Using event-related potentials, it was found that selective attention to valence facilitates early affective discrimination of words with task-relevant valence and inhibits affective processing of words with task-irrelevant valence. This attention-based modulation of affective processing presumably relies on prior associative learning linking visual word forms with their affects. To investigate this hypothesis, we employed a valence-detection task and manipulated lexical (length, frequency) and affective (arousal) word features. Since we assumed that these features strongly influence the strength of visual form–affect associations, we expected them to play a crucial role in early affective discrimination. Fifty-eight participants made speeded responses only to words of one predefined target level of valence (negative, neutral, or positive), which varied across three blocks. As expected, the visual P1 component yielded greater valence discrimination for the target than for nontarget words. This interactive effect was most prominent for short, high-frequency and low-arousal words, respectively. Regarding the N170 component, low-frequency words showed higher amplitudes when they were either positive low-arousing or negative high-arousing compared with the other two sets of words, independently of target status. Additionally, an average-referenced EPN-like posterior negativity (150–270 ms) revealed a target-independent interaction between valence and arousal and increased amplitudes for negative target words. Results extend previous research in showing that particularly short and highly frequent valent word forms can be tuned by selective attention to valence, facilitating early affective discrimination. Finally, findings support the notion that valence and arousal interact during early preattentive, bottom-up processing which is interpreted within the valence–arousal conflict theory.

1 INTRODUCTION

Every day, humans are surrounded by a multitude of environmental stimuli. As this stream of information often exceeds our capacities, specific classes of stimuli are given preferential consideration. Among these are emotional stimuli, which are highly relevant for survival and well-being (Brosch et al., 2010; Pourtois & Vuilleumier, 2006). A poisonous snake signals an immediate threat, while a smiling face communicates a potential ally that can be approached. Thus, detecting, selecting, and responding to emotional stimuli rapidly and efficiently is advantageous and crucial for guiding behavior. To investigate emotional processing, researchers often rely on emotional scenes such as erotic pictures or images of mutilations (e.g., Olofsson et al., 2008; Sabatinelli et al., 2011) and human faces (e.g., Schindler & Bublatzky, 2020; Vuilleumier & Pourtois, 2007). Such stimuli undoubtedly play a significant role in our social environment. However, humans have evolved to convey and receive affective signals by means of another category of stimuli in everyday life, that is, emotional words (Citron, 2012; Kissler et al., 2006). For example, a love letter, even though it consists of strings of characters that have no emotional connotation on their own, is highly likely to rapidly elicit an affective response in the reader. Interestingly, part of the emotional content of such words is already processed prior to the retrieval of semantic meaning (see Hinojosa et al., 2020) and without conscious perception (Banse, 1999; Gibbons, 2009). Event-related brain potentials (ERPs) with their excellent temporal precision are perfectly suited to improve our understanding of the neurocognitive mechanisms underlying the rapid processing of emotional words.

Emotional words are usually described among a two-dimensional structure of affect, including emotional valence as the degree of pleasantness and arousal as the extent of activation associated with affect (Barrett & Russell, 1998; Bradley & Lang, 2000; Russell, 1980). Neural processing of these two emotional word features has been most consistently found during or after lexico-semantic processing which is thought to start at around 200 ms (see Citron, 2012; Hinojosa et al., 2020; Kissler et al., 2006). In particular, larger amplitudes of the early posterior negativity (EPN) as a marker of “natural selective attention” (Olofsson et al., 2008) and the late positive potential as an index of sustained controlled attention (Hajcak et al., 2010) have been reported for emotional compared with neutral words (e.g., Citron et al., 2013; Frühholz et al., 2011; Gibbons, Kirsten, & Seib-Pfeifer, 2022; Herbert et al., 2008; Schacht & Sommer, 2009b; Schindler & Kissler, 2016). However, several ERP studies have indicated that affective word features might be processed, at least partially, at pre-lexical stages and thus before any controlled semantic analysis (for reviews, see Hinojosa et al., 2020; Kissler et al., 2006). In particular, emotion effects were reported for the lateral-posterior P1, peaking around 100 ms. This component is suggested to mainly reflect low-level perceptual analysis of word features (such as word length, Dien, 2009); nevertheless, emotion effects have emerged including both decreased (Bayer et al., 2012; Gibbons et al., 2023; Kuchinke et al., 2014) and increased P1 amplitudes for negative words as compared with positive or neutral words (Schindler et al., 2019; Zhang et al., 2014). These effects were sometimes restricted to highly arousing words (Hofmann et al., 2009; Scott et al., 2009). Similarly to P1, the word N1701 generated in the left fusiform area also known as the visual word form area (Posner & Petersen, 1990) is primarily thought to be involved in early structural word processing (Dien, 2009). Findings on visual emotional words indicate that the N170 might also be modulated by emotional content (Gibbons et al., 2023; Kissler et al., 2009; Kissler & Herbert, 2013; Scott et al., 2009; Yao et al., 2016; Zhang et al., 2014). Even though the literature is somewhat inconclusive as several studies did not report emotion effects on early perceptual components (Bayer & Schacht, 2014; Kissler et al., 2007; Palazova et al., 2011; Recio et al., 2014; Schacht & Sommer, 2009a), a growing body of findings suggests that emotion effects can occur before lexico-semantic word analysis.

