Deazevedo, L.C., 2001. Photic driving in
the electroencephalogram of children and adolescents: harmonic structure and re-
lation to the resting state. Braz. J. Med. Biol. Res. 34, 1573–1584.
MacLean, M.H., Arnell, K.M., 2011. Greater attentional blink magnitude is associated
with higher levels of anticipatory attention as measured by alpha event-related
desynchronization (ERD). Brain Res. 1387, 99–107.
MacLean,M.H.,Arnell, K.M., Cote,K.A., 2012. Resting EEG inalpha and beta bands predicts
individual differences in attentional blink magnitude. Brain Cogn. 78, 218–229.
Martin, E.W., Enns, J.T., Shapiro, K.L., 2011. Turning the attentional blink on and
off: opposing effects of spatial and temporal noise. Psychon. Bull. Rev. 18,
Mast, J., Victor, J.D., 1991. Fluctuations of steady-state VEPs: interaction of driven evoked
potentials and the EEG. Electroencephalogr. Clin. Neurophysiol. 78, 389–401.
Mathewson, K.E., Gratton, G., Fabiani, M., Beck, D.M., Ro, T., 2009. To see or not to see:
prestimulus alpha phase predicts visual awareness. J. Neurosci. 29, 2725–2732.
Mathewson, K.E., Fabiani, M., Gratton, G., Beck, D.M., Lleras, A., 2010. Rescuing stimuli
from invisibility: inducing a momentary release from visual masking with pre-
target entrainment. Cognition 115, 186–191.
Mathewson, K.E., Lleras, A., Beck, D.M., Fabiani, M., Ro, T., Gratton, G., 2011. Pulsed out
of awareness: EEG alpha oscillations represent a pulsed inhibition of ongoing cor-
tical processing. Front. Psychol. 2.
Olivers, C.N., Meeter, M., 2008. A boost and bounce theory of temporal attention.
Psychol. Rev. 115, 836–863.
Raymond, J.E., Shapiro, K.L., Arnell, K.M., 1992. Temporary suppression of visual pro-
cessing in an RSVP task: an attentional blink? J. Exp. Psychol. Hum. Percept. Per-
form. 18, 849–860.
Romei, V., Gross, J., Thut, G., 2010. On the role of prestimulus alpha rhythms over
occipito-parietal areas in visual input regulation: correlation or causation?
J. Neurosci. 30, 8692–8697.
Sakamoto, H., Inouye, T., Shinosaki, K., 1993. Preservation of alpha rhythm shortly after
photic driving. Int. J. Neurosci. 73, 227–233.
Schack, B., Klimesch, W., 2002. Frequency characteristics of evoked and oscillatory
electroencephalic activity in a human memory scanning task. Neurosci. Lett. 331,
Shapiro, K.L., Raymond, J.E., Arnell, K.M., 1994. Attention to visual pattern information
produces the attentional blink in rapid serial visual presentation. J. Exp. Psychol.
Hum. Percept. Perform. 20, 357–371.
Shapiro, K.L., Raymond, J.E., Arnell, K.M., 1997. The attentional blink. Trends Cogn. Sci.
Shils, J.L., Litt, M., Skolnick, B.E., Stecker, M.M., 1996. Bispectral analysis of visual inter-
actions in humans. Electroencephalogr. Clin. Neurophysiol. 98, 113–125.
Silberstein, R.B., 1995. Steady-state visually evoked potentials, brain resonances, and
cognitive processes. In: Nunez, P. (Ed.), Neocortical Dynamics and Human EEG
Rhythms. Oxford University Press, New York, pp. 272–303.
Silberstein, R.B., Schier, M.A., Pipingas, A., Ciorciari, J., Wood, S.R., Simpson, D.G., 1990.
Steady-state visually evoked potential topography associated with a visual vigi-
lance task. Brain Topogr. 3, 337–347.
Silberstein, R.B., Ciorciari, J., Pipingas, A., 1995. Steady-state visually evoked potential
topography during the Wisconsin card sorting test. Electroencephalogr. Clin.
Neurophysiol. 96, 24–35.
Tallon-Baudry, C., Bertrand, O., Delpuech, C., Pernier, J., 1996. Stimulus specificity of
phase-locked and non-phase-locked 40 Hz visual responses in human. J. Neurosci.
Thut, G., Schyns, P.G., Gross, J., 2011a. Entrainment of perceptually relevant brain oscilla-
tions by non-invasive rhythmic stimulation of the human brain. Front. Psychol. 2, 170.
Thut, G., Veniero, D., Romei, V., Miniussi, C., Schyns, P., Gross, J., 2011b. Rhythmic TMS
causes local entrainment of natural oscillatory signatures. Curr. Biol. 21,
Vinck, M., van Wingerden, M., Womelsdorf, T., Fries, P., Pennartz, C.M., 2010. The
pairwise phase consistency: a bias-free measure of rhythmic neuronal synchroni-
zation. NeuroImage 51, 112–122.
Ward, R., Duncan, J., Shapiro, K., 1996. The slow time-course of visual attention.
Cogn. Psychol. 30, 79–109.
A. Zauner et al. / NeuroImage 63 (2012) 674–686