Cortical Dynamics Underlying Perceived Visibility: An Event‐Related Potential Investigation of Forward Masking

ABSTRACT

Forward masking has been a fruitful approach for understanding the temporal organization of perception and the dynamics of visual processing. However, there is still limited knowledge on the neural correlates of forward masking and cortical processes underlying the modulations of perceived visibility under this masking type. Notably, the ERP components (event‐related potentials) associated with this paradigm have not been identified yet. Therefore, we designed an ERP study combined with a forward‐masking paradigm and recorded cortical activities while observers performed a contour discrimination task on the target. We manipulated stimulus onset asynchrony (SOA) and mask‐to‐target (M/T) contrast ratio to have different masking conditions. In line with previous research, the perceived visibility of the target significantly depended on SOA, with a strong suppression elicited by the preceding mask at short SOAs. More importantly, the magnitude of this brief inhibition was significantly enhanced by the M/T contrast ratio. ERP analyses revealed nonlinear interactions in both early and late positive components. In particular, we identified nonlinear suppressions in early P1 and late positivity, and the magnitude of these suppressions became larger for the short SOA and high M/T contrast ratio. On the other hand, the nature of interactions in N1 was facilitatory and became dominant in the longer SOA. We discuss these findings in light of previous empirical and modeling studies suggesting a three‐process explanation of forward masking. Our findings provide important insights into the ERP correlates of forward masking and highlight the diverse, multiprocess nature of contextual influences on perceptual dynamics.

This study examined target visibility under forward masking to uncover cortical processes underlying perceived visibility. EEG recording revealed nonlinear suppression of early (P1) and late positive components, while N1 showed facilitatory effects due to masking. These results provide ERP correlates of proposed neural pathway interactions in human vision and highlight the diverse and multiprocess nature of contextual influences on perceptual dynamics.

Abbreviations

ANOVA

analysis of variance

ECG

electrocardiogram

EEG

electroencephalogram

ERP

event‐related potential

ICA

independent component analysis

LP

late positive component

M

mask‐only

M/T

mask‐to‐target

MT

mask–target

P‐pathway

parvo‐dominated pathway

RECOD

retino‐cortical dynamics

SEM

standard error of mean

SOA

stimulus onset asynchrony

T

target‐only

V1

primary visual cortex

VAN

visual awareness negativity

1. Introduction

Vision illustrates a sophisticated dynamic system that enables organisms to effectively deal with vast amounts of information with the recruitment of diverse processing pathways. Visual masking has been a powerful approach to shed light on the dynamic interactions within and across processing pathways. In general, visual masking refers to a reduction in the visibility of a target stimulus due to the presentation of a mask (Bachmann and Francis 2013; Breitmeyer and Ogmen 2006). Under masking, the target stimulus appears to have low visibility (e.g., a decrease in perceived brightness/contrast) or might be entirely invisible to the observer. The onset timing between the target and the mask (stimulus onset asynchrony [SOA]) is one of the critical determinants of the masking‐induced suppression and the perceived visibility of the target, and this dependency is typically displayed with a masking function (Bachmann 1994; Breitmeyer and Ogmen 2000, 2006).

Various types of masking have been found to be fruitful in understanding the mechanisms underlying visual perception and sensory dynamics. By combining theoretical and experimental paradigms, previous research revealed distinct processes involved in the perceived visibility of an object under a forward‐masking paradigm called paracontrast (Breitmeyer et al. 2006; Kafaligonul et al. 2009). In this masking type, a spatiotemporally contiguous mask is presented before the target. Based on the SOA values, the visibility of the target stimulus can be suppressed (inhibition) or increased (facilitation). The inhibitory processes that interact with a facilitatory process lead to a complex morphology in the masking function. Accordingly, three putative processes have been proposed to be involved in paracontrast masking: brief inhibition, facilitation, and prolonged inhibition (Breitmeyer et al. 2006). Brief inhibition is effective when the mask precedes the target at short SOAs ranging from −10 to −30 ms, whereas prolonged inhibition becomes dominant over large SOAs of −450 ms. The simulation results of a model (retino‐cortical dynamics [RECOD] model) based on dual‐channel theory (Breitmeyer and Ganz 1976; Breitmeyer et al. 2006; Ogmen 1993; Ogmen et al. 2003) and neurophysiological findings (e.g., Benardete and Kaplan 1997; Maffei et al. 1970; Poggio et al. 1969; Singer and Creutzfeldt 1970) suggest that the brief inhibition is mainly driven by a relatively fast intrachannel inhibition in the (parvo‐dominated) sustained channel and realized in the center‐surround antagonism of classical receptive fields. On the other hand, the prolonged inhibition has been attributed to slower, higher level cortical processing and has been modeled as recurrent processing in the sustained channel (Berman et al. 1991; Nelson 1991; Ursino and La Cara 2004a; Ursino and La Cara 2004b). An enhancement in target visibility, called facilitation, is mainly observed in the SOA values between −20 and −110 ms. Based on the findings and approach by Bachmann (1994), the subcortical (nonspecific) pathways have been proposed to be the origin of facilitation (Kafaligonul et al. 2009). In the dual‐channel model of visual masking, the activity of a subcortical network multiplicatively gates the sustained activities, and this modulatory effect successfully accounts for the observed enhancement in the range of −20‐ and −110‐ms SOA values for different criterion content (perceptual tasks) and stimulus features (Breitmeyer et al. 2006).

The proposed mechanisms underlying paracontrast provide a comprehensive framework to investigate the involvement of distinct processes in perceived visibility and sensory dynamics at different stages. However, we have a limited understanding of the neural correlates because only a few neurophysiological studies incorporated paracontrast (forward) masking conditions in their experimental designs. In one of these studies, Macknik and Livingstone (1998) recorded activity from the primary visual cortex (V1) of monkeys and identified SOA‐dependent inhibition related to paracontrast. Under the conditions of paracontrast, a powerful suppression of the early activities occurred along with some suppression of a later component. Breitmeyer (2007) interpreted that the paracontrast mask primarily affects the early feedforward activity elicited by the target and, secondarily, the late reentrant (recurrent) activity. In other words, the secondary modulation of the reentrant activities was interpreted as the product of the changes in the early feedforward sweep. Kondo and Komatsu (2000) later examined the temporal characteristics of V4 activity under different SOA conditions of paracontrast. Compared to the Macknik and Livingstone (1998) study, the target and mask had shorter durations. For this brief stimulation, it is hard to identify early and late modulations in the activities corresponding to the target. However, similar to the findings by Macknik and Livingstone (1998), the results indicated an SOA‐dependent suppression. Maximum inhibition occurred with the simultaneous presentation of the target and mask. As the onset timing between the mask and target increased, the inhibition got weaker. Although these invasive recordings from different visual areas provide evidence for inhibitory mechanisms in the cortex, no behavioral measure was acquired during recordings, and concomitant behavioral indices of masking were not investigated. In an early study, Kaitz et al. (1985) recorded electroencephalogram (EEG) for different paracontrast conditions while observers performed a subjective rating on the target visibility. They identified SOA‐dependent modulations in both behavioral measures and cortical evoked potentials. In particular, the results indicated interactions in the evoked potentials elicited by temporally disparate target and mask. However, the temporal profile (SOA dependency) of these masking effects was different from that observed in behavioral rating measures. Therefore, there is still a limited understanding of cortical dynamics underlying these changes and associated mechanisms involved in perceived visibility.

