{"paper_id":"193dde2f-4797-4def-96c3-d8fa254d09fa","body_text":"1 \n \n \nChanges in aperiodic (1/f slope) activity during a picture-word interference task: \nEffects of congruency and sequence manipulations \n \nVirginia Tronelli*1,2, Patrycja Kałamała*3, Gabriele Gratton*4,5, Monica Fabiani4,5, Mate \nGyurkovics6, Kathy A. Low5, Maurizio Codispoti2, and Andrea De Cesarei2  \n \n1 Center for Mind/Brain Sciences (CIMeC), Universitá degli Studi di Trento-Rovereto, \nItaly \n2 Department of Psychology, Universitá degli Studi di Bologna, Italy \n3 Centre for Cognitive Science, Jagiellonian University, Poland \n4 Department of Psychology, University of Illinois Urbana-Champaign, USA \n5 Beckman Institute for Advanced Science and Technology, University of Illinois \nUrbana-Champaign, USA \n6 School of Psychology, University of East Anglia, Norwich, UK  \n \n*Drs. Tronelli, Kałamała, and Gratton contributed equally to this article as first authors. \n \nCorresponding Authors:  \nProf. Gabriele Gratton, grattong@illinois.edu \nProf. Andrea De Cesarei, andrea.decesarei@unibo.it \n \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n2 \n \nAbstract \nAperiodic neural activity (1/ f EEG) has been proposed to reflect the balance between \nexcitatory and inhibitory (E:I) inputs, with steeper spectral slopes reflecting increased \ninhibition and flatter slopes indicating excitation. This activity also reflects the temporal \ncoordination of n eural firing, offering insights into fundamental brain dynamics. Recent \nstudies have shown that the 1/f slope is sensitive to stimulus onset, characterized by initial \ninhibitory shifts followed by excitatory rebounds, which ma y reflect cognitive control \nmechanisms involved in suppressing distractions and preparing goal-directed responses. \nHowever, previous works have relied on fixed temporal windows and  insufficient control \nof ERP contamination, limiting our understanding of rapid control dynamics. Here we used \nnewly developed time-resolved analyses to study 1/ f spectral slope modulation during a \nPicture-Word Interference task, focusing on two canonical cognitive control markers: the \nCongruency Effect (CE) and Congruency Sequence Effect (CSE). Forty-nine participants \ncategorized pictures while ignoring congruent or incongruent words. Behaviorally, we \nreplicated robust CE and CSE patterns. Spectral slope analyses showed that incongruent \ntrials elicited steeper slopes — consistent with increased inhibition — particularly in frontal \nand central regions, reflecting conflict -related control engagement. Moreover, CSE \nanalyses revealed dynamic slope modulations across frontal, central, and occipital \ncomponents over time, suggesting contr ol adjustments influenced by previous trial \ncongruency. These results provide the first fine -grained evidence that aperiodic 1/f EEG \nactivity can track both immediate conflict resolution and cognitive adjustments, offering a \ntemporally sensitive neural marker of cognitive control through modulation of E:I balance.  \n \n \n \n \nKey Words: aperiodic 1/f EEG; cognitive control; neural noise; spatial distribution; \ncongruency effect; congruency sequential effect   \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n3 \n \n1. Introduction \nAperiodic neural activity (also called 1/ f noise due to its shape in the power \nspectrum) has gained increasing recognition as a potential marker of cortical \nExcitation:Inhibition (E:I) balance in electroencephalographic (EEG) research (Ahmad et \nal., 2022; Gao et al., 2017). Recently, cognitive neu roscientists have begun to link \naperiodic activity and E:I balance with cognitive control mechanisms (Donoghue et  al., \n2020; Gyurkovics et al., 2022; Voytek et al., 2015; Waschke et al., 2021; see also Gratton \net al., 2018). However, how aperiodic parameters evolve over time to support information \nprocessing remains unclear. In this study, we use time -resolved analyses to investigate \nthe temporal evolution of aperiodic EEG during congruency and congruency sequence \neffect — two canonical markers of cognitive  control (Gratton et al., 2018; von Bastian, \n2020) — offering a novel perspective on the neurophysiological mechanisms underlying \nthese phenomena.  \n1.1 Cognitive Control . Cognitive control enables the adaptive regulation of \nbehavior in response to changing environmental demands (Gratton et al., 2018; von \nBastian, 2020). A key behavioral marker of cognitive control is the Congruency Effect \n(CE)—a robust phenomenon in confl ict tasks whereby performance systematically \ndepends on the congruency between task -relevant and task -irrelevant information \n(Eriksen & Eriksen, 1974; Gratton et al., 1992). Incongruent trials, which require resolving \ncompeting stimulu s–response associations, typically elicit slower responses and more \nerrors than congruent trials. The CE emerges from the interplay of distinct processes \nengaged during task performance, including perceptual, attentional, and executive \nmechanisms. Nevertheless, its most prominent interpretation is as an index of cognitive \ncontrol: the interference captured by the CE represents the very conditions in which \ncognitive control must be engaged, i.e., a situation where task -irrelevant information \nimpedes goal -directedness. Consequently, greater interference necessitates stronger \ntop-down regulation, which is why the CE is widely regarded as a central marker of \ncognitive control (Eriksen & Eriksen, 1974; Gratton et al., 1992; von Bastian, 2020).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n4 \n \nBeyond this immediate performance cost, cognitive control also shows \nadjustments based on trial history. This is reflected in the Congruency Sequence Effect \n(CSE) (Gratton et al., 1992; Egner, 2007) —a reduction of the CE following incongruent \nrelative to c ongruent trials, typically interpreted as a transient upregulation of cognitive \ncontrol (Botvinick et al., 2001). The CSE has been explained by several theoretical \naccounts, with some proposing the engagement of cognitive control and others \nsuggesting it operates without direct control involvement. The most prominent cognitive \ncontrol account, the conflict-monitoring theory (Botvinick et al., 2001), posits a dedicated \nconflict-monitoring system that detects the simultaneous activation of competing \nresponse tendencies, as occurs in incongruent trials. The resulting conflict signal triggers \nan upregulation of cognitive control, which enhances performance on subsequent \nincongruent trials, reducing the CE. At the neural level, activity in the conflict -monitoring \nunit has been linked to the anterior cingulate cortex (ACC), while the upregulation of \ncontrol is thought to rely on the dorsolateral prefrontal cortex (DLPFC; Botvinick et al., \n2001). Alternative non -control related accounts emphasize the involvement of learning \nand memory processes rather than cognitive control. For instance, episodic retrieval , \nfeature integration, and contingency learning accounts propose that trial-to-trial changes \nmay be driven by automatic processes, such as retrieval of stimulus -response bindings, \nstimulus repetitions, or learned associations between stimuli and responses (for reviews, \nsee Braem et al., 2019). Importantly, research shows that the CSE can still occur in the \nabsence of these factors (Jiménez & Méndez, 2014; Kim & Cho,  2014; Schmidt & \nWeissman, 2014; Weissman et al., 2014; Gyurkovics, Stafford, & Levita, 2020; \nGyurkovics, Kovacs, et al., 2020; Gyurkovics & Levita, 2021). Overall, evidence indicates \nthat the CSE likely reflects a combination of cognitive control and non -control \nmechanisms, making it a well -established albeit impure indicator of control adjustments \n(Abrahamse et al., 2016; Egner, 2023). \n  Accordingly, both the CE and CSE have been extensively investigated as \nmarkers of cognitive control in behavioral and neurophysiological research (e.g., Botvinick \net al., 2001; Cavanagh & Frank, 2014; Cohen & Donner, 2013; De Cesarei et al., 2023; \nGratton et al., 1992; Gyurkovics & Levita, 2021; Schiltenwolf et al., 2024; Tronelli et al., \n2025; van Maanen et al., 2009). To date, most of the EEG evidence about the neural \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n5 \n \nunderpinnings of CE and CSE has come from event -related potential (ERP) and time -\nfrequency research, which primarily emphasize phase -locked or rhythmic brain activity \nassociated with conflict resolution and cognitive adjustments, respectively. More recently, \ngrowing interest has emerged in the non -rhythmic, non-phase-locked component of the \nEEG signal —aperiodic activity —as a complementary source of information about \ncognitive function (Gao et al., 2017).  \n1.2 1/ f EEG activity . Aperiodic neural activity, also called 1/ f noise, is \ncharacterized in the frequency domain by a gradual decline in power with increasing \nfrequency. This component is best quantified in log -log space, where its two key \nparameters—exponent and offset—can be extracted from the slope and intercept of the \npower spectrum after separating out periodic (oscillatory) activity. Specifically, the \nrelationship can be described by Eq. (1) below: \n𝑙𝑜𝑔(1/𝑓𝑥)  =  −𝑥 · 𝑙𝑜𝑔(𝑓) (1) \nsuch that a steeper (more negative) slope, corresponding to a higher (more positive) \nexponent (x in the formula), reflects relatively greater power at lower frequencies (f in the \nformula). Conversely, a flatter (less negative) slope, corresponding to a lower (less \npositive) exponent, indicates relatively greater power at higher frequencies.  \nAt the level of neural mechanisms, aperiodic activity has been linked to the balance \nbetween excitatory and inhibitory inputs in neural circuits (called E:I balance, Gao et al., \n2017). Within this framework, steeper spectral slopes are thought to reflect i ncreased \ninhibitory activity and flatter spectral slopes indicate a relative increase in excitatory \nactivity, suggesting that properties of aperiodic activity may provide insight into \nfundamental aspects of brain functioning (Ahmad et al., 2022; Donoghue e t al., 2020; \nGao et al., 2017; Gyurkovics et  al., 2022; Voytek et  al., 2015; Waschke et  al., 2021). \nRelatedly, the spectral slope is also theorized to serve as an index of the temporal \ncoordination of neural firing: steeper slopes suggest more coordinated,  less noisy \nsignaling, while flatter slopes are associated with noisier neural activity (Chini et al., 2022; \nHe, 2014; He et  al., 2019; Voytek & Knight, 2015). These interpretations align with \nneurophysiological models of cognitive control, which propose that low-frequency power \nreflects feedback loops involving inhibitory connections, while high -frequency power \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n6 \n \nreflects feedforward, excitatory activity, supporting both auxiliary and task -relevant \nprocesses, not necessarily limited to cognitive control (Gratton, 2018). Within this \nframework, stimuli, task demands, or contextual variability can transiently recruit inhibitory \nmechanisms that support cognitive performance, producing steeper spectral slopes. \nGyurkovics et al. (2022) were among the first to demonstrate systematic stimulus-\ninduced changes in the aperiodic EEG. They reported more negative post -stimulus \nspectral slopes (compared to the pre -stimulus interval) that were independent of \nconcurrent ERPs and scaled with the cognitive demands of an auditory task. This pattern \nhas been interpreted as a stimulus-driven shift in E:I balance toward increased inhibition, \npotentially reflecting a transient suppression of ongoing excitatory activity. Such a \nmechanism may facilitate the reallocation of neural resources and the flexible updating of \nmental representations in response to task -relevant stimuli, which is a function of \ncognitive control.  \nSubsequent studies have provided further evidence that aperiodic EEG can be \nmodulated by experimental manipulations (e.g., Akbarian et al., 2024; Frelih et al., 2024; \nJia et al., 2024; Kałamała et al. 2024; Lu et al., 2024; Manyukhina et al., 2024; Yan et al., \n2024; Zhang et al., 2023). For example, Kałamała et al. (2024) provided the first evidence \nof within-trial modulations in aperiodic activity. By segmenting each trial into 500 -ms \nintervals, the authors showed that the presentation of a stimulus (a cue in a cued flanker \ntask) initially produced a more negative spectral slope compared to the pre -cue period \n(as found in Gyurkovics et al., 2022), which progressively transitioned into a more positive \nslope as time passed. They interpreted this pattern as re flecting an initial increase in \ninhibition, likely serving to suppress ongoing neural activity and facilitate the processing \nof task-relevant information, followed by a rise in excitation in support of motor response \npreparation for the upcoming imperative stimulus. These mechanisms are all associated \nwith cognitive control, which fundamentally aims to ensure goal -directedness in the face \nof distractions (Gratton et al., 2018; Von Bastian et al., 2020). Relatedly, Jia et al. (2024) \nshowed that the aperiodic component of EEG reflects changes in the CE across adjacent \ntrials in a flanker task. Specifically, spectral slopes were more negative in the incongruent \ncondition compared to the congruent condition during the current trial, but this pattern \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n7 \n \nreversed in the subsequent trial, with more negative slopes observed in the congruent \ncondition than in the incongruent condition. This reversal suggests that the post-stimulus \nspectral slope is not simply a continuation of pre-stimulus activity but instead indicates a \nshift in neural dynamics across trials.  \nWhile existing studies offer promising evidence that stimulus -induced changes in \nthe aperiodic component of the EEG may reflect cognitive control mechanisms, their \nconclusions are limited by several methodological shortcomings. Some of them rely on \nsimple paradigms with limited performance measures (e.g., Gyurkovics et al., 2022), \nmaking it difficult to assess the behavioral relevance of the observed spectral slope \nchanges. Moreover, most studies do not account for ERPs (e.g., Jia et al., 2024; Zhang \net al., 2023), which, as transient, non -oscillatory patterns of activity, can confound the \nestimation of aperiodic parameters (for a detailed argument, see Gyurkovics et al., 2022). \nCritically, these analyses often rely on averaging neural responses over broad, fixed time \nwindows (typically longer than 500 ms), which may obscure the more rapid temporal \ndynamics of aperiodic activity. This last limitation is particularly important because \ncognitive control is a transient, adaptive process that unfolds rapidly foll owing conflict-\nrelated information (Gratton et al., 1992, 2018; Botvinick et al., 2001). Capturing moment-\nto-moment fluctuations in aperiodic activity is therefore essential for understanding how \ncognitive control is dynamically deployed and adjusted over time. While Kałamała et al. \n(2024) challenged the notion that the aperiodic signal remains stable throughout a task \ntrial, no prior study has tracked time-resolved changes in aperiodic activity in response to \ncanonical manipulations of cognitive control, such as the CE and CSE (but see Frelih et \nal., 2024, for a recent time-resolved analysis in an n-back task). \n  1.3 The current study. In this study, we investigated the aperiodic characteristics \nof post-stimulus EEG related to the CE and CSE using a time -resolved approach. To \nachieve this, we analyzed scalp EEG from young adults performing a Picture -Word \nInterference (PWI) task (Rosinsk i, 1977; Tronelli et al., 2025), in which participants are \nasked to categorize pictures as either animals or vehicles while ignoring a superimposed \nword. The word can either be congruent (e.g., a picture of an animal with the word \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n8 \n \n\"animal\") or incongruent (e.g., a picture of an animal with the word \"vehicle\"), manipulating \nconflict level and cognitive control engagement.  \nTo capture the temporal dynamics of aperiodic activity, stimulus -locked epochs \nunderwent time -frequency decomposition using wavelet convolution, and the spectral \nslopes were estimated from the resulting power spectral densities time point by time point \nusing a censored regression-based approach (Kałamała et al., 2025). We expected that \nboth the CE and CSE would be reflected in changes in aperiodic EEG activity. Building \non theories that attribute CE and CSE to the engagement of cognitive control (Botvinick \net al., 2001; Gratton et al., 1992), as well as on research linking steeper slopes to higher \ncognitive demands (Gyurkovics et al., 2022; Kałamała et al., 2024; Lu et al., 2024), \npotentially through increased cortical inhibition, we hypothesized that incongruent trials \nwould be associated with more negative spectral slopes after stimulus presentation \ncompared to congruent trials. Furthermore, we exp ected that the preceding trial would \nmodulate the E:I ratio, as captured by the spectral steepness of the aperi odic activity in \nthe current trial. Specifically, we predicted that the difference in spectral slope steepness \nbetween current incongruent and congruent trials would be greater following congruent \ntrials than following incongruent trials, mirroring the behavioral CSE pattern. \nBy tracking the moment -to-moment evolution of the spectral slope, we aimed to \nreveal how E:I balance shifts over time, translating into conflict resolution and cognitive \ncontrol adjustments. This approach offers a novel perspective on the neural mechanisms \nunderlying CE and CSE, potentially uncovering transient shifts in cortical E:I balance that \ncontribute to flexible cognitive control.  \n2. Method \n2.1 Participants \nThe study was conducted at the University of Bologna, Italy. Fifty Italian-speaking \nparticipants with normal or corrected vision took part in the experiment. Data from 49 \nparticipants (20 females, mean age ± SD = 26.27 ± 3.38) were analyzed due to one \nparticipant failing to comply with task instructions. The experimental protocol adhered to \nthe principles outlined in the Declaration of Helsinki and received approval from the \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n9 \n \nEthical Committee of the University of Bologna. Written informed consent was obtained \nfrom all participants. \n2.2 Stimuli \nThe Picture Word Interference (PWI) task involved a set of 488 different pictures \nsourced from the internet, which were employed in previous studies investigating visual \ncategorization and cognitive control (De Cesarei et al., 2019, 2021, 2023; Tronelli et  al., \n2025). Each picture depicted either one or two animals or vehicles in an indoor or outdoor \nsetting, resulting in eight equally probable combinations across three orthogonal \ndimensions: content (animal vs. vehicle), number of foreground elements (one vs. two), \nand scenario (indoors vs. outdoors). To ensure consistency, all pictures met the following \ncriteria: only one or two clearly visible animals or vehicles in the foreground, no additional \nanimals or vehicles in the background, and a clearly distinguishable setting. \nThe pictures were presented in two blocks: one block in which pictures did not \nrepeat across trials (i.e., Novel Picture Block) and the other block in which the same \npictures were repeatedly shown across trials (i.e., Frequent Picture Block). In the Novel \nPicture Block et, 480 pictures out of the total 488 were used. In the Frequent Picture \nBlock, the remaining eight pictures were used to create four pairs of pictures, each \nconsisting of an animal and a vehicle, and during the presentation of Frequent Pictu re \nBlock participants saw one of the four repeated picture pairs. Hence, participants viewed \n480 pictures, all different in the Novel Picture Block, and a pair of pictures out of the four \nrepeated 240 times in the Frequent Picture Block. In the Frequent Pi cture Block, the \npictures were repeated across trials so that response repetition to the picture category \n(i.e., animal or vehicle) from one trial to another also implied the repetition of the same \npicture. In the Novel Picture Block, all the pictures were  different, so the repetition of the \nresponse to the picture category only implied the repetition of the same response \ncategory (animal or vehicle), as shown in Figure 1. \nThe pictures were displayed in color, with brightness and contrast adjusted to \nensure consistency. The average pixel intensity was 153 ( SD = 5.05) on a scale from 0 \nto 255. Each full-screen picture was resized to a 1280 × 1024 pixel monitor, subtending \n20°30' horizontal × 16°20' vertical degrees of visual angle. Each picture was shown \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n10 \n \ntogether with a word that was congruent or incongruent with the content of the picture. \nThe words were in Italian: \"veicolo\" (\"vehicle\") and \"animale\" (\"animal\") (e.g., a picture \nthat represents a cat with the word \"animale\" - i.e., \"animal\" - is a congruent trial, while \nthe same picture with the word \"veicolo\" – i.e., \"vehicle\" - is an incongruent trial). The \nwhite words in Courier New font (size 70) were centered on the screen within a black \nrectangle (15°8' horizontally × 2°33' vertically) superimposed the picture.  \n \nFigure 1. Overview of experimental design.  Examples of two consecutive trials in the changed response category \ncondition (Panel A) and in the repeated response category condition (Panel B), shown separately for the Novel Picture \nCondition (left panels) and the Frequent Picture Condition (right panels).  