Edited by: John J. Foxe, Albert Einstein College of Medicine, USA
Reviewed by: Hans-Peter Frey, Albert Einstein College of Medicine, USA; Seppo P. Ahlfors, Massachusetts General Hospital/Harvard Medical School, USA; István Czigler, Hungarian Academy of Sciences, Hungary
*Correspondence: Daniel Schneider, Leibniz Research Centre for Working Environment and Human Factors, Ardeystr., 67, D-44139 Dortmund, Germany e-mail:
This article was submitted to the journal Frontiers in Human Neuroscience.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The present study investigated the neural mechanisms that contribute to the detection of visual feature changes between stimulus displays by means of event-related lateralizations of the electroencephalogram (EEG). Participants were instructed to respond to a luminance change in either of two lateralized stimuli that could randomly occur alone or together with an irrelevant orientation change of the same or contralateral stimulus. Task performance based on response times and accuracy was decreased compared to the remaining stimulus conditions when relevant and irrelevant feature changes were presented contralateral to each other (contralateral distractor condition). The sensory response to the feature changes was reflected in a posterior contralateral positivity at around 100 ms after change presentation and a posterior contralateral negativity in the N1 time window (N1pc). N2pc reflected a subsequent attentional bias in favor of the relevant luminance change. The continuation of the sustained posterior contralateral negativity (SPCN) following N2pc covaried with response times within feature change conditions and revealed a posterior topography comparable to the earlier components associated with sensory and attentional mechanisms. Therefore, this component might reflect the re-processing of information based on sustained short-term memory representations in the visual system until a stable target percept is created that can serve as the perceptual basis for response selection and the initiation of goal-directed behavior.
According to the biased competition account of visual selective attention, goal-directed behavior in a rich visual environment depends on the representation of relevant visual signals in higher-level executive instances that are capable to process only a strongly limited amount of information. The biased competition account suggests that the strongest represented signals prevail in competitive visual processing and are therefore represented in highler-level areas (Desimone and Duncan,
Furthermore, information processing is not implemented in an exclusively unidirectional way with slow competitive interactions ranging from sensory to higher-level structures. It rather consists of iterative circuits between and within sensory and higher-level areas that can be differentiated into the feed-forward stream and the re-entrant processing of information (Lamme and Roelfsema,
The present study made use of event-related potentials (ERPs) of the EEG that were shown to be indicative of such feedback or re-entrant mechanisms (see Robitaille and Jolicoeur,
Subsequent to this posterior positivity, lateralized feature changes trigger multiple posterior asymmetric deflections in the ERP that are not specific to stimulus change and are caused by an increase in negativity contralateral to the most salient or currently relevant stimulus. While the posterior contralateral negativity in the N1 time window (N1 posterior contralateral or N1pc; 150–200 ms after stimulus presentation) is based on the physical characteristics of presented stimuli (Shedden and Nordgaard,
In the present study, participants were instructed to locate a relevant luminance change in two lateralized bars that was randomly presented simultaneous with an irrelevant orientation change of the same or contralateral bar (Wascher and Beste,
However, if SPCN is indeed associated with sustained representations in vSTM required for evaluating if selected visual information corresponds to current response tendencies or behavioral goals, this should also be indicated by a co-variation of its continuation with differing RTs within stimulus conditions. Therefore we defined 10 bins of trials within each change condition that were arranged according to the time required for responding to the location of the relevant luminance change. We hypothesize that a longer lasting pattern of posterior asymmetric activation in favor of the relevant luminance change is shown for slow compared to fast response trials. Additionally, if the mentioned posterior asymmetries associated with sensory and attentional processing (i.e., change-related positivity, N1pc, N2pc) and SPCN reveal comparable posterior topographies (see Robitaille and Jolicoeur,
12 participants (seven female) with an age between 20 and 30 years (
Participants were seated in front of a 22-inch CRT monitor (viewing distance 120 cm, 100 Hz) in a dimly lit chamber. Stimulus presentation was controlled by a VSG 2/5 graphic accelerator (Cambridge Research Systems, Rochester, UK). In the first stimulus display, two bars with an area of 0.76 cm2 and a 1:2.41 length-to-width ratiowere presented 1.1° lateral to a central fixation cross. Their luminance was either brighter (45 cd/m2) or darker (20 cd/m2) than the background (30 cd/m2), leading to a constant Michelson contrast of 0.2 to keep stimulus saliency comparable between dark and bright stimuli (Michelson,