One possible explanation for these early emotion effects relies on associative learning mechanisms, that is, the linking of the visual word form with its affective connotation or valence (see Fritsch & Kuchinke, 2013; Gibbons et al., 2023; Hinojosa et al., 2020; Hofmann et al., 2009; Keuper et al., 2014; Kissler et al., 2006). Over repeated encounters, we learn to associate a word's valence with its visual form. Consequently, each presentation of the word form should simultaneously activate neural representations of the associated valence (hypothesis of valent word forms; for a more detailed explanation, see Gibbons et al., 2023; for similar concepts, see Bayer et al., 2019; Keuper et al., 2014; Kissler et al., 2006). Consistent with this idea, recent ERP studies have shown that pseudowords and neutral words which previously acquired an affective connotation showed a modulation of the P1 amplitude (Bayer et al., 2019; Fritsch & Kuchinke, 2013; Kuchinke et al., 2015; Kulke et al., 2019). Here, an increased P1 for negatively associated pseudowords relative to new pseudowords is the most prominent finding (Bayer et al., 2019; Kuchinke et al., 2015; Kulke et al., 2019). Learned associations between word form and valence, however, cannot only explain P1 emotion effects but also offer novel predictions. If we selectively attend to a specific valence, neural representations encoding this valence should be pre-activated before the appearance of the stimulus (attentional tuning as a concept of feature-based visual attention, see Battistoni et al., 2017; Maunsell & Treue, 2006; Peelen & Kastner, 2011). This activation should then spread to neural networks coding for word forms that are associated with the currently attended valence. Eventually, the pre-activated neural representations of word forms should facilitate rapid visual integration and perception of the word shape which should be reflected in amplitude modulations of P1 and N170 as markers of early perceptual word form processing.

This hypothesis has recently been tested by Gibbons et al. (2023) employing a valence-detection task (paradigm first introduced by Schupp et al., 2007). They examined 40 positive, neutral, and negative words each whose specific valence level was either selectively attended (as a target) or not (as a nontarget). A valence main effect revealed a larger P1 amplitude for positive than negative words. Crucially, an interaction effect of (target) status and valence supported top-down tuning of valent word forms, in terms of greater P1 valence discrimination (positive > negative) for targets compared with nontargets. Additionally, emotional target words elicited a larger N170 amplitude than neutral target words while the pattern was reversed for nontarget words. This indicates that selective attention to valent words might not only facilitate the processing of relevant words but also involve the inhibition of currently irrelevant words. There is only one comparable study investigating emotional words in a valence-detection task (Schindler & Kissler, 2016); however, no results were reported for early ERP components (P1 and N170) as the focus was on subsequent processes. While the facilitated neural processing of selectively attended emotional words at early stages is an intriguing finding, more research needs to be conducted to corroborate and expand the previous findings, and the present study is one such attempt.

1.1 The present study

We aimed to continue the investigation of valent word forms from previous ERP research by Gibbons et al. (2023). While the former study investigated effects involving the factors valence (positive, neutral, and negative words) and status (target or nontarget), a novel study including further adjustments was designed to corroborate and expand our prior findings. First, in the present study, we employed a more basic two-level valence-detection task. In particular, only positive and negative words but no neutral words were included in the task. The underlying rationale was that in the absence of neutral words, the discrimination between positive and negative words is more unambiguous and more precise due to the greater distance in terms of valence. High error rates when attending to neutral words in the three-level valence-detection task support this reasoning (Gibbons et al., 2023). Consequently, top-down tuning processes and hence the early visual integration of affective words should be improved, presumably increasing the likelihood of uncovering target-dependent ERP valence discrimination effects.