In the present study, we aimed to understand the neural correlates and cortical dynamics underlying the perceived visibility of a target under forward (paracontrast) masking. Our primary focus was inhibitory mechanisms, and previous research indicated that the inhibitory processes involved in paracontrast become dominant when participants are engaged in a contour discrimination task (Breitmeyer et al. 2006; Kafaligonul et al. 2009). Moreover, manipulations in the contrast level of the mask (mask‐to‐target [M/T] contrast ratio) have been found to be effective in controlling the amplitude of the masking effects (e.g., Breitmeyer et al. 2006). Accordingly, we acquired EEG while participants performed a contour discrimination task on a visual target under different contrast ratios and SOA conditions. The modulations of perceived visibility were quantified by changes in behavioral performance values. Similar to previous research, we expected to have significant effects of SOA and M/T contrast ratio on target visibility. In the first stage of EEG analyses, we focused on identifying event‐related potentials (ERPs) over which nonlinear mask–target interactions take place. Then, we tested the SOA and contrast‐level dependency of these modulations. In line with prior behavioral and neurophysiological studies, we hypothesized that stronger masks would reduce target visibility and modulate target‐evoked ERPs in an SOA‐ and contrast‐dependent manner. According to the findings by Macknik and Livingstone (1998) and the interpretations of Breitmeyer (2007), we specifically predicted that masked targets would elicit reduced amplitudes in early visual ERP components, reflecting the suppression of feedforward processing, and potential modulations of later components associated with later cortical (and possibly reentrant/recurrent) processing. To our knowledge, there is no systematic and comprehensive study on the ERP correlates of forward masking/paracontrast, even though this masking paradigm has been commonly used in vision research. This approach, therefore, allowed us to determine the ERP components associated with forward/paracontrast masking and, consequently, to shed light on the cortical dynamics underlying perceived visibility.

2. Materials and Methods

2.1. Behavioral Prestudy

Previous research indicated that three processes (brief inhibition, facilitation, and prolonged inhibition) interact with different weights and lead to distinct paracontrast masking functions under different criterion content and stimulus parameters (Breitmeyer et al. 2006; Kafaligonul et al. 2009). Prior to the main EEG experiment, we conducted a behavioral prestudy to evaluate masking functions under our current settings and stimulus parameters. Using a contour discrimination task, we particularly aimed to identify key SOA conditions at which the contrast level of the mask (M/T ratio) leads to significant changes in behavioral performance and the perceived visibility of a target.

2.1.1. Participants

Eight adult volunteers (age range: 22–32 years) participated in this behavioral experiment. This sample size is consistent with previous behavioral studies of paracontrast masking (e.g., Breitmeyer et al. 2006; Kafaligonul et al. 2009) and has been shown to be sufficient for capturing changes in the morphology of the masking function and the three proposed processes underlying paracontrast masking. All participants had normal or corrected‐to‐normal vision. None of the participants had neurological disorders or psychiatric diagnoses. Participants gave their written informed consent and filled out a prescreening form before the experiment. All procedures were in accordance with the Declaration of Helsinki (World Medical Association 2013) and approved by the local ethics committee at Bilkent University.

2.1.2. Apparatus and Stimuli

Stimulus presentation, experimental paradigm, and data acquisition were controlled by MATLAB (The MathWorks, Natick, MA) with the Psychtoolbox extension (Brainard 1997; Kleiner et al. 2007; Pelli 1997). A 27‐in. LCD monitor with 1920 × 1080 resolution and 120‐Hz refresh rate was used to display the visual stimuli. The monitor was gamma corrected with SpectroCAL (Cambridge Research Systems, Rochester, Kent, UK) photometer. A photodiode connected to a digital oscilloscope (Rigol DS 10204B, 167 GmbH, Puchheim, Germany) was used to ensure accurate timing of stimuli and event markers, rather than relying solely on nominal values provided by the presentation software or hardware. The distance between the observer and the screen was approximately 57 cm. A chin rest was used to constrain head movements and maintain a stable viewing distance. Behavioral responses were collected via a standard computer keyboard. The experiment was held in a silent and dark room.

A central black fixation, which was a combination of bull's eye and crosshair, was chosen to minimize eye movements during experimental sessions (Thaler et al. 2013). The inner and outer diameters of the fixation were 0.2° and 0.6°, respectively (Figure 1a,b). The target and mask were vertically aligned and presented 3° above the fixation. The target was a 1.2° radius disk with a notch (0.3° × 0.3° square) either on its right or left. The mask was a surrounding and contiguous annulus with a 1.3° outer and 2° inner radius, having identical gaps (0.3° × 0.3° square) on both sides. These gaps were introduced to minimize feature inheritance/attribution from the target's notch to the mask (e.g., Werner 1935; Stewart and Purcell 1970; Herzog and Koch 2001; Ogmen et al. 2006). Both the target and the mask had a duration of 16.66 ms. The target and the background were displayed at fixed luminance values of 38 and 50 cd/m², respectively. The luminance of the mask was either 32 or 14 cd/m², corresponding to 1.5 and 3.0 M/T Weber contrast‐ratio conditions, respectively.

FIGURE 1.

Experiment design and visual stimuli. (a) Schematic representation of the stimuli and timeline for target trials. Each trial began with a variable prestimulus interval (1000 ± 150 ms). In the mask–target conditions (MT_Low and MT_High), a mask was presented for 16.66 ms, and after a variable SOA, the target was shown for 16.66 ms. There was no mask in the target‐only (T) condition. A notch was randomly positioned on the left or right side of the target, and participants were instructed to indicate the notch location via keypress within 1 s. (b) Schematic representation of the stimuli and timeline for nontarget trials. In these trials, there was either only a mask (M_Low and M_High) or no stimulus (NS). The observers passively fixated and did not perform any additional tasks.