The distractor word could be congruent with \nthe picture (e.g., the word animale — animal in English — on an image of an animal) or incongruent (e.g., the word \nveicolo — vehicle in English — on an image of an animal). \n2.3 Procedure \nDuring the PWI task, participants were seated in the experimental room, where the \nillumination was set to 3 lux, as measured with a diode -type digital luxmeter. The \nexperiment comprised two blocks of 480 trials: the Novel Picture Block (480 trials, each \nfeaturing a unique picture) and the Frequent Picture Block (480 trials, in which two \npictures were shown 240 times each), totalling 960 trials. Participants were instructed to \nrespond to the picture while ignoring the word, giving equal importance to response speed \nand accuracy. They pressed one of two keys (J or N) on the computer keyboard using \ntwo fingers of their dominant hand. The order of task blocks and response keys were \ncounterbalanced across participants. The experiment lasted approximately 60 min, with \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n11 \n \neight practice trials preceding each experimental block to familiarize participants with the \ntask. The distance between the monitor and the participant was 94 cm. \nEach trial began with a 500 -ms picture presentation, followed by a response -\nstimulus interval (RSI) of either 1000, 2000, 3000, or 5000 ms. Out of the 960 trials, half \nwere congruent and half incongruent. The sequence of trials was designed such that each \nof the four possible sequences (e.g., previous trial congruent or incongruent, followed by \na congruent or incongruent trial) occurred with equal probability (120 trials per sequence \nper block). Additionally, each sequence was equally paired with each RSI c ondition, \nresulting in 30 trials per condition in each block. In summary, the PWI task comprised five \nexperimental manipulations: Stimulus Repetition (Novel Picture Block vs. Frequent \nPicture Block), Category-Response Repetition (Same Category vs. Changed Category), \nRSI (1000, 2000, 3000, or 5000 ms), Previous Congruency (Previous -Congruent vs. \nPrevious-Incongruent), and Current Congruency (Congruent vs. Incongruent). Since the \npresent study focuses specifically on the well -established CE and CSE, we limite d our \nanalysis to these two manipulations. All other conditions were collapsed, as they fall \noutside the scope of the current research.  The experiment was run using E -Prime 2.0 \nProfessional. \n2.4 Behavioral Analysis \nThe dependent variables were mean accuracy and mean reaction time (RT). Practice \ntrials and the first trial of each block were excluded. For the RT analysis, we additionally \nexcluded trials with incorrect responses, trials following an error, and trials wi th RTs \nexceeding ±2.5 standard deviations ( SDs) from the participant's mean. Both accuracy \nand RT data were analyzed with a repeated-measures ANOVA, with the following within-\nsubject factors: Previous Congruency (two levels: Previous -Congruent, Previous -\nIncongruent) and Current Congruency (two levels: Congruent, Incongruent). Huynh–Feldt \ncorrection for lack of sphericity was used when appropriate. Follow-up paired t-tests were \nperformed in the case of significant interactions. A p-value < .05 was considered the \nthreshold for statistical significance. For all tests, partial eta squared (ηp²) and Cohen’s d \nwere calculated and reported as measures of effect size for ANOVAs and t -tests, \nrespectively.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n12 \n \n2.5 EEG Data Acquisition and Preprocessing \nScalp EEG was recorded from 64 Ag/AgCl active electrodes using a BioSemi \nActiveTwo® system (Amsterdam, The Netherlands). The electrodes were secured by an \nelastic cap according to the extended 10 -20 international electrode placement system \n(Jasper, 1958). Horizontal and vertical electrooculograms (EOGs) were also recorded to \nmonitor ocul ar artifacts. The sampling rate was 512 Hz. During recording, data were \nreferenced to the common mode sense (CMS) electrode and were filtered online with a \nlow pass filter equal to 1/5 of the sampling rate (i.e., 102.4 Hz). \nThe data were pre -processed using custom MATLAB 2022b codes (The \nMathWorks) incorporating EEGLAB 13.6.5 (Delorme & Makeig, 2004) and ERPlab 6.1.3 \n(Lopez-Calderon & Luck, 2014). The EEG was first re-referenced to the average mastoids \nand bandpass filtered w ith 0.1 and 40 Hz cut -off frequencies. The data were then \nsegmented into 2000-ms epochs, spanning from 500 ms before to 1500 ms after stimulus \nonset. After excluding epochs with amplifier saturation and performing ocular correction \n(Gratton et al., 1983), epochs with peak-to-peak voltage fluctuations at any EEG channel \nexceeding 150 μV (600 -ms window width, 100 -ms window step) were discarded. \nAdditional epochs were excluded if they contained an incorrect response, were followed \nan error, were the first tria l of a block, or had a RT exceeding ±2.5 SDs from the \nparticipant's mean to match the RT analysis criteria. Data from electrodes Fp1, Fp2, AF3, \nAF4, AF7, AF8, AFz, and FPz were excluded as they often contain minor residual ocular \nartifacts even after ocula r correction. This left 56 electrodes for further analysis. The \naverage number of artifact -free epochs per participant was 694 ( SD = 153, min = 366, \nmax = 893). \n2.6 Spectral Analysis \nTo examine the temporal variation of the aperiodic EEG component, stimulus -\nlocked epochs underwent wavelet decomposition, and the spectral slopes were estimated \nfrom the resulting power spectral densities at each time point using a censored \nregression-based approach, as described below (see also Kałamała et al., 2025). For a \ndescription of the spectral analysis workflow see Fig. 2. Data and code necessary to \nreproduce the statistical analyses (section 3.2.3) will be available at https://osf.io/z3qsg . \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n13 \n \n \nFigure 2. Flowchart of the spectral analysis procedure. The diagram shows the main processing steps. \n \nFirst, we removed the average event-related potentials (ERPs) from the EEG data \nto avoid distortions in aperiodic parameter estimation. Since ERPs contribute to the \noverall EEG spectrum in a broadband fashion, removing them allows for a more accurate \nisolation of the aperiodic signal (for further discussion, see Gyurkovics et al., 2022). This \nwas done by computing the average ERP for each Participant × Condition combination \nand subtracting it from the corresponding epochs, following established procedures \n(Gyurkovics et al., 2022; Kałamała et al., 2024). \nNext, the single epochs were convolved with sliding cosinusoidal and sinusoidal \nwavelet pairs, which were combined to provide \"instantaneous\" power estimates for each \nfrequency. This new approach provides a sufficient temporal resolution to follow unfolding \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n14 \n \ntask dynamics as they occur: one -cycle wavelets were used to maximize temporal \nresolution. Wavelets were not weighted with Gaussian tapers and simply constituted a \nsingle cycle of a sine or a cosine wave at a given frequency. The analysis focused on a \nlimited set of frequencies to optimize computational time and minimize redundancy due \nto overlap between nearby frequencies.  Because the emphasis here was on temporal \nresolution, this necessarily limited the frequency resolution of the analyses. This was \nconsidered acceptable since the focus is on broadband rather than narrowband activity. \nSpecifically, we selected values corresponding to successive powers of 2 within the \navailable frequency range —i.e., the 2nd, 4th, 8th, 16th, and 32nd frequencies from the \nset—resulting in the following values: 2.5, 5, 7.5, 15, and 25 Hz. Restricting the frequency \nrange facilitates more accurate estimation of the spectral slope by excluding frequencies \nmost likely to reflect periodic activity. Prior research has shown that most scalp-recorded \noscillatory activity during wakefulness falls within the theta-alpha-beta range (e.g., Myrov \net al., 2024). Accordingly, power values within the 7.5 –15 Hz range were excluded from \nslope and intercept estimation. This procedure yields estim ates that more accurately \ncapture the aperiodic component.  \nThe resulting power estimates and corresponding frequencies were log-transformed, \nand regression slopes were calculated for each time point using a least-squares approach \nto provide instantaneous estimates of the aperiodic component. To temporally smooth \nthe data, slope estimates were filtered using a 5-point moving average within each epoch. \nEpochs with slope values exceeding ±4 were considered outliers and excluded. On \naverage, 659 epochs per participant were retained ( SD = 168, min = 274, max = 891). \nThe remaining slopes were then averaged across epochs for each EEG electrode (56 \nelectrodes in total) and experimental condition (Stimulus Repetition × Category Repetition \n× RSI × Previous Congruency × Current Congruency) within each participant. \nParametrization and data reduction. To reduce data dimensionality, improve the \nsignal-to-noise ratio, and increase statistical power, the Participant × EEG Electrode × \nCondition × Time data were subjected to a Principal Component Analysis (PCA). To \nminimize the edge artifacts introduced by t he wavelet decomposition, the first and last \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n15 \n \n200 ms of each epoch were excluded, leaving a -300 to 1300 ms window around the \nstimulus for PCA analysis. \nThe Empirical Kaiser Criterion (Braeken & Van Assen, 2017; Steiner & Grieder, \n2020) was applied to determine the number of principal components to retain. The factor \nloadings were rotated using varimax rotation (Kaiser, 1958), which optimizes the \nalignment of factors (components) with physiologically meaningful sources by \nemphasizing high-contrast loadings. Finally, any component explaining less than 1/ x of \nthe total variance (where x is the total number of components) was excluded from further \nanalysis. \nFirst, the PCA was performed on the EEG electrode dimension (referred to as spatial \nPCA). The EEG data were then projected onto the spatial loadings using the regression \nmethod (i.e., EEG data × unstandardized factor loadings × inverse covariance matrix; \nThomson, 1938; Thurstone, 1935) to obtain spatial (factor) scores (Participant × Spatial \nComponent × Condition × Time). \nTo establish the optimal temporal resolution for statistical analyses, a PCA was also \nconducted on the time dimension (temporal PCA). Since the PCA loadings plotted as a \nfunction of time showed a Gaussian -shaped pattern (see Figure 5A), the Full Width at \nHalf Maximum (FWHM) was applied to estimate the average duration of temporal \ncomponents. This FWHM value was then used to segment the spatial PCA epochs (-300 \nto 1300 ms) into consecutive time windows anchored at the stimulus onset (0 ms). \nFinally, the reduced data (Participant × Spatial Component × Condition × Time \nWindow) were subjected to statistical testing. Analyses were conducted separately for \neach spatial component on the corresponding spatial scores (derived from spatial PCA), \naveraged within each FWHM-long time window (as determined by temporal PCA). Given \nthat examining the effects of experimental manipulation on aperiodic activity in a time -\nresolved manner is a novel approach in EEG research, we first assessed the global \nstimulus-induced changes, independent of experimental condition. This was done by \ncomparing condition-averaged values in each post -stimulus PCA-based time window to \nthe pre -stimulus window using paired t-tests. The PCA, as well as the spectral \ndecomposition, were performed across all available conditions and then averaged, so that \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n16 \n \nonly Previous and Current Congruency were included in the analyses. Subsequently, we \nconducted a repeated -measures ANOVA with Previous Congruency and Current \nCongruency as within -subject factors to evaluate the expected experimental effects, \nspecifically the congruency and congruency sequence effects. Follow -up paired t-tests \nwere performed in the case of significant interactions. Baseline correction was not \napplied, as no significant effects were observed in the pre -stimulus time window. A p-\nvalue < .05 was considered the threshold for statistical significance. For all tests, partial \neta squared (ηp²) and Cohen’s d were calculated and reported as measures of effect size \nfor ANOVAs and t-tests, respectively. \n3 Results \n3. 1 Behavioral Data \n3.1.1 Accuracy \nThe main effect of Current Congruency was significant, F(1,48) = 30.24, p < .001, \nηp 2 = .39, with lower accuracy for incongruent (M = 96.74%, SD = 3.15) compared to the \ncongruent trials (M = 97.53%, SD = 2.68), indicating the CE. The main effect of Previous \nCongruency was also significant, F(1,48) =6.80, p = .012, ηp 2 = .12, with higher accuracy \nin trials following incongruent trials (M = 97.33%, SD = 2.73) compared to trials following \ncongruent trials (M = 96.93%, SD = 3.14).  \nFinally, the two -way interaction between Current and Previous Congruency was \nsignificant, F(1, 48) = 12.02, p = .001, ηp² = .20. There was no significant difference in \naccuracy between congruent trials preceded by congruent trials (M = 97.63%, SD = 2.64) \nand those preceded by incongruent trials (M = 97.43%, SD = 2.94), t(48) = 0.93, p = .357, \nd = .13. In contrast, the accuracy on incongruent trials was significantly higher when they \nwere preceded by incongruent trials ( M = 97.24%, SD = 2.67) compared to when they \nwere preceded by congruent trials ( M = 96.23%, SD = 3.79), t(48) = 4.12, p < .001, d = \n.59. When trials were preceded by a congruent trial, the accuracy was lower in \nincongruent trials compared to congruent trials, t(48) = -5.37, p < .001, d = -.77. There \nwas no significant difference between incongruent trials and congruent trials when they \nwere preceded by an incongruent trial, t(48) = -.98, p = .332, d = -.14 (indicating the CSE). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n17 \n \n3.1.2 Reaction Times \nThe main effect of Current Congruency was significant, F(1,48) = 16.68, p < .001, \nηp 2 = .26, with slower responses for incongruent (M = 656.39 ms, SD = 147.35) compared \nto the congruent trials (M = 641.84 ms, SD = 132.27), indicating the CE. The main effect \nof Previous Congruency was not significant, F(1,48) =.12, p = .732, ηp 2 <.001. \nThe two -way interaction between Current and Previous Congruency was \nsignificant, F(1,48) = 18.33, p < .001, ηp 2 = .28 (see Figure 3). Responses to incongruent \ntrials were slower when they were preceded by a congruent compared with an \nincongruent trial, and responses to congruent trials were slower when they were preceded \nby an incongruent compared with a congruent trial, t(48) = 3.52, p =.001, d = .50  and \nt(48) = 2.87, p =.006, d = .41; respectively. There was no significant difference between \nincongruent and congruent trials when they were preceded by incongruent trials, t(48) = \n1.44, p =.158, d = .21. Responses were slower during incongruent trials than during \ncongruent trials when they were preceded by congruent trials, t(48) = 5.18, p <.001, d = \n.74 indicating the CSE. \n \nFigure 3. Current Congruency as a function of Previous Congruency for Reaction Times.  Mean reaction times for \ncurrent congruency, broken down by previous congruency. Error bars represent ±1 within -subject standard errors of \nthe mean (Cousineau, 2005). \n3.2 EEG Data \n3.2.1 Spatial PCA  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n18 \n \nThe Empirical Kaiser Criterion suggested a five -component solution. Figure 4A \nillustrates the distribution of loadings across electrodes, with each component showing a \nprominent peak in different scalp regions. Components were named after the region \nwhere they showed the maximum activity: occipital, frontal, central, left temporal,  and \nright temporal. Temporal location components were excluded from further analysis as \nthey accounted for less than one -fifth of the total variance (see Figure 4B). As a result, \nstatistical analyses focused on the occipital, frontal, and central components. \n \nFigure 4. Results of the Spatial Principal Component Analysis. Distribution of electrode -wise standardized loadings \nafter varimax rotation (Panel A), and percentage of variance explained by each corresponding component (Panel B). \nComponents are ordered from highest to lowest explained variance. Those outlined in green in Panel B accounted for \nmore than one-fifth of the variance and were retained for further analysis.  \n \n3.2.2 Temporal PCA  \nThe Empirical Kaiser Criterion indicated a 13 -component solution. Figure 5A shows the \ndistribution of loadings across time points, with each component showing a prominent \npeak in a different period. Five components were excluded as they accounted for less \nthan 1/13 of the total variance (see Figure 5B). The mean FWHM across the remaining \neight components was 162 ms ( Mdn = 161 ms, SD = 25 ms). Based on this, a 160 -ms \nperiod was selected as the effective time window for statistical analysis. Accordingly, the \nfollowing time windows were defined: [ -160,0], (0,160], (160,320], (320,480], (480,640], \n(640,800], (800,960], (960,1120], and (1 120,1280], each containing 8 to 9 data points. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n19 \n \nSquare brackets indicate the inclusion of the boundary value, while parentheses indicate \nexclusion. Time points at the edges of the epochs, where a full 160-ms time window could \nnot be created, were excluded. For statistical testing, values within each tim e window \nwere averaged separately for each spatial PCA component. \n \nFigure 5. Results of the Temporal Principal Component Analysis. Distribution of time-wise standardized loadings after \nvarimax rotation (Panel A) and percentage of variance explained by each component (Panel B). Components in Panel \nB are ordered from highest to lowest explained variance. Those outlined in green accounted for more than 1/13 of the \nvariance and were retained for further analysis. FWHM, Full Width at Half Maximum. \n3.2.3 Spectral Slope Analysis \nPaired t-tests were conducted to assess global (condition -average) changes \nbetween the pre -stimulus and post -stimulus periods during the trial, while repeated -\nmeasures ANOVA was used to examine the CE and CSE. These analyses were \nperformed on the spatial scores for the three spatial components—occipital, frontal, and \ncentral (see Spatial PCA section)—with their values averaged across 9 time windows \n(see Temporal PCA section). \nFigure 6 illustrates the time course of the spectral slope before and after PCA \ndecomposition. As expected, the spectral slope exhibits a widespread negativity, with all \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n20 \n \nvalues remaining below zero. Subsequent analysis of global (condition -averaged) \nchanges induced by the stimulus (see Figure 7; for statistics, see Table 1) revealed that, \ncompared to the pre -stimulus time window, the stimulus induced an additional negative \nshift throughout the entire post -stimulus period for the central component and during all \npost-stimulus time windows except the last for the frontal and occipital components. \n \n \nFigure 6. Changes in Aperiodic Activity (Spectral Slope) as a Function of Time. Stimulus -locked spectral slopes for \nCurrent Congruency (blue line for congruent, red line for incongruent) by Previous Congruency (solid line for previous \ncongruent, dashed line for pre vious incongruent), shown for raw data from midline electrodes (Panel A) and for key \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n21 \n \ncomponents extracted from spatial PCA (Panel B). The dashed vertical line indicates stimulus onset (time zero). Panel \nA displays slope values from selected electrodes for illustrative purposes, whereas Panel B presents factor scores \nestimated from all electrodes using PCA weights; therefore, the panels are not directly comparable. \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n22 \n \nFigure 7. Global Stimulus-Induced Changes in Spectral Slope. The time course of condition -averaged spatial scores \nfor the frontal (top), central (middle), and occipital (bottom) components derived from spatial PCA. The dashed vertical \nline indicates stimulus onset (time zero). Shaded areas represent the time windows defined by temporal PCA. Lines at \nthe bottom of each subplot denote statistically significant effects from paired t-tests comparing post-stimulus values to \nthe pre-stimulus interval (values with in the time windows were averaged for t -tests; black horizontal line for p < .05, \ngreen horizontal line for p < .01). \n \nTable 1. Results of t-Tests Assessing Global Changes in Spectral Slope Across Time Windows and Scalp Regions. \n  FRONTA\nL \n   CENTRAL    OCCIPITA\nL \n \nTime \nWindow (ms)                            \nt p d  t p d  t p d \n(0,160] -4.23 \n \n< 0.001 \n \n-0.60 \n \n 3.67 \n \n< 0.001 \n \n0.52 \n \n -6.88 \n \n< 0.001 \n \n-0.98 \n \n(160,320] -7.44 \n \n< 0.001 -1.06  -4.60 \n \n< 0.001 \n \n-0.66 \n \n -14.82 \n \n< 0.001 \n \n-2.12 \n \n(320,480] -6.99 \n \n< 0.001 \n \n-1.00 \n \n -9.08 \n \n< 0.001 \n \n-1.30 \n \n -15.23 \n \n< 0.001 \n \n-2.18 \n \n(480,640] -6.38 \n \n< 0.001 \n \n-0.91 \n \n -10.80 \n \n< 0.001 \n \n-1.54 \n \n -11.84 \n \n< 0.001 \n \n-1.69 \n \n(640,800] -7.65 \n \n< 0.001 \n \n-1.09 \n \n -8.61 \n \n< 0.001 \n \n-1.23 \n \n -9.57 \n \n< 0.001 \n \n-1.37 \n \n(800,960] -7.15 \n \n< 0.001 \n \n-1.02 \n \n -7.06 \n \n< 0.001 \n \n-1.01 \n \n -4.58 \n \n< 0.001 \n \n-0.65 \n \n(960,1120] -4.48 \n \n< 0.001 -0.64 \n \n -4.88 \n \n< 0.001 \n \n-0.70 \n \n -1.13 \n \n0.262 \n \n-0.162  \n \n(1120,1280] -1.58 \n \n0.121 \n \n-0.225 \n \n -2.97 \n \n0.005 \n \n-0.42 \n \n 1.07 \n \n0.288 \n \n0.154 \nNote. Results of paired t-tests comparing post-stimulus values to the pre-stimulus interval (i.e., [–160, 0] ms). For time \nwindows, square brackets indicate inclusion of the boundary value, while parentheses indicate exclusion. Degrees of \nfreedom for all tests are (48). Significant effects (p < .05) are bolded. d represents Cohen’s d effect size. \n \nA series of repeated -measures ANOVAs assessing the effects of experimental \nmanipulation revealed significant main effects of Current Congruency and Previous \nCongruency, as well as significant interactions between these two factors, in various time \nwindows and locations (for statistics, see Table 2). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n23 \n \nTable 2 \nResults of Repeated-Measures ANOVA Assessing the Effects of Experimental Manipulation on Spectral Slope across \nTime Windows and Scalp Regions.  \n   FRONTA\nL  \n   CENTRA\nL  \n   OCCIPITA\nL  \n \nEffects Time \nWindow \nF p ηp 2  F p ηp 2  F p ηp 2 \nCC (0,160] 1.20 0.278 0.0\n2 \n 0.8\n1 \n0.373 0.0\n2 \n 0.9\n3 \n0.339 0.0\n2 \n (160,320] 0.09 0.770 0.0\n0 \n 0.0\n3 \n0.858 0.0\n0 \n 0.0\n4 \n0.849 0.0\n0 \n (320,480] 2.79 0.101 0.0\n5 \n 2.9\n0 \n0.095 0.0\n6 \n 2.5\n7 \n0.115 0.0\n5 \n (480,640] 4.45 0.040 0.0\n8 \n 1.1\n5 \n0.288 0.0\n2 \n 4.5\n8 \n0.038 0.0\n9 \n (640,800] 4.70 0.035 0.0\n9 \n 2.1\n2 \n0.152 0.0\n4 \n 2.8\n5 \n0.098 0.0\n6 \n (800,960] 1.65 0.205 0.0\n3 \n 0.9\n5 \n0.336 0.0\n2 \n 7.2\n6 \n0.010 0.1\n3 \n (960,1120] 0.56 0.460 0.0\n1 \n 1.8\n1 \n0.185 0.0\n4 \n 0.3\n3 \n0.566 0.0\n1 \n (1120,1280] 8.07 0.007 0.1\n4 \n 9.7\n3 \n0.003 0.1\n7 \n 7.0\n2 \n0.011 0.1\n2 \n             \nPC  (0,160] 1.51 0.225 0.0\n3 \n 0.0\n1 \n0.920 0.0\n0 \n 0.6\n7 \n0.417 0.0\n1 \n (160,320] 1.25 0.270 0.0\n3 \n 1.0\n0 \n0.323 0.0\n2 \n 0.2\n8 \n0.602 0.0\n1 \n (320,480] 0.04 0.836 0.0\n0 \n 1.1\n9 \n0.280 0.0\n2 \n 4.6\n0 \n0.037 0.0\n9 \n (480,640] 0.38 0.538 0.0\n1 \n 0.3\n7 \n0.545 0.0\n1 \n 1.3\n8 \n0.246 0.0\n3 \n (640,800] 0.08 0.782 0.0\n0 \n 0.1\n5 \n0.696 0.0\n0 \n 0.1\n3 \n0.722 0.0\n0 \n (800,960] 0.86 0.358 0.0\n2 \n 0.6\n0 \n0.443 0.0\n1 \n 0.3\n6 \n0.552 0.0\n1 \n (960,1120] 0.25 0.617 0.0\n1 \n 1.3\n6 \n0.249 0.0\n3 \n 0.4\n0 \n0.532 0.0\n1 \n (1120,1280] 6.05 0.018 0.1\n1 \n 0.4\n4 \n0.512 0.0\n1 \n 0.3\n8 \n0.542 0.0\n1 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n24 \n \n             \nCC × \nPC \n(0,160] 0.06 0.807 0.0\n0 \n 0.0\n0 \n0.951 0.0\n0 \n 1.0\n4 \n0.312 0.0\n2 \n (160,320] 4.14 0.048 0.0\n8 \n 5.2\n1 \n0.027 0.1\n0 \n 0.2\n8 \n0.600 0.0\n1 \n (320,480] 0.26 0.611 0.0\n0 \n 4.0\n4 \n0.050 0.0\n8 \n 0.9\n4 \n0.337 0.0\n2 \n (480,640] 0.29 0.596 0.0\n1 \n 0.1\n5 \n0.700 0.0\n0 \n 7.4\n0 \n0.009 0.1\n3 \n (640,800] 1.54 0.221 0.0\n3 \n 0.9\n9 \n0.326 0.0\n2 \n 3.3\n5 \n0.073 0.0\n7 \n (800,960] 0.97 0.330 0.0\n2 \n 1.4\n1 \n0.241 0.0\n3 \n 0.5\n9 \n0.447 0.0\n1 \n (960,1120] 6.58 0.013 0.1\n2 \n 0.0\n0 \n0.995 0.0\n0 \n 0.3\n7 \n0.545 0.0\n1 \n (1120,1280] 1.82 0.184 0.0\n4 \n 3.4\n6 \n0.069 0.0\n7 \n 0.0\n3 \n0.862 0.0\n0 \nNote. Results of 2 (Current Congruency [CC]) × 2 (Previous Congruency [PC]) ANOVAs. For time windows, square \nbrackets indicate inclusion of the boundary value, while parentheses indicate exclusion. Degrees of freedom for all \neffects are (1, 48). Significant effects (p < .05) are bolded. \nSignificant effects of Current Congruency were observed in three time windows of \nthe frontal component (480–640 ms, 640–800 ms, and 1120–1280 ms), one time window \nof the central component (1120 –1280 ms), and three time windows of the occipital \ncomponent (480–640 ms, 800 –960 ms, and 1120 –1280 ms) (see Figure 8A). Overall, \nslope values were more negative for the incongruent condition than for the congruent \ncondition. However, in the final time window of the occipital component (1120–1280 ms), \nthis pattern rev ersed, with more negative values observed in the congruent than the \nincongruent condition. \nOnly two significant effects of Previous Congruency were observed (see Figure \n8B). Slope values were significantly more negative for the previous congruent condition \nthan the previous incongruent condition in the 320 —480 ms window for the occipital \ncomponent and the 1120 —1280 ms window for the frontal component. No other \nstatistically significant effects of Previous Congruency were found. \nSignificant interactions between Current Congruency and Previous Congruency \nwere observed in several time windows across the three components (see Figure 8C and \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n25 \n \nFigure 9). When considering how previous congruency influenced the current trial, we \nobserved four statistically significant effects in total —three for the current congruent \ncondition and one for the current incongruent condition. In the frontal (160–320 ms, 960–\n1120 ms) and central (160 –320 ms) components, Previous Congruency influenced the \ncurrent congruent trials. Specifically, in the frontal component, slope values for congruent \ntrials were more negative when preceded by an incongruent trial compared to  when \npreceded by a congruent trial,  tfrontal:160–320ms(48) = 2.33,  p < .05, d = .34, and tfrontal:960–\n1120ms(48) = 2.17, p < .05, d = .31. In contrast, the central component showed the opposite \npattern: slope values for congruent trials were more negative when preceded by another \ncongruent trial than by an incongruent one, tcentral:160–320ms(48) = 2.28, p < .05, d = .33. No \nsignificant effects of Previous Congruency on incongruent trials were found in the frontal \nor central components ( ps > .05). However, in the occipital component (480 –640 ms), \nPrevious Congruency significantly influenced the current incongruent condition, \ntoccipital:480–640ms(48) = 3.13, p < .05, d = .45, with more negative slope values when the \nincongruent trial was preceded by a congruent one compared to an incongruent one. No \neffects were observed for the current congruent trials in the occipital component ( ps > \n.05). When analyzing the difference between current incongruent and congruent trials \nseparately for each type of preceding trial (i.e., preceded by congruent and preceded by \nincongruent), we found two significant effects. Specifically, when preceded by a \ncongruent trial, the slope was more negative for current incongruent than for current \ncongruent trials in the late frontal time window (960–1120 ms) and in the occipital window \n(480–640 ms), tfrontal:960–1120ms(48) = 2.18, p < .05, d = .32, and toccipital:480–640ms(48) = 3.57, \np < .001, d = .52. No other significant effects were observed (p’s > .05).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n26 \n \n \nFigure 8. Effects of Experimental Manipulation on the Spectral Slope. The time course of differences in spatial scores \nfor Current Congruency (Panel A), Previous Congruency (Panel B), and their interaction (Panel C) for the frontal (top), \ncentral (middle), and occipital (bottom) components derived from spatial PCA. The dashe d vertical line indicates \nstimulus onset (time zero). Shaded areas represent the time windows defined by temporal PCA. Lines at the bottom of \neach subplot denote statistically significant effects  from repeated-measures ANOVA (values within the time windows \nwere averaged for testing); black horizontal line for p < .05, green horizontal line for p < .01; Con, congruent; Inc, \nincongruent. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n27 \n \n \nFigure 9. Current Congruency as a Function of Previous Congruency for Spectral Slope. Mean spectral slope values \nfor Current Congruency as a function of Previous Congruency in the time windows and spatial components where \nsignificant interactions between these factors were observed: 160–320 ms and 960–1120 ms in the frontal component \n(top left and top right panels, respectively), 160–320 ms in the central component (bottom left panel), and 480–640 ms \nin the occipital component (bottom right panel).  \n \n \n \n4. Discussion \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n28 \n \nThis study investigated whether and how the aperiodic component of EEG \nactivity—specifically the time-resolved spectral slope—reflects the CE and CSE during a \nPWI task. Our findings offer some novel insights into the temporal dynamics and neural \nmechanisms of cognitive control, demonstrating that fluctuations in post -stimulus \naperiodic activity, likely reflecting changes in the E:I balance of underlying neural circuits, \nare linked to the engagement and adaptive modulation of cognitive control. \nAt the behavioral level, we observed the classic CE, characterized by slower and \nless accurate responses on incongruent trials compared to congruent ones. We also \nobserved the CSE (Gratton et al., 1992). Specifically, responses to incongruent trials were \nslower and less accurate when they followed congruent rather than incongruent trials, \nand likewise, responses to congruent trials were slower and less accurate when they \nfollowed incongruent rather than congruent trials. Notably, the CE was statistically \nsignificant only following congruent trials but not following incongruent ones, consistent \nwith the idea that experiencing conflict triggers temporary adjustments in control that \nreduce susceptibility to interference on subsequent trials (Botvinick et al., 2 001; Gratton \net al., 1992; Grant & Weissman, 2019; 2023; Weissman, 2019).  \nAt the EEG level, we quantified the aperiodic spectral slope in a time -resolved \nmanner using a wavelet approach to capture the dynamics of neural processing. This \nhigh-temporal-resolution method allowed us to track rapid fluctuations in aperiodic activity \nthat traditional time -averaged analyses might overlook. To reduce data dimensionality, \nwe then applied two PCAs: a spatial PCA and a temporal PCA. The spatial PCA yielded \nthree interpretable components corresponding to occipital, frontal, and central \ntopographies. The temporal PCA provided a resolution of ~160 ms, segmenting each \nepoch into eight non-overlapping time windows. As such, subsequent analyses focused \non the three spatial components (i.e., frontal, central and occipital) across these windows. \nVisual inspection of the spectral slope time course revealed a widespread negative \ndeflection, with slope values consistently remaining below zero across the scalp (Figure \n6). This is in line with the expectation that the power spectrum of neural data is negative- \ngoing under normal physiological circumstances (Brake et al., 2024; He, 2014; Gao et al., \n2017). Subsequent analysis of global (condition -averaged) changes induced by the \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n29 \n \nstimulus showed that the stimulus further increased the negativity of the slope relative to \nthe baseline (pre-stimulus) period, in line with findings of post-stimulus steepening of the \nspectrum by Gyurkovics et al. (2022) and Kalamala et al. (2024). Here, this effect—a \nnegative spectral shift —was particularly robust in the central component, where it \npersisted throughout the entire post -stimulus period, adding some spatial specificity to \nprevious reports. In contrast, in the frontal and occipital components , the effect was \npresent across most time windows, except the final ones. These findings suggest that the \nstimulus induced a shift in the aperiodic background activity, presumably reflecting a \nsustained state of increased cortical inhibition during task en gagement. Such an \ninterpretation could align with the well-established pattern of default mode network (DMN) \ndeactivation observed during externally oriented, goal -directed tasks (  Buckner et al., \n2008; Shulman et al.1997; Mazoyer et al. 2001). The DMN typically shows reduced activity \nwhen attention is directed toward goal -oriented behavior, reflecting the suppression of \ninternally directed cognition to support efficient cognitive control and performance \n(Raichle et al., 2015). Thus, the obser ved increase in cortical inhibition may represent a \nneurophysiological correlate of this large-scale network reconfiguration that accompanies \nengagement in externally focused cognitive demands. \nThe spectral slope was further modulated by current congruency ( Figure 8A). In \nline with our hypothesis, steeper slopes (more negative slopes) were observed during \nincongruent trials compared to congruent ones in several time windows. This effect was \nstatistically reliable in frontal time intervals: 480 –640 ms, 640 –800 ms, and 1120–1280 \nms after stimulus onset. A similar effect was observed in one, relatively late, central time \ninterval (1120–1280 ms), and in occipital time intervals: from 480 to 640 ms, from 800 to \n960 ms. Interestingly, in the final occipital time interval (1120–1280 ms), a different pattern \nwas observed: steeper slopes were observed in the congruent condition than in the \nincongruent condition.  \nAs described in the Introduction, spectral slopes are thought to index E:I balance \nwithin synaptic circuits (Ahmad et al., 2022; Gao et al., 2017). The presentation of a \nstimulus, which engages in a task, can lead to a temporary steepening of the spectral \nslope. Such changes are interpreted as reflecting the increased recruitment of inhibitory \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n30 \n \nmechanisms, which help regulate excessive neural activity and thereby support more \nefficient cognitive performance (Gratton, 2018). In line with this interpretation, the steeper \nslopes observed during incongruent compared to congruent trials may reflect a conflict-\nrelated neural response that filters out irrelevant signals to support efficient response \nselection. This may serve as an index of reactive control wherein neural activity is \nmodulated after conflict detection to support accurate performance. Indeed, congruency-\nrelated differences in slope were observed relatively late after stimulus onset (around 500 \nms), suggesting that control processes reflected in the aperiodic background were \nengaged after conflict detection. The increased cortical inhibition (and/or reduction in \ncortical excitability) could serve as a neural filtering mechanism, suppressing interference \nfrom task-irrelevant signals and enhancing the precision of task -relevant processing, as \nobserved by Gyurkovics et al. (2022) and Kałamała et al. (2024). This mechanism would \nsupport more efficient conflict resolution and facilitate more accurate response selection. \nThe proposed interpretation aligns with theoretical accounts positing that an increase in \nspectral steepness during incongruent tr ials may indicate a shift toward slower activity, \nwhich has been linked to greater engagement of control mechanisms (Gratton, 2018).  \nNotably, in the final occipital time interval (1120 –1280 ms), negative slopes were \nless pronounced during incongruent compared to congruent trials, opposite to the effects \nseen fronto-centrally. Although these results do not align with our hypothesis, we m ay \nspeculate that this pattern reflects the distinct functional roles of these cortical areas: \noccipital regions primarily supports sensory processing and feature analysis, which may \nbe enhanced during congruent trials, whereas frontal and central regions are more \ninvolved in top-down cognitive control and conflict resolution, which may be heightened \nduring incongruent trials. The presence of steeper slopes for the congruent condition in \nthe late occipital window suggests that sensory processing either re -emerges once \nconflict has been resolved or remains active throughout the trial but becomes detectable \nonly after resolution of the conflict. Taken together, these opposing patterns may reflect \na dynamic interplay between sensory -driven and control-related mechanisms, with their \nrelative dominance shifting as a function of task demands and cortical region. Future \nstudies are required to replicate this finding and tease apart these proposed mechanisms. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n31 \n \nDynamic, trial -to-trial changes in spectral slope steepness were observed over \nfrontal, central, and occipital components, suggesting a modulating effect of the previous \ncongruency on the current congruency (CSE; Figure 8C). In the frontal component (160–\n320 ms, 960 –1120 ms), congruent trials following incongruent trials showed more \nnegative slopes than those following congruent trials. In the central component (160–320 \nms), the opposite pattern emerged: more negative slopes were observed when congruent \ntrials followed congruent trials. In the occipital component (480 –640 ms), incongruent \ntrials preceded by congruent trials again showed more negative slopes than those \npreceded by another incongruent trial. \nThe frontal and occipital patterns were consistent with our predictions: congruent \ntrials preceded by another congruent trial showed the flattest spectral slopes, reflecting \nminimal engagement of inhibitory processes, whereas incongruent trials following a  \ncongruent trial exhibited the steepest slopes, indicating a stronger shift toward inhibition. \nFurthermore, at these two spatial components, the difference between current congruent \nand incongruent trials was not observed when trials followed incongruent t rials, similarly \nto our behavioral data where the congruency effect was significant only following the \ncongruent trials, suggesting a CSE pattern. Highlighting the usefulness of a time-resolved \napproach, these effects emerged at two distinct latencies, wit h the occipital effect \nfollowing the earlier frontal effect, indicating a temporal progression of conflict -related \nprocessing across cortical regions. Following the E:I balance framework (where steeper \nslopes indicate greater inhibition/lower excitation) a nd the CE interpretation proposed \nabove, the observed patterns at frontal and occipital sites jointly suggest that reduced \ninhibition of synaptic circuits occurred when conflict (i.e., interference from irrelevant \ninformation) was minimal, whereas increase d inhibition became evident as the level of \nconflict increased (this interpretation is reported in the model depicted in Fig. 10). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n32 \n \n \nFigure 10. From neuronal inhibition to cognitive control level. Schematic overview of the proposed mechanisms \nlinking neuronal-level changes in excitation and inhibition to cognitive control. (Top left) A shift in the local excitation–\ninhibition balance, captured by local field potentials (LFPs), reflects increased GABAergic inhibition relative to AMPA-\nmediated excitation. (Top right) This shift leads to spectral changes characterized by a steeper 1/f slope in the power \nspectrum. (Bottom right) At the brain-network level, increased inhibition manifests as enhanced activation in regions \ninvolved in resolving competing responses. (Bottom left) At the cognitive level, stronger cognitive control supports \nperformance in interference tasks, reflected by increased reaction times (RTs) for incongruent compared to congruent \ntrials. \n \nThis interpretation aligns with the conflict monitoring theory (Botvinick et al., 2001) \ndescribed in the Introduction. In our data, occipital effects occurred later than frontal \nchanges, consistent with a sequence in which conflict is detected in frontal (possibly ACC-\nrelated) regions and subsequently resolved through top -down modulation of occipital \nsensory processing.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n33 \n \nHowever, frontal and central components showed opposing patterns within the \nsame 160 –320 ms window. A potential explanation for this inconsistency is that the \nspectral slope may reflect multiple underlying mechanisms and/or neural substrates, with \ntheir relative contributions varying across scalp locations. As discussed below, our design \nwas not entirely free of potential confounds, such as those related to feature integration. \nTherefore, one possibility is that frontal activity reflects the recruitment of the \nfrontoparietal network during cognitive control processes (Niendam et al., 2012; Gratton, \nSun, & Petersen, 2018). In contrast, the central pattern may reflect the engagement of a \ndifferent network, such as the DMN. From this perspective, frontal activity would primarily \nsignal the implementation of cognitive control, whereas central activity might capture the \ncontribution of additional task-related mechanisms.  \n5. Limitations and Future Directions \nThis study is the first to examine the time -resolved aperiodic EEG in relation to \ncognitive control. While it provides novel insights, it also has certain limitations that point \nto directions for future research. First, the PWI task does not involve a full y confound-\nminimized design, making it difficult to disentangle feature integration effects from \ncognitive control engagement. This limitation may confound the interpretation of the CSE \nor increase the variability in our data, as feature integration proces ses can, at least to \nsome extent, obscure genuine adjustments in cognitive control (e.g., Duthoo et al., 2014; \nHommel et al., 2004; Mayr et al., 2003). Future research should adopt confound -\nminimized designs that disentangle feature integration from more t raditional cognitive \ncontrol processes to clarify the underlying mechanisms (Braem et al., 2019; Schmidt & \nLiefooghe, 2016; Weissman et al., 2014). It should be noted, however, that feature \nintegration and cognitive control likely operate in concert rather than in opposition to one \nanother (e.g., Abrahamse et al., 2016). \nSecond, the observed aperiodic effects are relatively small with partial ηp2 values \nranging from 0.08 to 0.14. While our sample size ( N = 49) is not small by EEG research \nstandards, future studies with larger sample sizes are needed to replicate these findings \nand ensure their robustness and generalizability (Clayson et al., 2019). Since this is the \nfirst study to apply this specific methodology, we lacked prior information to determine the \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n34 \n \noptimal sample size for reliable detection of aperiodic CSE effects, highlighting the \nimportance of follow -up research to confirm the stability of these effects. Relatedly, the \npresent findings were obtained using the PWI task, which primarily engages lexi cal-\nsemantic processing with concrete, non -abstract stimuli. It remains to be determined \nwhether similar time -resolved aperiodic dynamics would be observed in other cognitive \ncontrol paradigms, such as task -switching, flanker, or Stroop tasks (von Bastian et al., \n2020), which often involve abstract stimuli and place different demands on conflict \nmonitoring and control processes. \nMoreover, it should be acknowledged that the spectral slope could also reflect \nfluctuations in arousal (Mocchi et al., 2024), or motor preparation (Wilson et al., 2022), \nparticularly given the temporal proximity of the observed CE effects to response -related \nactivity. Furthermore, the spectral slope may reflect different processes—or combinations \nof processes —depending on the cortical region analyzed, underscoring a broader \nlimitation of surface-level aperiodic EEG studies. We speculated about distinct functional \ninterpretations of the spectral slope across scalp distributions. This reasoning parallels \nlong-standing findings in ERP research, where different components arise from distinct \nneural generators with specific spatiotemporal and functional propertie s (Fabiani et al., \n2007). To disentangle these contributions, future studies could combine aperiodic \nanalyses with source localization techniques or with techniques with better spatial \nresolution (e.g., EEG –fMRI) to better identify the cortical generators and functional \nsignificance of slope changes. Additionally, experimental manipulations targeting specific \nmechanisms—such as pharmacological modulation of the E:I balance or controlled \nchanges in arousal—could help clarify the functional specificity of aperiodic dynamics. \n6. Conclusions \nThe time-resolved analysis of the aperiodic EEG activity offers a novel approach \nto studying cognitive control, showing how inhibitory dynamics evolve throughout the trial. \nThis fine -grained approach demonstrates that the aperiodic activity indexes conflic t \nresolution and can capture adjustments in cognitive control processes. Specifically, our \nresults show slope modulations during incongruent trials consistent with increased \ninhibitory activity relative to congruent trials, which may be linked to control mechanisms, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n35 \n \nsuch as the suppression of irrelevant representations. In parallel, adjustments in cognitive \ncontrol driven by the previous trial context were reflected in slope modulations emerging \nfirst in frontal and central, and then occipital components, consistent w ith the temporal \nunfolding of the CSE (Botvinick et al., 2001). These CSE-related effects suggest that the \nspectral slope is sensitive not only to the immediate demands of the task, but also to \ninternal control states carried over from preceding trials, hi ghlighting its potential as a \nmarker of proactive adjustments in cortical excitability. Overall, this time -resolved \nframework highlights that aperiodic component could capture the spatial -temporal \ndynamics of cognitive control, offering a novel account of how the brain resolves conflict \nand flexibly adapts control in real time. \n \nAcknowledgments.  \nThis work was partly supported by NIA grant RF1AG062666 to M. Fabiani and G. Gratton. \nThis work was conducted in partial fulfillment of the Ph.D. degree requirements of Virginia \nTronelli.  \n \nReferences \nAkbarian, F., Rossi, C., Costers, L., D’hooghe, M. B., D’haeseleer, M., Nagels, G., & Van \nSchependom, J. (2024). Stimulus -related modulation in the 1/f spectral slope suggests an \nimpaired inhibition during a working memory task in people with multiple scle rosis. Multiple \nSclerosis Journal, 30(8), 1036–1046. https://doi.org/10.1177/13524585241253777 \nAbrahamse, E., Braem, S., Notebaert, W., & Verguts, T. (2016). Grounding cognitive control in \nassociative learning. Psychological Bulletin , 142(7), 693 –728. \nhttps://doi.org/10.1037/bul0000047  \nAhmad, J., Ellis, C., Leech, R., Voytek, B., Garces, P., Jones, E., ... & McAlonan, G. (2022). From \nmechanisms to markers: novel noninvasive EEG proxy markers of the neural excitation and \ninhibition system in humans. Translational Psychiatry, 12 (1), \n467.https://doi.org/10.1038/s41398-022-02218-z \nBotvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict \nmonitoring and cognitive control.  Psychological review , 108(3), \n624.https://doi.org/10.1037/0033-295X.108.3.624 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n36 \n \nBrake, N., Duc, F., Rokos, A., Arseneau, F., Shahiri, S., Khadra, A., & Plourde, G. (2024). A \nneurophysiological basis for aperiodic EEG and the background spectral trend. Nature \nCommunications, 15(1). https://doi.org/10.1038/s41467-024-45922-8 \nBraeken, J., & Van Assen, M. A. (2017). An empirical Kaiser criterion.  Psychological \nmethods, 22(3), 450. https://doi.org/10.1037/met0000074 \nBraem, S., Bugg, J. M., Schmidt, J. R., Crump, M. J. C., Weissman, D. H., Notebaert, W., & Egner, \nT. (2019). Measuring Adaptive Control in Conflict Tasks. Trends in Cognitive Sciences, 23(9), \n769–783. https://doi.org/10.1016/j.tics.2019.07.002 \nBuckner, R. L., Andrews ‐Hanna, J. R., & Schacter, D. L. (2008). The brain's default network: \nanatomy, function, and relevance to disease.  Annals of the new York Academy of \nSciences, 1124(1), 1-38. https://doi.org/10.1196/annals.1440.011 \nCavanagh, J. F., & Frank, M. J. (2014). Frontal theta as a mechanism for cognitive control. In \nTrends in Cognitive Sciences  (Vol. 18, Issue 8, pp. 414 –421). Elsevier Ltd. \nhttps://doi.org/10.1016/j.tics.2014.04.012 \nChini, M., Pfeffer, T., & Hanganu -Opatz, I. (2022). An increase of inhibition drives the \ndevelopmental decorrelation of neural activity. elife, 11, \ne78811.https://doi.org/10.7554/eLife.78811 \nClayson, P. E., Carbine, K. A., Baldwin, S. A., & Larson, M. J. (2019). Methodological reporting \nbehavior, sample sizes, and statistical power in studies of event‐related potentials: Barriers to \nreproducibility and replicability.  Psychophysiology, 56(11), e13437. \nhttps://doi.org/10.1111/psyp.13437 \nCohen, M. X., & Donner, T. H. (2013). Midfrontal conflict-related theta-band power reflects neural \noscillations that predict behavior. Journal of Neurophysiology , 110(12), 2752 –2763. \nhttps://doi.org/10.1152/jn.00479.2013 \nCousineau, D. (2005). Confidence intervals in within-subject designs: A simpler solution to Loftus \nand Masson’s method. Tutorials in quantitative methods for psychology, 1 (1), 42 -45. \nDOI:10.20982/tqmp.01.1.p042 \nDe Cesarei, A., Cavicchi, S., Micucci, A., & Codispoti, M. (2019). Categorization Goals Modulate \nthe Use of Natural Scene Statistics. Journal of Cognitive Neuroscience 31 (1): 109 –125. \ndoi:10.1162/jocn_a_01333  \nDe Cesarei, A., Cavicchi, S., Cristadoro, G., & Lippi, M. (2021). Do Humans and Deep \nConvolutional Neural Networks Use Visual Information Similarly for the Categorization of \nNatural Scenes? Cognitive Science, 45, e13009. doi: 10.1111/cogs.13009  \nDe Cesarei, A., D'Ascenzo, S., Nicoletti, R., Codispoti, M. (2023). Novelty and  learning \nin cognitive control: evidence from the Simon task. Psychological Research, 87, 2390–2406.  \nhttps://doi.org/10.1007/s00426-023-01813-z \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n37 \n \nDelorme, A., & Makeig, S. (2004). EEGLAB: An open source toolbox for analysis of single -trial \nEEG dynamics including independent component analysis. Journal of Neuroscience Methods, \n134(1), 9–21. https://doi.org/10.1016/j.jneumeth.2003.10.009 \nDonoghue, T., Haller, M., Peterson, E. J., Varma, P., Sebastian, P., Gao, R., ... & Voytek, B. \n(2020). Parameterizing neural power spectra into periodic and aperiodic components. Nature \nneuroscience, 23(12), 1655-1665. https://doi.org/10.1038/s41593-020-00744-x \nDuthoo, W., Abrahamse, E. L., Braem, S., Boehler, C. N., & Notebaert, W. (2014). The \nheterogeneous world of congruency sequence effects: An update. Frontiers in Psychology, 5, \n1001. https://doi.org/10.3389/fpsyg.2014.01001 \nEgner, T. (2007). Congruency sequence effects and cognitive control.  Cognitive, Affective, & \nBehavioral Neuroscience, 7(4), 380-390. https://doi.org/10.3758/CABN.7.4.380 \nEgner, T. (2023). Principles of cognitive control over task focus and task switching. Nature \nReviews Psychology, 2(11), 702–714. https://doi.org/10.1038/s44159-023-00234-4 \nEriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the identification of a target \nletter in a nonsearch task.  Perception & psychophysics , 16(1), 143 –149. \nhttps://doi.org/10.3758/BF03203267 \nFabiani, M., Gratton, G., & Federmeier, K. D. (2007). Event -related brain potentials: Methods, \ntheory, and applications.  Handbook of psychophysiology , 3, 85 –119. \nhttps://doi.org/10.1017/CBO9780511546396.004 \nFrelih, T., Matkovič, A., Mlinarič, T., Bon, J., & Repovš, G. (2024). Modulation of aperiodic EEG \nactivity provides sensitive index of cognitive state changes during working memory task. \nbioRxiv, 2024–05. https://doi.org/10.1101/2024.05.13.593835 \nGao, R., Peterson, E. J., & Voytek, B. (2017). Inferring synaptic excitation/inhibition balance from \nfield potentials. Neuroimage, 158, 70–78. https://doi.org/10.1016/j.neuroimage.2017.06.078 \nGrant, L. D., & Weissman, D. H. (2023). The binary structure of event files generalizes to abstract \nfeatures: A nonhierarchical explanation of task set boundaries for the congruency sequence \neffect. Journal of Experimental Psycholgy: Learning, Memory, and Cognition, 49 (7), 1033–\n1050. https://doi.org/10.1037/xlm0001148 \nGrant, L. D., & Weissman, D. H. (2019). Turning distractors into targets increases the congruency \nsequence effect. Acta Psychologica, 192, 31–41.  \nhttps://doi.org/10.1016/j.actpsy.2018.10.010Get rights and content \nGratton, G. (2018). Brain reflections: A circuit ‐based framework for understanding information \nprocessing and cognitive control.  Psychophysiology, 55(3), e13038. \nhttps://doi.org/10.1111/psyp.13038 \nGratton, G., Coles, M. G., & Donchin, E. (1992). Optimizing the use of information: strategic \ncontrol of activation of responses. Journal of Experimental Psychology: General, 121(4), 480. \ndoi: 10.1037//0096-3445.121.4.480. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n38 \n \nGratton, G., Coles, M. G. H., & Donchin, E. (1983). Technical Section A NEW METHOD FOR \nOFF-LINE REMOVAL OF OCULAR ARTIFACT t. In Electroencephalography and clinical \nNeurophysiology (Vol. 55). https://doi.org/10.1016/0013-4694(83)90135-9 \nGratton, G., Cooper, P., Fabiani, M., Carter, C. S., & Karayanidis, F. (2018). Dynamics of cognitive \ncontrol: Theoretical bases, paradigms, and a view for the future.  Psychophysiology, 55(3), \ne13016. https://doi.org/10.1111/psyp.13016 \nGyurkovics, M., Clements, G. M., Low, K. A., Fabiani, M., & Gratton, G. (2022). Stimulus-Induced \nChanges in 1/f-like Background Activity in EEG. Journal of Neuroscience, 42(37), 7144–7151. \nhttps://doi.org/10.1523/JNEUROSCI.0414-22.2022 \nGratton, C., Sun, H., & Petersen, S. E. (2018). Control networks and hubs. Psychophysiology, \n55(3), e13032. https://doi.org/10.1111/psyp.13032 \nGyurkovics, M., Kovacs, M., Jaquiery, M., Palfi, B., Dechterenko, F., & Aczel, B. (2020). \nRegistered Replication Report of Weissman, DH, Jiang, J., & Egner, T.(2014). Determinants \nof congruency sequence effects without learning and memory confounds.  Attention, \nPerception, & Psychophysics , 82(8), 3777-3787. https://doi.org/10.3758/s13414-020-02021-\n2 \nGyurkovics, M., & Levita, L. (2021). Dynamic Adjustments of Midfrontal Control Signals in Adults \nand Adolescents. Cerebral Cortex, 31(2), 795–808. https://doi.org/10.1093/cercor/bhaa258 \nGyurkovics, M., & Levita, L. (2021). Dynamic adjustments of midfrontal control signals in adults \nand adolescents. Cerebral Cortex, 31(2), 795-808. https://doi.org/10.1093/cercor/bhaa258 \nGyurkovics, M., Stafford, T., & Levita, L. (2020). Cognitive control across adolescence: Dynamic \nadjustments and mind -wandering. Journal of Experimental Psychology: General, 149 (6), \n1017–1031. https://doi.org/10.1037/xge0000698 \nHe, B. J. (2014). Scale -free brain activity: past, present, and future.  Trends in cognitive \nsciences, 18(9), 480–487. https://doi .org/10.1016/j.tics.2014.04.003 \nHe, W., Donoghue, T., Sowman, P. F., Seymour, R. A., Brock, J., Crain, S., ... & Hillebrand, A. \n(2019). Co -increasing neuronal noise and beta power in the developing brain. BioRxiv, \n839258. https://doi.org/10.1101/839258 \nHommel, B. (2004). Event files: Feature binding in and across perception and action. Trends in \ncognitive sciences, 8(11), 494–500. DOI: 10.1016/j.tics.2004.08.007 \nJasper, H. H. (1958). Ten -twenty electrode system of the international \nfederation. Electroencephalography and Clinical Neurophysiology, 10, 371–375. \nJia, S., Liu, D., Song, W., Beste, C., Colzato, L., & Hommel, B. (2024). Tracing conflict -induced \ncognitive-control adjustments over time using aperiodic EEG activity. Cerebral Cortex, 34(5). \nhttps://doi.org/10.1093/cercor/bhae185 \nJiménez, L., & Méndez, A. (2014). Even with time, conflict adaptation is not made of \nexpectancies. Frontiers in Psychology, 5, 1042. https://doi.org/10.3389/fpsyg.2014.01042 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n39 \n \nKaiser, H. F. (1958). The varimax criterion for analytic rotation in factor \nanalysis. Psychometrika, 23(3), 187–200. https://doi.org/10.1007/BF02289233 \nKałamała, P.*, Clements, G.M.*, Gyurkovics, M.*, Chen, T., Low, K.A., Fabiani, M., & Gratton, G. \n(revise and resubmit ). How to Improve the Reliability of Aperiodic Parameter Estimates in \nM/EEG: A Method Comparison and Recommendations for Best Practices. Psychophysiology. \nKałamała, P., Gyurkovics, M., Bowie, D. C., Clements, G. M., Low, K. A., Dolcos, F., Fabiani, M., \n& Gratton, G. (2024). Event -induced modulation of aperiodic background EEG: Attention -\ndependent and age -related shifts in E:I balance, and their consequences for behavior. \nImaging Neuroscience, 2, 1–18. https://doi.org/10.1162/imag_a_00054 \nKim, S., & Cho, Y. S. (2014). Congruency sequence effect without feature integration and \ncontingency learning. Acta Psychologica , 149, 60–6. \nhttps://doi.org/10.1016/j.actpsy.2014.03.004 \nLopez-Calderon, J., & Luck, S. J. (2014). ERPLAB: An open-source toolbox for the analysis of \nevent-related potentials. Frontiers in Human Neuroscience , 8(1 APR). \nhttps://doi.org/10.3389/fnhum.2014.00213 \nLu, R., Dermody, N., Duncan, J., & Woolgar, A. (2024). Aperiodic and oscillatory systems \nunderpinning human domain -general cognition. Communications Biology , 7(1), 1643. \nhttps://doi.org/10.1038/s42003-024-07397-7 \nManyukhina, V. O., Prokofyev, A. O., Obukhova, T. S., Stroganova, T. A., & Orekhova, E. V. \n(2024). Changes in high -frequency aperiodic 1/f slope and periodic activity reflect post -\nstimulus functional inhibition in the visual cortex. Imaging Neuroscience , 2, 1 –24. \nhttps://doi.org/10.1162/imag_a_00146 \nMayr, U., Awh, E., & Laurey, P. (2003). Conflict adaptation effects in the absence of executive \ncontrol. Nature neuroscience, 6(5), 450–452. https://doi.org/10.1038/nn1051 \nMazoyer, B., Zago, L., Mellet, E., Bricogne, S., Etard, O., Houdé, O., ... & Tzourio -Mazoyer, N. \n(2001). Cortical networks for working memory and executive functions sustain the conscious \nresting state in man. Brain research bulletin, 54(3), 287–298. https://doi.org/10.1016/S0361-\n9230(00)00437-8. \nMocchi, M., Bartoli, E., Magnotti, J., de Gee, J. W., Metzger, B., Pascuzzi, B., ... & Bijanki, K. \n(2024). Aperiodic spectral slope tracks the effects of brain state on saliency responses in the \nhuman auditory cortex. Scientific Reports, 14(1), 30751. https://doi.org/10.1038/s41598-024-\n80911-3 \nMyrov, V., Siebenhühner, F., Juvonen, J. J., Arnulfo, G., Palva, S., & Palva, J. M. (2024). \nRhythmicity of neuronal oscillations delineates their cortical and spectral \narchitecture. Communications biology, 7(1), 405. https://doi.org/10.1038/s42003-024-06083-\ny \nNiendam, T. A., Laird, A. R., Ray, K. L., Dean, Y. M., Glahn, D. C., & Carter, C. S. (2012). Meta-\nanalytic evidence for a superordinate cognitive control network subserving diverse executive \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n40 \n \nfunctions. Cognitive, Affective, & Behavioral Neuroscience, 12 , 241 –268. \nhttps://doi.org/10.3758/s13415-011-0083-5 \nRaichle, M. E. (2015). The brain's default mode network. Annual review of neuroscience, 38(1), \n433–447. https://doi.org/10.1146/annurev-neuro-071013-014030 \nRosinski, R. R. (1977). Picture-Word Interference Is Semantically Based  (Vol. 48, Issue 2). \nhttps://www.jstor.org/stable/1128667 \nSchiltenwolf, M., Kiesel, A., Frings, C., & Dignath, D. (2024). Memory for abstract control states \ndoes not decay with increasing retrieval delays. Psychological Research, 88(2), 547–561. \nhttps://doi.org/10.1007/s00426-023-01870-4 \nSchmidt, J. R., & Liefooghe, B. (2016). Feature integration and task switching: Diminished switch \ncosts after controlling for stimulus, response, and cue repetitions. PloS one, 11(3), e0151188. \nhttps://doi.org/10.1371/journal.pone.0151188 \nSchmidt, J. R., & Weissman, D. H. (2014). Congruency sequence effects without feature \nintegration or contingency learning confounds.  PLoS One , 9(7), e102337. \nhttps://doi.org/10.1371/journal.pone.0102337 \nSteiner, M. D., & Grieder, S. (2020). EFAtools: An R package with fast and flexible \nimplementations of exploratory factor analysis tools. Journal of Open Source Software, 5(53), \n2521. https://doi.org/10.21105/joss.02521 \nShulman, G. L., Fiez, J. A., Corbetta, M., Buckner, R. L., Miezin, F. M., Raichle, M. E., & Petersen, \nS. E. (1997). Common blood flow changes across visual tasks: II. Decreases in cerebral \ncortex. Journal of cognitive neuroscience , 9(5), 648 –663. \nhttps://doi.org/10.1162/jocn.1997.9.5.648 \nThomson, G. H. (1938). The Factorial Analysis of Human Ability. London: University of London \nPress \nThurstone, L. L. (1935). The vectors of mind: Multiple-factor analysis for the isolation of primary \ntraits. University of Chicago Press. https://doi.org/10.1037/10018-007 \nTronelli, V., Codispoti, M., & De Cesarei, A. (2025). EXPRESS: Cognitive control during scene \ncategorization: The role of identity repetition and timing in Congruence Sequence \nEffects. Quarterly Journal of Experimental \nPsychology, 0(ja). https://doi.org/10.1177/17470218251335293 \nvan Maanen, L., van Rijn, H., & Borst, J. P. (2009). Stroop and picture-word interference are two \nsides of the same coin. Psychonomic Bulletin and Review , 16(6), 987 –999. \nhttps://doi.org/10.3758/PBR.16.6.987  \nvon Bastian, C. C., Blais, C., Brewer, G., Gyurkovics, M., Hedge, C., Kałamała, P., ... & Wiemers, \nE. (2020). Advancing the understanding of individual differences in attentional control: \nTheoretical, methodological, and analytical considerations. \nhttps://doi.org/10.31234/osf.io/x3b9k \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint \n\n41 \n \nVoytek, B., Kramer, M. A., Case, J., Lepage, K. Q., Tempesta, Z. R., Knight, R. T., & Gazzaley, \nA. (2015). Age -related changes in 1/f neural electrophysiological noise. Journal of \nneuroscience, 35(38), 13257–13265. https://doi.org/10.1523/JNEUROSCI.2332-14.2015 \nVoytek, B., & Knight, R. T. (2015). Dynamic network communication as a unifying neural basis for \ncognition, development, aging, and disease.  Biological psychiatry , 77(12), 10891097. \nhttps://doi.org/10.1016/j.biopsych.2015.04.016 \nWaschke, L., Donoghue, T., Fiedler, L., Smith, S., Garrett, D. D., Voytek, B., & Obleser, J. (2021). \nModality-specific tracking of attention and sensory statistics in the human electrophysiological \nspectral exponent. elife, 10, e70068. https://doi.org/10.7554/eLife.70068 \nWeissman, D. H. (2019). Let your fingers do the walking: Finger force distinguishes competing \naccounts of the congruency sequence effect.  Psychonomic Bulletin & Review, 26(5), 1619–\n1626. https://doi.org/10.3758/s13423-019-01626-5. \nWeissman, D. H., Jiang, J., & Egner, T. (2014). Determinants of congruency sequence effects \nwithout learning and memory confounds.  Journal of Experimental Psychology: Human \nPerception and Performance, 40(5), 2022. https://doi.org/10.1037/a0037454 \nWilson, L. E., da Silva Castanheira, J., & Baillet, S. (2022). Time -resolved parameterization of \naperiodic and periodic brain activity. Elife, 11, e77348. https://doi.org/10.7554/eLife.77348 \nYan, J., Yu, S., Mückschel, M., Colzato, L., Hommel, B., & Beste, C. (2024). Aperiodic neural \nactivity reflects metacontrol in task -switching. Scientific Reports , 14(1), 24088. \nhttps://doi.org/10.1038/s41598-024-74867-7 \nZhang, C., Stock, A. K., Mückschel, M., Hommel, B., & Beste, C. (2023). Aperiodic neural activity \nreflects metacontrol. Cerebral Cortex , 33(12), 7941 –7951. \nhttps://doi.org/10.1093/cercor/bhad089 \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted December 22, 2025. ; https://doi.org/10.64898/2025.12.19.695375doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}