Second, we added the two lexical variables word length and word frequency as additional orthogonal factors in our design. They have previously been found to interact with word valence (Scott et al., 2009), and their inclusion allows for the possibility of examining the hypothesis of valent word forms in greater detail. High-frequency words, as compared with low-frequency words, are processed more quickly in lexical and semantic decision tasks (word frequency effect, Brysbaert et al., 2018). This is often interpreted in terms of a learning effect with activation levels of word representations varying as a function of frequency (Brysbaert et al., 2018). On a neural level, the earliest differences were reported within 200 ms after visual word presentation (P1, N1, e.g., Hauk et al., 2006; Hauk & Pulvermüller, 2004; Penolazzi et al., 2007), with frequent words eliciting lower amplitudes than rare words in early ERP components. Word frequency effects have also been found to be modulated by valence (Méndez-Bértolo et al., 2011; Palazova et al., 2011; Scott et al., 2009). Scott et al. (2009) revealed that highly frequent words were influenced by valence, with frequent negative words showing a smaller P1 than frequent positive or neutral words, while no differences were found for less frequent words. This interaction emphasizes the importance of word frequency for the early perceptual processing of emotional words. In addition to such empirical findings, the need to take word frequency into account can be derived from a theoretical perspective. As frequent words are encountered more often over one's lifetime, the representations of word form and valence are coactivated more frequently, yielding stronger associations between the respective neural networks for high-frequency words. Therefore, preparatory activity for a certain valence induced by selective attention (in the valence-detection task) should primarily spread to frequent word forms, thus facilitating particularly strong early integration and processing of words of that valence.

Word length is also associated with early visual word processing (Hauk et al., 2009), although behavioral findings regarding reaction times are less consistent than for word frequency (Oganian et al., 2016). What appears to be the case, however, is that short words are encoded perceptually more quickly than long words (Oganian et al., 2016), differing in their respective ERPs as early as 100 ms after stimulus presentation (Hauk et al., 2006; Hauk & Pulvermüller, 2004). Processing the more complex visual features of long words is reflected in higher amplitudes of the lateral parieto-occipital P1 (Hauk et al., 2006). To the best of our knowledge, there are no empirical studies investigating early word processing while manipulating (or taking into account) both word length and valence. However, the value of doing so can again be based on theoretical considerations. Short word forms have less complex and more easily distinguishable visual features which should facilitate stronger, less ambiguous associations with valence over time. In contrast, long word forms are harder to distinguish from one another and should have weaker associations with valence. Hence, selectively attending to one specific valence should favor the preactivation of the associated short word forms, resulting in a greater signal-to-noise ratio and valence discrimination specifically for short valent word forms. Finally, including both word length and frequency in one experiment allows to detect potential interaction effects between these two lexical variables on behavior and ERP components (Balota et al., 2004; Penolazzi et al., 2007) in the valence-detection task.

The third extension to the previous study by Gibbons et al. (2023) involves the investigation of arousal as an additional orthogonal factor. The two key dimensions of affect have been found to exert both independent as well as interactive effects on word processing (Carretié et al., 2008; Citron et al., 2013; Delaney-Busch et al., 2016; Hinojosa et al., 2009; Hofmann et al., 2009; Recio et al., 2014; Yao et al., 2016; for review, see Citron, 2012). Specifically for early processing at P1, Hofmann et al. (2009) reported facilitation, that is, a higher amplitude for high-arousal negative but not positive words. While former studies of valence detection in words have not investigated arousal (Gibbons et al., 2023; Schindler & Kissler, 2016), the full factorial manipulation may help disentangle their respective influence at early stages in the current study. Additionally, it is plausible that the word's arousal plays an important role within the concept of valent word forms. According to Mather (2007), highly arousing objects elicit focused attention that enhances within-object memory binding, thereby strengthening associations between the object and its features. Hence, it is reasonable to assume that the association between visual word form and valence (both within-object features) is particularly strong in highly arousing words. Applied to the current task, it needs to be investigated whether selective attention to one target level of valence may indeed facilitate the early processing, particularly of arousing valent word forms.