2.1.3. Task and Procedure

Each trial started with the presentation of fixation for a variable prestimulus interval (1000 ± 150 ms). Then, the mask stimulus was shown for 16.66 ms. After a variable SOA, the target was presented for 16.66 ms. A notch was randomly placed either on the left or the right side of the target. After the stimulus offset, the observers used a keyboard to report the location of the notch on the target stimulus, whether it was on the right or left (two‐alternative forced‐choice, contour discrimination task). The participants were allowed to respond within 1000 ms. If no response was acquired within this time window, the trial was repeated later in the session (see below).

We employed a (2 × 9) repeated‐measures design with two levels of M/T contrast ratio (low: 1.5, high: 3.0) and nine stimulus onset asynchronies (SOAs: 0, −8.33, −16.66, −41.66, −83.33, −116.66, −158.33, −200, and −350 ms). There was an additional target‐only condition to measure baseline performance. An experimental session included 40 repetitions of each condition (18 mask–target and 1 target‐only), resulting in a total of 760 trials. All conditions were randomly presented and counterbalanced throughout a session. Each participant completed five experimental sessions, and there were brief rest breaks between sessions. In the end, each participant completed 200 trials per condition.

2.2. EEG Study

2.2.1. Participants

Twenty‐one volunteers took part in the main EEG study. Two participants were excluded due to excessive EEG artifacts and one participant was excluded due to having a poor performance on the contour discrimination task. This observer failed to follow instructions in more than 50% of the trials in some experimental blocks. Therefore, the data from 18 observers (age range: 20–32 years) were included in the data analyses. Two of these observers also participated in the behavioral prestudy. All other inclusion/exclusion criteria and ethical approval were the same as those used in the behavioral prestudy.

2.2.2. Design and Procedure

Based on the behavioral results of the prestudy, we identified two critical SOA (16.66 and 83.33 ms) conditions for the main EEG experiment. Together with the combination of low and high M/T contrast ratio, this led to four mask–target conditions (MT_Low or MT_High; 16.66‐ or 83.33‐ms SOA). There was an additional target‐only (T) condition. Most of the theories on masking formulate the effects of visual masking with nonlinear interactions between the neural signals elicited by the target and mask (Breitmeyer and Ogmen 2006; Francis 2000, 2003). Therefore, our EEG analyses used a baseline linear model to identify deviations from linearity as manifestations of nonlinear interactions. To carry out these analyses, we also included mask‐only (M_Low or M_High) and no‐stimulus (NS) conditions in the design (Figure 1b). The timeline and event markers in these conditions were the same as those in the target‐only (T) condition, but the stimulation was different. There was either a low or a high contrast mask in the mask‐only condition. In the NS condition, there was no visual stimulation except the central fixation point. In both mask‐only and NS trials, the observers passively fixated and did not perform a contour discrimination task. The target (i.e., MT_Low, MT_High, and T) and nontarget trials (i.e., M_Low, M_High, and NS) were presented in separate blocks of trials. The order of these blocks and the conditions in each block were randomized. At the end of these blocks, each observer completed 80 trials per condition. With the exception of these changes, all other stimulus parameters and experimental procedures were the same as those used in the behavioral prestudy.

Two of the observers took part in the behavioral prestudy. The observers, who did not participate in any vision experiment or behavioral prestudy, completed some practice sessions on the day before the main EEG experiment. Each participant first attended a brief fixation training with the eye‐tracker to test whether they could fixate according to the instructions. Then, each participant completed a short practice to become familiar with the visual stimuli and contour discrimination task. This session was a shorter version of the behavioral prestudy and included all the conditions with only seven trials per condition.

2.2.3. EEG Recording and Preprocessing

The recording and preprocessing of electrophysiological signals were similar to those described previously (Akyuz et al. 2020; Catak et al. 2022; Kaya and Kafaligonul 2019). Briefly, a 64‐channel MR‐compatible EEG system (Brain Products GmbH, Gilching, Germany) was used to record high‐density signals. The placement of scalp electrodes (Ag/AgCl) was according to the extended 10–20 system. Two of the electrodes were used as reference (FCz) and ground (AFz). To reduce noise level and achieve stable recording, the electrode impedances were kept below 10 kΩ by applying conductive paste (ABRALYT 2000 FMS, Herrsching‐Breitbrunn, Germany) and monitored during recording sessions. The signals were sampled at 1 kHz, band‐pass filtered between 0.016 and 250 Hz, and recorded with event markers via BrainVision Recorder software (Brain Products, GmbH).

The raw EEG signals were initially preprocessed offline using Brain Vision Analyzer 2.0 (Brain Products, GmbH). The data were down‐sampled to 500 Hz, and the cardiobalistic artifacts were removed using the signals from the ECG (electrocardiogram) channel. Then, the signals were offline rereferenced to common average and filtered with band‐pass (zero‐phase shift Butterworth, 0.5–100 Hz, 24 dB/octave) and notch (50 ± 2.5 Hz, 16th order) filters. The data were segmented into epochs (−850, 1000 ms) by setting target onset (corresponding time point in nontarget trials) as zero. An infomax independent component analysis (ICA) was applied to detect and correct ocular artifacts. An epoch was rejected when the eye blinks were concurrent with stimulus presentations. Bad channels were spatially interpolated using spherical spline interpolation (Perrin et al. 1989). In the end, a semiautomatic artifact rejection process was applied to detect and remove trials contaminated with oscillations over 50 mV/ms, voltage changes of more than 200 mV in 200 ms, or changes of less than 0.5 mV in a 100‐ms window. After these preprocessing steps, 3% of epochs/trials (SEM = 1.24%) were removed from the data. These trials were excluded from both behavioral and ERP analyses.