Participants were instructed to respond by pressing a key with their left or right index finger on the side of the relevant luminance change. The orientation change was declared as irrelevant and was not predictive for the location of the luminance change. Accordingly, all trials only containing an irrelevant orientation change (ORI) were no-go trials. Participants were instructed to respond as quickly and accurately as possible. The characteristics of all bars in the first and second stimulus display were randomly intermixed to assure that the observer could not anticipate a luminance or orientation change on basis of the first stimulus display and thus had to process the feature change for solving the task. The intertrial interval varied between 2000 and 2500 ms (uniform distribution). The experiment was divided into eight blocks with 384 trials each. A 3-min break followed after each block to counteract confounding effects of fatigue during the experiment. This led to an overall experimental time of about 3 h. Compared to prior experiments with the same experimental setup (Wascher and Beste,
Responses were recorded from presentation of the first stimulus display up to 1200 ms after onset of the second display. Responses prior to 150 ms after the second display were categorized as “fast guesses.” Error trials involved missed responses (no response in the 1200 ms interval), fast guesses, responses at the wrong stimulus side and misplaced responses in the ORI no-go trials. The experimental effects on error rates were quantified by an analysis of variance (ANOVA) with the within-subject factors
EEG was recorded from 60 Ag/AgCl active electrodes (ActiCap, Brain Products, Gilching, Germany) affixed across the entire scalp according to the extended 10/20 System (Pivik et al.,
Posterior ERLs (PO7/PO8) were calculated from the ERPs of the CSD transformed EEG by subtracting the activation ipsilateral from the activation contralateral to a laterally presented stimulus. This difference wave served for addressing the visuospatial processing of the presented feature changes and provides valuable information about the contralateral vs. ipsilateral processing preference in the visual system (Wascher and Wauschkuhn,
The posterior ERLs for each change condition revealed multiple deflections in the course of visual processing that were classified according to their temporal occurrence and polarity (i.e., change-related positivity, N1pc, N2pc, SPCN). Mean amplitudes in a 20 ms interval oriented at the peak in the grand averaged ERL were used as indicators of these posterior asymmetries (see Figure 3A). Their reliable occurrence was tested by means of
Since ERLs cannot be obtained in single-trial EEG data (multiple trials are required for subtracting ipsilateral from contralateral activity referred to a lateralized visual event), and we were interested in the co-variation of ERLs and RTs, we introduce a new analysis procedure for estimating RT dependent ERLs (cf. Poli et al.,
In the present study, RTs were collected from online event triggers in the EEG. Initially, RTs were sorted and the corresponding trial number was collected, also. Following this, the single-trial data of the ERL channels of interest (i.e., PO7/PO8) were sorted according to the RTs, using the collected trial numbers. Then, RTs and single-trial EEG data were binned (separately for each channel). In the present study, 10 bins were calculated. Finally, ERLs were computed for each bin (see Figure 3B) according to the procedure already described. This resulted in an averaged number of trials per bin of 39.26 (
In order to yield more reliable estimates of the RT bins, a time range of interest, i.e., where the ERL effects were most prominent, was selected for statistical quantification. More specifically, the time range from 160 to 510 ms was divided into 10 segments (35 ms each), from which the corresponding averages were computed (see Figure 6). For statistical quantification, all averaged segments in the derived time window of interest in the RT bins for each experimental condition (except the NoGo ORI condition), were tested against zero via non-parametric bootstrapping (Efron and Tibshirani,
Error rates varied with change condition,
Comparable results were obtained on the basis of RTs. RTs varied with change condition,
The first posterior deflection in the PO7/PO8 ERL locked to the change-relevant second stimulus display appeared in a time window from 70 to 120 ms subsequent to stimulus presentation (change-related positivity; see Figure
Comparable results were observed for the subsequent N1pc component (150–190 ms after stimulus presentation; see Figure
In the current study, only the grand average ERL waveforms of the LUM and LOB conditions revealed both a negative peak in the N2 time window (200–250 ms subsequent to feature change presentation) and SPCN time window (at least 300 ms subsequent to feature change presentation). In the LOU condition, N2pc was merged to the strong N1pc complex and also SPCN appeared earlier compared to the LUM and LOB conditions (see Figure
Posterior ERLs revealed a negativity in the N2 time window (N2pc) for both the LUM,
Figure
Both the LUM and LOU condition revealed an early positivity (change-related positivity) and subsequent negativity (N1pc) in the ERL with a constant latency across RT bins. While N2pc in the LOU condition was merged to N1pc for all RT bins, it was more distinctly represented in the LUM condition, but also showed no consistent variation between RT bins. The early deflection in the ERL of the LOB condition (change-related positivity) was not so consistently shown, possibly due to the generally lower amplitude compared to the remaining change conditions. Yet, both N1pc (positive deflection) and N2pc (negative deflection) were clearly represented in the vincentized ERLs of this change condition and also revealed a constant latency across RT bins.