Furthermore, including arousal and valence allowed us to investigate dimensional models of affect. In these, the two affective (word) dimensions are often conceptualized as being independent (Barrett & Russell, 1998; Russell & Barrett, 1999). However, it has been suggested that during the initial stages of emotional appraisal and processing, arousal and valence actually interact (Robinson, 1998; Robinson et al., 2004). Positive valence and low arousal are associated with approach tendencies, whereas negative valence and high arousal lead to withdrawal (Robinson, 1998). These tendencies are initiated independently at preattentive stages. Integration of the affective dimensions and processing should thus be facilitated for congruent stimuli, that is, for positive low arousal (PL) and negative high arousal (NH) stimuli, compared with positive high arousal (PH) and negative low arousal (NL) words (valence–arousal conflict theory, Robinson et al., 2004). A very similar idea is captured in the evaluative space model (ESM, Cacioppo et al., 1997; Cacioppo & Berntson, 1994), which proposes two biases of emotional processing. At low levels of emotional input (low arousal situations), responses to positive affect are stronger compared with negative affect, which is called the positivity offset. However, for high levels of emotional input (high intensity), a stronger response to negative stimuli can be observed, known as the negativity bias (see Norris et al., 2010).

The conflict theory is supported by results from explicit (Citron et al., 2016; Robinson et al., 2004) and implicit affective tasks (Citron et al., 2014; Eder & Rothermund, 2010; Yao et al., 2016), yielding faster behavioral responses to congruent compared with incongruent stimuli. Correspondingly, ERP studies revealed interaction effects between valence and arousal during affective word processing (Citron et al., 2013; Hofmann et al., 2009; Recio et al., 2014; Yao et al., 2016). During the N170 time range, NH words elicited higher amplitudes than NL words with no differences between PH and PL words (Yao et al., 2016). For EPN, Citron et al. (2013) found marginally larger amplitudes for PH and NL words than for PL and NH words, interpreting these as facilitated processing of congruent words at the early stages of word recognition. Similarly, Recio et al. (2014) reported a processing advantage for PL compared with PH words (“positivity offset”) during the EPN window. However, some studies found only independent main effects of valence and arousal for P1 and EPN (e.g., Bayer et al., 2012; Delaney-Busch et al., 2016). Recently, Gibbons, Schmuck, and Kirsten (2022) pursued a more indirect way of testing congruency biases in the valence-detection task. We reanalyzed data on nontarget ERPs (Gibbons, Kirsten, & Seib-Pfeifer, 2022) as a function of both the word's valence and the (different) valence of the attentional set of the current block. Under standard conditions (neutral set), a smaller lateral-posterior negativity (resembling EPN but with an earlier onset; N170/EPN) for positive compared with negative nontarget words was observed. According to Robinson et al. (2004), integration of bias-congruent words will be faster and therefore processing at N170/EPN should be mainly driven by NH words and PL words, for which integrated affect is already available. Given that the EPN is particularly sensitive to differences in arousal, this explains smaller amplitudes for positive compared with negative nontarget words in Gibbons, Schmuck, and Kirsten (2022).

While neural evidence for the valence–arousal conflict theory originates mainly from implicit tasks (e.g., lexical decision task; LDT), the question remains whether benefits in the processing of bias-congruent words also occur in a valence-detection task. As the evaluation and integration of the valence and arousal dimensions are supposed to take place at preattentive stages (Robinson et al., 2004) and might therefore reflect a bottom-up process, the early posterior ERP components involved in visual emotional word processing (P1, N170) and early markers of automatic attention toward emotional words (EPN) are of particular interest. Following the idea that preattentive integration of valence and arousal in congruent words should be facilitated, differences in neural processing (indicated by ERP amplitudes) between bias-congruent and incongruent words can be expected for early components (Citron et al., 2013; Hofmann et al., 2009; Recio et al., 2014; Yao et al., 2016).

1.2 Hypotheses

Due to the lack of reaction times in nontarget trials (where no response is required), we were unable to derive any behavioral hypothesis regarding attentional tuning, which would pertain to target-nontarget differences. However, in line with valence–arousal conflict theory, we predicted shorter (hit) reaction times (H1) and fewer errors (misses; H2) for bias-congruent target words (PL, NH) than for incongruent target words (PH, NL, Citron et al., 2014; Robinson et al., 2004; Yao et al., 2016).