2.2.4. ERP Analysis

Further ERP analyses were performed on data from all 63 scalp electrodes (excluding the ECG electrode). As mentioned above, electrodes with poor signal quality were interpolated using spherical spline interpolation. These analyses were carried out with Fieldtrip Toolbox (Oostenveld et al. 2011) and custom MATLAB (The MathWorks) scripts. Each epoch was baseline corrected according to the mean voltage of the prestimulus baseline period (target trials: [−200, −85] ms; nontarget trials: [−150, 0] ms range), and the signals from each channel were averaged across trials to compute ERPs time‐locked to the stimulus onset. In visual masking paradigms, the observers perform a task on a primary target, and the mask acts as a secondary task‐irrelevant stimulation. A significant masking effect implies that the signals elicited by the mask interfere and interact with the processing primarily driven by the target. Theoretical studies on masking also emphasize dynamic nonlinear interactions between the representations of the target and the mask (Breitmeyer and Ogmen 2006; Francis 2000, 2003). Similar to previous research (Aydin et al. 2021; see also Kaya and Kafaligonul 2021), we used an application of the additive model to identify nonlinear interactions in neural signals. Our approach was to compare the ERPs elicited by the mask–target sequence (MT_Low and MT_High) with the synthetic summation (linear model) of target‐only (T) and corresponding mask‐only (M_Low and M_High) activities. Before the summation, the waveforms of the mask‐only condition were shifted in time to align with the mask onset in the corresponding mask–target sequences, according to the physical SOA values. Even though the observers did not perform any explicit task and passively fixated in the mask‐only conditions, there may be confounding factors in the summed waveforms due to the summation of common activities that exist in both target‐only and mask‐only trials (e.g., anticipatory slow potentials as reported by Walter et al. 1964). While baseline correction from −200 to −85 ms largely reduces such potentials, small residual contributions may remain. To minimize these residual effects and eliminate the double summation of common factors, we subtracted the waveform of NS trials, as implemented in Aydin et al. (2021). This additional correction was important to prevent potential misinterpretation of interaction effects when contrasting these M + T with those of comparable MT. In the end, these corrected synthetic waveforms were compared with those of corresponding target–mask conditions (MT vs. [M + T–NS]).

We first wanted to identify ERP components associated with the nonlinear suppression of a preceding mask. To achieve this goal, we used the neural signals of the short (16.66 ms) SOA condition because behavioral results indicated significant inhibition of a preceding mask‐only under this condition (see Section 3.1). We subtracted the synthetic waveforms from the corresponding mask–target condition (MT – [M + T–NS]) and combined these difference waveforms across different contrast‐ratio conditions. Significant deviations (p < 0.05) from the baseline zero level in the combined difference waveforms indicate nonlinear interactions. We further applied a cluster‐based permutation test to determine the spatiotemporal profile of significant deviations and to identify associated ERP components (Groppe et al. 2011; Maris and Oostenveld 2007). This approach clusters spatially and temporally adjacent samples with p values exceeding an uncorrected alpha level of 0.05 (i.e., a data point significantly deviating from the zero level). We required at least three neighboring electrodes to form a cluster. Then, the cluster‐level statistics were calculated by taking the sum of t values within a spatiotemporal cluster. A distribution of cluster‐level statistics was generated using Monte Carlo simulations with 10,000 permutations of the original data. In the end, the observed/empirical cluster‐level statistics were compared with the generated distribution. When the observed cluster was in the highest or lowest 2.5th percentile of the Monte Carlo generated distribution, it was considered to be unique and significant.

The electrodes that were part of the identified spatiotemporal cluster for at least 20 ms were chosen as exemplar electrodes. Based on the time range and exemplar electrode locations of identified clusters, we determined the ERP components associated with paracontrast masking. Within the range of each identified component, we calculated the peak latency and/or mean amplitude of the difference waveforms (TM – [T + M–NS]) for each SOA and M/T contrast‐ratio condition. To test the SOA and contrast‐ratio dependency of these metrics, we further applied a two‐way repeated‐measures ANOVA with SOA and contrast ratio as factors. Post hoc pairwise t tests were performed to elucidate the nature of a significant two‐way interaction. Multiple comparisons were corrected with Bonferroni correction.

3. Results

3.1. Behavioral Prestudy

For each observer, the accuracy of contour discrimination under target‐only (baseline) condition was subtracted from the performance value of each target–mask condition (9 SOAs × 2 M/T contrast ratios). This subtraction procedure allowed us to observe the masking effect in the contour discrimination task relative to the baseline level. A negative difference value indicates a suppression in target visibility, whereas a difference value above this baseline level points to facilitation. The averaged values for each contrast ratio were separately plotted as a function of SOA and displayed in Figure 2a. For both M/T ratios, a clear drop below baseline can be seen when the SOA was brief (SOA > −83.33 ms). As expected, the amount of this brief suppression was smaller for the low contrast‐ratio condition. When the SOA was around −83.33 ms, there was a recovery from this brief masking effect and the overall performance got close to the baseline level. Compared to the baseline level, there was a performance decrease in long SOA conditions (e.g., −350 ms), indicating prolonged inhibition of target visibility. The overall morphology of masking functions agrees well with previous studies of paracontrast and forward masking in contour discrimination tasks (Breitmeyer et al. 2006; Kafaligonul et al. 2009).

FIGURE 2.

The percent correct difference between mask–target and baseline (Δ performance) for each SOA and contrast‐ratio condition. The target‐only condition (dotted line) corresponds to the baseline (zero) level. The masking magnitude was quantified with the absolute value of Δ performance values, and the negative and positive values correspond to suppression and facilitation, respectively. The Δ performance values are plotted as a function of SOA for each contrast‐ratio condition (red curve: MT_Low, blue curve: MT_High). Error bars indicate standard error (±SEM) across participants. (a) Results of behavioral prestudy (n = 8). (b) Behavioral results of EEG experiment (n = 18).

A two‐way repeated‐measures ANOVA with SOA and M/T ratio as factors revealed a significant effect of SOA (F _8,56 = 35.58, p < 0.01, ${\eta }_p^2=0.836$) and M/T ratio (F _1,7 = 44.81, p < 0.01, ${\eta }_p^2=0.865$) and a significant two‐way interaction between these factors (F _8,56 = 13.87, p < 0.01, ${\eta }_p^2=0.665$). To further understand the nature of the two‐way interaction, we performed post hoc comparisons across M/T ratio conditions (low vs. high) for each SOA. Bonferroni‐corrected paired t tests indicated significant differences between contrast‐ratio conditions at 0 (t ₇ = 4.87, p = 0.018, Cohen's d = 1.72), −8.33 (t ₇ = 4.16, p = 0.036, Cohen's d = 1.47), and −16.66 ms (t ₇ = 6.73, p < 0.001, Cohen's d = 2.38) of SOAs. The magnitude of suppression significantly increased for the high contrast‐ratio condition at these SOA values. These behavioral results overall showed significant effects of contrast ratio in the range of brief inhibition. On the other hand, the prolonged inhibition was insensitive to contrast ratio.