Regarding SPCN, both the vincentized ERL of the LOB condition and the LUM condition indicated a longer duration of posterior contralateral negativity with increasing RTs. This result pattern was further supported by the statistical analyses shown in Figure
Additionally, a positive ERL deflection subsequent to SPCN was shown for all change conditions containing a relevant luminance change (see Figure
The present study investigated the neural mechanisms that contribute to the detection of visual feature changes. Participants were instructed to locate a change of luminance in a fast succession of bilaterally presented bars that was randomly accompanied by an irrelevant change of orientation of the same or contralateral bar. Accuracy and RTs for change localization were impaired compared to the remaining luminance change conditions (LUM, LOU) when the task-irrelevant orientation change was presented contralateral to the target (LOB—contralateral distractor condition). Given the short presentation times and the short temporal offset of the two stimulus displays, the irrelevant change of orientation might have induced a lateral motion transient that distracted attention when relevant and irrelevant information were presented spatially separated and thus interfered with target processing (Beste et al.,
We suggest that different processes are involved in the localization of relevant feature changes that can be studied by means of event-related lateralizations of the EEG (ERLs; i.e., contralateral—ipsilateral difference) referred to the bilateral stimuli. Furthermore, we propose a new vincentizing method based on these ERLs to more closely investigate the link between perceptual or attentional mechanisms and behavioral performance.
A first positive deflection in the ERL appeared with a latency of about 90 ms subsequent to the feature change(s) and was associated with the dishabituation of neurons encoding the feature change compared to those encoding the contralateral non-changing stimulus (Luck and Hillyard,
Also subsequent N1pc was associated with the sensory representation of the bilateral stimuli. However, prior studies revealed an attentional bias in favor of relevant signals already at this early stage of processing (see Schneider et al.,
As laid down in the introduction, the later posterior contralateral negativity (SPCN) is associated with the maintenance of representations in vSTM subsequent to attentional selection until further cognitive operations with the fading neural representation of the stimulus material are completed (e.g., Prime et al.,
Furthermore, Figure
However, no comparable co-variation of SPCN continuation and RT was shown in the LOU condition, although the occurrence of SPCN in the grand averaged data suggests that a lateralized representation of relevant information was at least frequently created in vSTM (see Figure
One additional point should be considered in further research. In the current study, only exogenous factors (the different stimulus conditions), but not the attentional setting of the observer was varied. Participants were instructed to focus on luminance changes during the whole experimental session. Concerning the generalizability of the current results, it should also be investigated if a comparable late posterior asymmetry co-variating with RTs can be observed when participants attend to orientation changes and have to ignore occasionally presented luminance changes.
In conclusion, the present study replicated that an irrelevant orientation change interfered with the processing of the luminance change target when it was presented at a contralateral position (see Wascher and Beste,
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Khatuna Parkosadze for her help in participant recruitment and in organizing and conducting the experiments.
1We referred to SPCN as a delayed N2pc in prior studies. However, the suggested underlying mechanisms were the same described in the current manuscript.
2The positive ERL deflections in the N2 and subsequent time windows of the No-Go ORI condition might in fact represent an active suppression of the orientation change distractor that is required for response inhibition.
3It can be argued whether the late N2pc in the study of Hickey et al. (