Our investigation of ERPs was restricted to early components (P1, N170, EPN). On the one hand, attentional tuning to valent word forms should mainly affect early perceptual processes even before lexico-semantic processing (see above and Gibbons et al., 2023). On the other hand, valence–arousal conflict theory states that the integration of affective dimensions takes place at preattentive stages. Hence, we focused on early components including EPN as an early marker of automatic attention which has previously been associated with facilitated processing for congruent words (e.g., Citron et al., 2013). In line with attentional tuning toward word forms, we expected a greater valence discrimination in words whose valence was currently selectively attended (target words), compared with words displaying a nontarget level of valence. In particular, this should be reflected in greater lateral parieto-occipital P1 amplitude differences between valent (positive vs. negative) target words compared with nontarget words (H3, see Gibbons et al., 2023).2 Derived from theoretical considerations and related empirical findings (e.g., Scott et al., 2009), we further hypothesized that the P1 valence discrimination in words with the target valence should be carried more by frequent than rare words (H4), more by short compared with long words (H5) and more by high-arousal than low-arousal words (H6).

Finally, we put forward our last hypothesis derived from the valence–arousal conflict theory. The mean amplitudes of P1, N170, and/or EPN should be greater for bias-congruent words (PL, NH) than for incongruent words (PH, NL; H7). The respective effects were expected to be unaffected by status, or even stronger in nontarget words, because they are bottom-up in nature and not useful for performance on the valence-detection task. In addition to our preceding hypotheses, we also investigated the modulation of valence × arousal interactions (related to the valence–arousal conflict theory) by frequency and length in an exploratory manner for all three components.

2 METHOD

2.1 Sample

Sixty-six German native speakers participated in the current EEG study. The majority were recruited at the Department of Psychology at the University of Bonn and received partial course credit for their participation. Due to technical problems during EEG recording and/or large artifacts, eight participants were excluded from further analysis. The resulting 58 participants were on average 22.60 years old, ranging from 18 to 33 years (SD = 3.49). The sample consisted of 50 females and 8 males; approximately 90% were right-handed. All participants had normal or corrected-to-normal vision, reported no history of neurological or psychiatric disorders, and gave written informed consent. The principal study protocol was approved by the local ethics committee (#15-08-10) and was in accordance with the Declaration of Helsinki.

2.2 Stimulus and apparatus

2.2.1 Initial stimulus selection

The word stimuli for the present study were selected based on the findings from an inhouse online pre-study. In this pre-study, 67 participants rated 294 adjectives from German databases (Schwibbe et al., 1994; Võ et al., 2009) complemented by ~20 newly added adjectives. Ratings were given for valence and arousal on Likert scales ranging from 1 to 7, while word length was defined by the number of letters and word frequency was taken from the Leipzig Corpora Collection on German words (wortschatz.uni-leipzig.de; accessed 1 February 2021). This rating process was done to ensure compatibility between both the word ratings from different lexical databases and the new words and to select the most positive and negative but otherwise still comparable words (see below). The final selection consisted of 72 positive and 72 negative adjectives (144 adjectives in total) that did not differ significantly on arousal ratings, word length and word frequency, and several nonaffective control parameters (letters, syllables, mean bigram frequency), all t(142) < 1.06, p > .291. Within-valence median splits were computed to subdivide positive and negative words into categories of low- and high-arousal words, short and long words, and frequent and rare words. This resulted in 16 groups of words according to the combinations of the four experimental factors valence (positive/negative), arousal (high/low), word length (short/long), and word frequency (low/high).

An inspection of the behavioral results after data collection of the present experiment revealed that three words (herzlos [heartless], lieblos [loveless], and kritikfähig [being able to take criticism]) produced particularly many false responses and had particularly high reaction times on correct trials. This was probably due to the fact that these words were composite words starting with a word part that had a valence opposite to the valence of the word as a whole (e.g., the highly positive word “love” ending in “less,” making the overall compound “loveless” highly negative). It is plausible that this triggered numerous false alarms in our valence-detection task under speed instructions (e.g., in the attend-positive block, the word part “love” in “loveless” produced many quick responses although the response actually had to be withheld as “loveless” is a negative word).