3.2. EEG Study: Behavioral Results

Two SOA conditions, −16.66 and −83.33 ms, were selected for the EEG experiment. According to the prestudy results, there was a significant masking effect in the brief inhibition range, which was also modulated by the contrast level of the mask. The −16.66‐ms SOA condition was chosen to understand the neural correlates of these significant modulations. Alternatively, shorter SOA values (0 and −8.33 ms) could have been selected. The 0‐ms SOA was not chosen because it represents simultaneous, rather than forward, masking. The −8.33‐ms condition is close to the simultaneous masking and thus can show mixed effects of forward and simultaneous masking processes. Hence, we selected the −16.66‐ms condition to obtain a clearer picture of forward masking. On the other hand, there was no masking and contrast‐ratio modulation at −83.33 ms. This condition was included in the design because it could potentially serve as a control/baseline condition for both masking and contrast‐ratio modulations.

Behavioral data obtained during the EEG recording sessions are shown in Figure 2b. A two‐way repeated‐measures ANOVA on the difference values revealed significant main effects of SOA (F _1,17 = 69.57, p < 0.01, ${\eta }_p^2=0.804$) and contrast ratio (F _1,17 = 49.94, p < 0.01, ${\eta }_p^2=0.746$), as well as a two‐way interaction (F _1,17 = 55.35, p < 0.01, ${\eta }_p^2=0.765$). When the SOA was −16.66 ms, the masking effect became stronger as the contrast ratio was increased (t ₁₇ = −7.94, p < 0.001, Cohen's d = 1.87). However, there was no masking for −83.33 ms, and the normalized values were similar in both contrast‐ratio conditions. These results confirm that the main characteristics of the masking functions in the behavioral prestudy were preserved during EEG recordings.

3.3. ERP Results

Similar to the behavioral results, we found robust modulations in the evoked activities. We performed a cluster‐based permutation test on the combined difference waveform (MT vs. [M + T–NS]) of short SOA condition. By doing so, we first aimed to identify ERP components associated with nonlinear interactions leading to the brief suppression in forward masking. The test revealed two separate spatiotemporal clusters associated with nonlinear interactions between the representations of target and mask in the P1 component (110–160 ms, cluster‐level t _sum  = −1605.3, p = 0.0046) and the N1 component (160–200 ms, cluster‐level t _sum  = 2936.3, p = 0.0018) ranges. These clusters were primarily located over occipital and parieto‐occipital electrodes (Figures 3 and 4; see also evoked activities to target‐only in Figure 6). Figures 3 and 4 illustrate the grand‐averaged ERPs for these clusters, with the upper panels (a and b) showing the waveforms under different experimental conditions, and the bottom panel (c) depicting the corresponding voltage topographies averaged across the identified time windows for each cluster. Electrodes (exemplar sites) contributing to the significant clusters are highlighted on the scalp maps, emphasizing the occipital and parieto‐occipital distributions of the effects. We further tested the SOA and contrast‐ratio dependency of these interactions by applying two‐way repeated‐measures ANOVA tests to the mean amplitudes of difference waveforms in the P1 and N1 ranges. For the difference waveform in the P1 range, the ANOVA test on mean amplitudes showed significant main effects of SOA (F _1,17 = 12.20, p < 0.01, ${\eta }_p^2=0.418$) and contrast ratio (F _1,17 = 7.49, p < 0.02, ${\eta }_p^2=0.306$). There was no significant two‐way interaction between these factors (Table 1). There was a subadditive interaction (i.e., negativity due to MT < [M + T–NS]) in P1, and the magnitude of this interaction decreased (i.e., less negativity) as the SOA got longer. The magnitude of this interaction increased when the mask's contrast level (contrast ratio) was high. With regards to the N1 component, there was a supra‐additive interaction (MT > [M + T–NS]), and the magnitude of this interaction was SOA‐dependent (F _1,17 = 27.94, p < 0.01, ${\eta }_p^2=0.622$). The magnitude of the interaction became larger for the longer SOA value. There was no main effect of contrast ratio, but the two‐way interaction was significant (F _1,17 = 16.19, p < 0.01, ${\eta }_p^2=0.488$). Bonferroni‐corrected post hoc comparisons revealed that the mean amplitude did not significantly differ between low and high contrast ratios both at the 16.66 ms (t ₁₇ = 1.29, p = 1, Cohen's d = 0.304) and 83.33‐ms SOA (t ₁₇ = −2.66, p = 0.085, Cohen's d = 0.626). Further comparisons indicated that the amplitude of difference waveforms for both contrast ratios became larger as the SOA increased, but only the low contrast ratio remained significant after correction (low: t ₁₇ = 6.49, p < 0.01, Cohen's d = 1.529; high: t ₁₇ = 2.85, p = 0.055, Cohen's d = 0.672).

FIGURE 3.

Grand‐averaged activities across all participants and derived waveforms from the exemplar scalp sites (electrodes from the identified cluster, see Section 2.2.4) of the identified cluster in the P1 component range (n = 18). The activities for short (a) and long (b) SOA conditions are shown in separate columns. In the upper plots, the evoked activities to mask–target (MT) and derived waveforms (M + T–NS) are displayed with solid and dotted curves, respectively. Each contrast‐ratio condition is shown with different colors (red: low, blue: high). In the lower plots, the difference waveforms (MT – [M + T–NS]) are displayed by using the same conventions. The activities are time‐locked to the onset of the target (dashed line), and the onset of the mask is indicated by the dotted line. Shaded areas in the difference waveforms represent ±SEM, illustrating variability across participants in the measures used for further statistical analyses. The time window of identified clusters is highlighted by gray rectangles. (c) Voltage topographical maps of the difference (MT – [M + T–NS]) waveforms averaged within the identified time range (110–160 ms). The voltage topographical maps of contrast‐ratio and SOA conditions are shown in separate rows and columns. The final subtracted difference waveforms across contrast ratios are shown in the topographical maps (left: short SOA = 16.66 ms, right: long SOA = 83.33 ms). The electrodes that were part of the identified spatiotemporal cluster for at least 20 ms were chosen as exemplar electrodes and marked by filled circles in the middle topographical maps.

FIGURE 4.

Grand‐averaged activities across all participants and derived waveforms from the exemplar scalp sites (electrodes from the identified cluster, see Section 2.2.4) of the identified cluster in the N1 component range (n = 18). The activities for short (a) and long (b) SOA conditions are shown in separate columns. In the upper plots, the evoked activities to mask–target (MT) stimulation and derived waveforms (M + T–NS) are displayed. The difference waveforms (MT − [M + T–NS]) are displayed in the lower plots. (c) Voltage topographical maps of the difference (MT – [M + T–NS]) waveforms averaged within the identified time range (160–200 ms). The voltage topographical map of contrast‐ratio and SOA conditions are shown in separate rows and columns, respectively. The final subtracted difference waveforms across contrast ratios are indicated in the middle topographical maps (left: short SOA, right: long SOA). Other conventions were the same as those in Figure 3.

FIGURE 6.

Grand‐averaged activities across all participants in target‐only and nontarget trials (n = 18). The activities from the exemplar scalp sites (electrodes from the identified clusters, see Section 2.2.4) of the identified clusters in the P1 (a) and N1 (b) and late positivity (c) are shown in separate columns. In each ERP plot on the left, each condition is displayed in a different color (target‐only: black, red: mask‐only with low contrast, blue: mask‐only with low contrast, no stimulus: gray). The activities were time‐locked to the onset of the target or corresponding event marker in nontarget trials. The time windows of identified clusters are highlighted by gray rectangles. The exemplar sites are marked by filled circles on a head model. Voltage topographical maps averaged within the identified time window of each component are indicated on the right. M_High: mask‐only with high contrast, M_Low: mask‐only with low contrast, NS: no stimulus, T: target‐only.

TABLE 1.

The outcome of two‐way repeated‐measures ANOVAs.

		Factor	F 1,17	p	${\eta }_p^2$
P1	Mean amplitude	SOA	12.20	0.003	0.418
		Contrast ratio	7.49	0.014	0.306
		SOA × CR	0.12	0.732	0.007
	Peak latency	SOA	8.41	0.010	0.331
		Contrast ratio	2.35	0.143	0.122
		SOA × CR	0.66	0.428	0.037
N1	Mean amplitude	SOA	27.94	< 0.001	0.622
		Contrast ratio	0.35	0.561	0.020
		SOA × CR	16.19	< 0.001	0.488
	Peak latency	SOA	2.48	0.134	0.127
		Contrast ratio	4.17	0.057	0.197
		SOA × CR	5.60	0.030	0.248
Late positivity	Mean amplitude	SOA	32.29	< 0.001	0.655
		Contrast ratio	6.93	0.017	0.290
		SOA × CR	0.75	0.398	0.042

Note: The threshold for significance was set at p < 0.05, and significant p values are highlighted in bold.

An inspection of the difference waveforms suggests that peak latencies may be affected by SOA and contrast ratio. For instance, the average waveforms illustrate that the long SOA conditions peaked earlier in the P1 component range. To determine whether there were consistent changes in latencies, we computed the peak latencies of difference waveforms in the range of each component and performed two‐way repeated‐measures ANOVA tests. The outcome of ANOVA tests revealed that there was a significant main effect of SOA (F _1,17 = 8.41, p = 0.010, ${\eta }_p^2=0.331$) on the early interactions in the P1 range, indicating earlier peaks for the short SOA conditions. However, there was no significant differences across contrast ratios or a significant two‐way interaction (Table 1). The ANOVA test on the interactions in the N1 component range did not reveal any significant main effects of SOA or contrast ratio. However, the two‐way interaction between these factors was significant (F _1,17 = 5.60, p = 0.030, ${\eta }_p^2=0.248$). Bonferroni‐corrected pairwise comparisons were conducted to further examine this interaction. For the low contrast ratio, when the SOA became longer, the difference waveform peaked earlier (t ₁₇ = −2.34, p = 0.032, Cohen's d = 0.551). On the other hand, the change in peak latency was in the opposite direction for the high contrast‐ratio condition, and the peak was significantly later for the long SOA (t ₁₇ = 2.84, p = 0.011, Cohen's d = 0.670).

We identified two additional clusters in the late positive activities (300–350 ms, cluster‐level t _sum  = −864.5, p = 0.0230; 350–600 ms, cluster‐level t _sum  = −8851.5, p = 0.0002). These clusters were mainly over central, parietal, and centro‐parietal electrodes and extended to occipital sites (Figure 5, see also evoked activities to target‐only condition in Figure 6). Figure 5 illustrates the grand‐averaged ERPs for these clusters, with the upper panels (a and b) showing the waveforms under different experimental conditions, and the bottom panel (c) depicting the corresponding voltage topographies averaged across the identified time windows for each cluster. Electrodes contributing to the cluster are highlighted on the scalp maps, emphasizing their centro‐parietal distributions. The mask suppressed the late positive activities (MT < [M + T–NS]). We further tested the SOA and contrast‐ratio dependency of these nonlinear modulations. For each SOA and ratio condition, we computed the mean difference potentials within 300–600 ms over the identified electrodes and then applied a two‐way repeated‐measures ANOVA on these mean values. The ANOVA test revealed a significant main effect of SOA (F _1,17 = 32.29, p < 0.001, ${\eta }_p^2=0.655$). Compared to the 83.33‐ms SOA, the mean difference potentials were lower, and the magnitude of subadditive interaction was larger for the brief SOA condition. There was also an overall main effect of contrast ratio (F _1,17 = 6.93, p < 0.02, ${\eta }_p^2=0.290$). Similarly, an increase in the contrast level of the mask led to larger interactions. However, there was no significant two‐way interaction between SOA and contrast ratio (Table 1).

FIGURE 5.

Grand‐averaged activities across all participants and derived waveforms from the exemplar scalp sites (electrodes from the identified cluster, see Section 2.2.4) of the identified clusters in the late positivity (n = 18). The activities for short (a) and long (b) SOA conditions are shown in separate columns. In the upper plots, the evoked activities to mask–target (MT) and derived waveforms (M + T–NS) are displayed. The difference waveforms (MT − [M + T–NS]) are displayed in the lower plots. (c) Voltage topographical maps of the difference (MT – [M + T–NS]) waveforms averaged within the identified time range (300–600 ms). The voltage topographical map of contrast‐ratio and SOA conditions are shown in separate rows and columns. The final subtracted difference waveforms across contrast ratios are indicated in the middle topographical maps (left: short SOA, right: long SOA). Other conventions were the same as those in Figure 3.

We further examined the contrast dependency of evoked activities in the mask‐only conditions. In particular, we tested whether the contrast level alters the averaged amplitudes in the range of identified components (Figure 6). There were no significant differences across the mean ERP amplitudes of contrast levels (M_Low vs. M_High) in the identified range of the P1 component (110–160 ms, t ₁₇ = 1.51, p = 0.149, Cohen's d = 0.356) and the late positivity (300–600 ms, t ₁₇ = 1.79, p = 0.091, Cohen's d = 0.422). As shown in Table 1, the contrast level of the mask significantly modulated the nonlinear interactions in these components. Therefore, the observed significant effects of mask contrast level on these interactions were not driven solely by contrast level changes in the evoked activities to mask‐only stimulation, highlighting the importance of neural activities driven by both mask and target stimulations (i.e., MT) and their interactions. In the range of the N1 component, the high contrast level (M_High) elicited significantly larger mean amplitudes compared to the low (M_Low) condition (160–200 ms, t ₁₇ = 2.43, p = 0.027, Cohen's d = 0.572). On the other hand, there was no main effect of contrast ratio (or a significant two‐way interaction between contrast ratio and SOA) on the supra‐additive interaction observed in the component range (Table 1). It is important to note that the amplitudes of all components got larger in target‐only conditions because the visual saliency/parameter of the target is different and the observers performed a contour discrimination during these trials (Figure 6). In particular, both P1 and N1 components are known to index early contour and low‐level visual feature processing. In our data, the P1 amplitude occasionally exceeds the N1 amplitude, likely reflecting the contour salience and low‐level visual features of the stimuli (Brodeur et al. 2008), consistent with prior reports using similar visual displays.

4. Discussion

In the present study, we investigated the perceived visibility of a target under different onset timing (SOA) and the contrast level of a preceding mask (M/T contrast ratio) by employing a contour discrimination task on the target. In agreement with previous studies, the behavioral results indicated significant suppressive effects of the mask on target visibility, particularly when the SOA between the mask and target was brief. The neurophysiological findings further provided important insights into the cortical correlates of the brief inhibitory process involved in forward masking. The analyses of the difference waveforms revealed nonlinear neural interactions between the representations of the mask and the target in both early ERP components located over occipital and occipito‐parietal electrodes and late positivity spread over centro‐parietal sites. Furthermore, the M/T contrast ratio significantly influenced these interactions, albeit in different ways for the early and late components. Our findings highlight the diverse nature of forward‐masking effects on the cortical processing of perceived visibility.

4.1. Alterations in Perceived Visibility

Previous research suggested that the morphology of the paracontrast function is determined by a combination of brief inhibition, facilitation, and prolonged inhibition, and the contribution of each process depends on criterion contents and stimulus parameters (Breitmeyer et al. 2006; Kafaligonul et al. 2009). Notably, the enhancement/facilitation in target visibility becomes dominant at SOAs of about −50 to −100 ms when subjects perform a surface‐based brightness judgment task. On the other hand, the prolonged and brief inhibitions mainly contribute to the masking function obtained with a contour detection/discrimination task. In line with these previous observations, the brief and prolonged inhibitions were dominant for both M/T contrast‐ratio conditions because we used a contour discrimination task in our design. Another important factor is stimulus parameters. For instance, when the M/T contrast ratio is greater than 1.0, strong Type A forward‐masking effects are typically obtained, and the masking function becomes a monotonic decreasing function in the negative SOA range, reaching its minimum at SOA = 0 (Breitmeyer and Ogmen 2006). Our behavioral results are mainly in agreement with these basic characteristics, and indeed, a Type A masking function was expected because both of the M/T contrast‐ratio values were higher than 1.0.

Compared to the SOA range of prolonged inhibition, the contrast‐ratio modulation of the masking effect was much more prominent in the brief inhibition, and an increase in the M/T contrast ratio strengthened masking at short but not long SOAs. Within the framework of the dual‐channel approach and RECOD model, it has been suggested that the brief inhibition is mainly driven by lateral inhibitions and center‐surround antagonistic receptive field properties of neurons within the parvo‐dominated (P) pathway (Ogmen et al. 2003; Kafaligonul et al. 2009). On the other hand, the model simulations suggest that long‐lasting recurrent inhibitions within the P‐pathway lead to prolonged inhibitions (Breitmeyer et al. 2006). Therefore, the brief inhibition is expected to be more prone to low‐level stimulus properties. In line with this prediction, Kafaligonul et al. (2009) revealed that a change in the polarity of the mask can lead to major changes in the brief inhibition and successfully explained these changes with lateral inhibition. Similar to the findings by Breitmeyer et al. (2006), we additionally found that a change in the contrast level of the mask mainly affects the brief inhibition observed in the short SOAs. These experimental findings overall confirm that low‐level stimulus properties lead to relatively smaller changes in the long‐lasting recurrent inhibitions within the P‐pathway. It is important to note that, while our design closely followed earlier paracontrast masking studies (e.g., Breitmeyer et al. 2006; Kafaligonul et al. 2009), the target and mask were presented 3° above fixation with a target size of 1.2°. Therefore, these findings should be interpreted with care when generalizing to foveal presentations or different target sizes, as masking effects may vary with eccentricity and stimulus dimensions (Breitmeyer and Ogmen 2006).

4.2. ERP Correlates

We identified the ERP components associated with the neural interactions between representations of mask and target. There was a subadditive interaction (MT < [M + T–NS]) in the P1 component range, suggesting a suppression elicited by a preceding mask. The magnitude of this suppression was dependent on SOA and contrast ratio. Similar to the behavioral results, the amount of inhibition was smaller for the long SOA and got larger when the contrast level of the mask was increased. On the other hand, there was no interaction between these two factors, and the contrast level similarly modulated this suppression for both SOAs. In masking studies, the P1 component has been associated with early processing of visual information and feedforward inhibitory mechanisms (Fahrenfort et al. 2007; Railo et al. 2011). Given that significant neural interactions were revealed in the SOA range of brief inhibition (Breitmeyer et al. 2006; Kafaligonul et al. 2009), P1 modulations may reflect cortical representations of lateral inhibitions. Together with the evoked activities of the mask‐only conditions, our findings indicate that both evoked activities and the neural interactions in the P1 component range depend on low‐level stimulus properties such as contrast level.

Using a different timeline of events (target/mask durations, interstimulus intervals/onset timing) in paracontrast and metacontrast, Macknik and Livingstone (1998) differentiated paracontrast and metacontrast with the inhibitions elicited by the mask on the V1 (primary visual cortex) neurons. They indicated that paracontrast leads to a main inhibition of the initial response to the target (onset) along with some suppression of later components/responses. On the other hand, according to their findings, metacontrast is mainly characterized by the inhibition of after‐discharge/later components. Indeed, it had virtually no effect on the initial response. The early nonlinear interactions over the P1 component observed in the current study had a similar time range to the initial inhibitions elicited by paracontrast and may be driven by these feedforward inhibitions in V1. Breitmeyer (2007) argued that the later modulations might be a reflection of these initial response changes at the later stages of visual processing (e.g., recurrent/reentrant activities). In other words, because the initial feedforward activity is inhibited by paracontrast, the later activities and processing stages emanating from the initial activity will also be activated less and lead to modulations and suppressions in later components.

In line with this interpretation, our EEG results indicated modulations in later ERP components. We found an SOA‐dependent modulation in the range of the N1 component. This component has been associated with discrimination tasks at the attended location (Vogel and Luck 2000) and with awareness of the target stimulus in both masking paradigms (e.g., Fahrenfort et al. 2007; Foxe et al. 2022) and nonmasking paradigms using near‐threshold stimuli (Koivisto and Revonsuo 2010). However, the nature of masking effects was different than those in the P1 component range. As indicated by the difference waveforms, there was a supra‐additive interaction (MT > [M + T–NS]), and interestingly, the magnitude of this facilitation increased when the SOA got longer. Rather than a main effect of contrast ratio, there was a differential effect (i.e., two‐way interaction), and the difference between the two ratio conditions became significant for the long SOA condition. Beyond the N1 component, we found additional interactions in the late positive component (LP) spreading over centro‐parietal electrodes. Similar to the P1 component, there was an SOA‐dependent supra‐additive interaction, and the magnitude of this suppression increased when the SOA became shorter. Moreover, an increase in contrast ratio enhanced this suppressive effect of the mask. The modulations in this late component were also observed in previous backward masking studies (Aydin et al. 2021; Catak et al. 2024; Del Cul et al. 2007; Fahrenfort et al. 2007). It has been argued that these late modulations reflect the neural correlates of perceptual reports and awareness, while the relatively later interactions beyond 150 ms represent subliminal recurrent processing over occipital‐temporal sites. In the current study, the suppressions elicited by the preceding mask in the P1 and late positivity ranges were similar, and the main effects of SOA and contrast ratio in these suppressions were in line with the changes in behavioral percentage values. These early and late modulations in ERP components may be associated with the mechanisms underlying brief inhibition. The additional modulations in the N1 component range might reflect the proposed facilitatory mechanisms incorporated in the dual‐channel approach and RECOD model because these modulations were supra‐additive neural interactions and became dominant at the long SOA condition (i.e., 83.33 ms falling in the SOA range of proposed facilitation).

It is also conceivable that the late visual‐evoked potentials observed in our study reflect not only central sensory processing but also higher order cognitive processes such as working memory. Previous research has shown that late positive components, including the P3, are modulated during working memory tasks (Kok 2001; Polich 2007). However, our paradigm consisted of a low load (single item, the target) and a perceptual discrimination task that required no sequential comparisons or active maintenance of stimulus representations, which are hallmark features of working memory paradigms. Thus, while the absence of such task demands and the lack of prolonged behavioral inhibition in our results favor a low‐level sensory origin for the observed effects, the possibility of minor contributions from working memory or attentional mechanisms cannot be fully excluded. This interpretation is consistent with theoretical accounts linking late ERP components to conscious access and working memory engagement (Dehaene and Changeux 2011; Sergent and Dehaene 2004), whereas earlier components (e.g., N2 and visual awareness negativity [VAN]) are more closely associated with phenomenal consciousness (Förster et al. 2020). Moreover, visual masking studies indicate that conscious registration depends on whether stimuli are transferred from sensory memory into working memory, with metacontrast masking strongly modulating this transfer (Bachmann 1994; Breitmeyer and Ogmen 2006; Ogmen and Herzog 2016). Taken together, our findings are best explained by sensory inhibition mechanisms, though the possibility of additional working memory‐related contributions to late ERP modulations remains an interesting question for future investigation.

Our analyses were restricted to post–target‐onset activity, in line with our primary aim of investigating how a preceding mask influences target‐evoked responses. This focus allowed us to characterize the temporal dynamics and ERP components associated with mask–target interactions across different SOA and contrast‐ratio conditions. While appropriate for our paradigm, this approach does not account for potential modulatory effects of prestimulus neural activity, such as slow cortical potentials or ongoing oscillatory dynamics (e.g., alpha power), which have been shown to influence poststimulus ERP amplitudes and perceptual outcomes (e.g., Iemi et al. 2019). Future studies could systematically examine these factors, for example, through analyses of baseline oscillatory power, slow potential shifts, or phase‐dependent effects, to more fully elucidate the interplay between prestimulus brain states and forward‐masking mechanisms. Such investigations would complement the current findings and provide a more comprehensive understanding of the oscillatory dynamics underlying perceived visibility.

5. Conclusion

Although many electrophysiological studies investigated the neural correlates of backward masking (e.g., metacontrast), only a few studies addressed forward masking and paracontrast. To our knowledge, our study provides the first systematic ERP investigation of paracontrast. We identified nonlinear suppressions in both early P1 and late positivity. Based on the SOA and contrast‐ratio dependency, these modulations may be associated with the brief inhibition involved in forward masking and paracontrast. Our analyses revealed an additional facilitation in the N1 component range that became dominant at longer SOA values close to −100 ms. These interactions might be a cortical representation of the proposed facilitatory process. Our findings overall highlight the diverse nature of masking and paracontrast effects on the dynamic processing of visual information.

Author Contributions

Afife Turker: data curation, formal analysis, investigation, methodology, software, visualization, writing – original draft. Haluk Ogmen: conceptualization, supervision, validation, visualization, writing – review and editing. Hulusi Kafaligonul: conceptualization, funding acquisition, project administration, resources, supervision, validation, visualization, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/ejn.70284.

Turker, A. , Ogmen H., and Kafaligonul H.. 2025. “Cortical Dynamics Underlying Perceived Visibility: An Event‐Related Potential Investigation of Forward Masking.” European Journal of Neuroscience 62, no. 8: e70284. 10.1111/ejn.70284.

Data Availability Statement

The dataset and analysis tools of the current study are available from the corresponding author on request. Any access to the dataset will be in accordance with the informed consent signed by the participants.