2.2.2 Final selection

The above-mentioned three words were excluded from all analyses, resulting in a final sample of 141 words. As shown by control analyses based on 144 words, this did not alter the central ERP findings. A summary of the descriptive statistics (means and standard deviations) for each combination of the factors on the relevant affective variables is shown in Table S1. The valence ratings differed significantly between negative words (M = 2.13, SD = 0.34), which were rated lower than positive words (M = 5.85, SD = 0.36), t(139) = 63.47, p < .001, d = 10.7. Highly arousing words (M = 4.84, SD = 0.30) had higher arousal scores than less arousing words (M = 3.97, SD = 0.35), t(135) = 15.56, p < .001, d = 2.62. Further comparisons for word length and frequency revealed that longer words (M = 9.20, SD = 1.29) had significantly more letters than short words (M = 5.89, SD = 1.19), t(138) = 15.84, p < .001, d = 2.67, while frequent words (M = 17.99, SD = 23.31) had higher mean word occurrence per million words than rare words (M = 1.53, SD = 1.22), t(71) = 5.98, p < .001, d = 1.00. All comparisons were calculated using the Welch's t-test.

To assess any potential dependencies among the word-related factors arousal, valence, frequency, and length, we carried out 4 three-way ANOVAs with one of the factors as continuous dependent variable and the other three factors as dichotomous predictors, respectively. The three-way ANOVA with valence ratings as dependent variable revealed no modulations by any (combination) of the three other factors arousal, length, and frequency, all F(1,133) < 2.31, p > .131, ${\eta}_p^2$ < 0.02. Similarly, all further three-way ANOVAs with dependent variables arousal [all F(1,133) < 3.41, p > .067, ${\eta}_p^2$ < 0.03], word length [all F(1,133) < 1.75, p > .188, ${\eta}_p^2$ < 0.03], and word frequency [all F(1,133) < 1.85, p > .176, ${\eta}_p^2$ < 0.02] revealed no significant influences by the three remaining factors in each case.

Words were shown in 30-point white sans-serif font (Arial) on a black background in the center of a 23″ TFT screen running at a resolution of 1920 × 1080 at a viewing distance of ~80 cm. The experiment was run on Presentation 21.1 (Neurobehavioral Systems Inc., Berkeley, CA).

2.3 Experimental design

In the current experiment, participants performed a valence-detection task with emotional words. At the beginning of each block, they were instructed to respond with a single button press (space bar) only to words of the designated target valence for this block (either positive or negative). The response to these target trials should be given as quickly and accurately as possible. If the target valence for the block and the valence of the word did not match, participants had to withhold their response in these nontarget trials. The experiment consisted of two blocks (target levels positive and negative) with 144 trials each. Each trial started with the presentation of a white fixation cross (40-point) in the center of the screen for a random duration between 1000 and 1500 ms. Next, a single emotional word was shown in the center of the screen for a maximum of 1500 ms. On target trials, the word disappeared after the button press and a new trial started. If no response was given within 1500 ms, the feedback “Too slow!” (shown in German as “Zu langsam!”) was displayed for 1000 ms. Conversely, a response on nontarget trials resulted in the feedback “Wrong!” (shown in German as “Falsch!”) for 1000 ms. All 144 adjectives were presented in random order once per block while the order of the blocks (positive or negative as target valence) was counterbalanced across participants. Between blocks, participants could take a self-paced break.

2.4 EEG acquisition and preprocessing

EEG data were recorded from 61 Ag/AgCl electrodes of the 10% system (Homan et al., 1987) using a digital 64-channel BrainAmp system (Brain Products, Gilching, Germany). The ground was located between AFz and AF4, while FCz served as the online reference. Vertical and horizontal electrooculogram (EOG) were monitored from electrodes below and above the right eye and the outer left and right canthi, respectively. EEG was recorded at 500-Hz sampling rate, the band pass included frequencies between 0.1 and 70 Hz. All electrodes showed impedances below 5 kΩ.

EEG data were processed and analyzed using VisionAnalyzer 2 (Brain Products, Gilching, Germany). First, bad channels were replaced by the average of neighboring channels. On average, 0.31 channels per participant had to be replaced (maximum eight channels). The continuous data were re-referenced against algebraically linked mastoids (TP9, TP10), and exclusively for EPN analysis, an average reference was computed. This follows the recommendation to analyze the EPN in average reference which allows a better visualization of the component while its identification can be difficult in mastoid-referenced data (Junghöfer et al., 2006) and is in accordance with the procedure in our previous valence-detection study (see Gibbons et al., 2023; Gibbons, Schmuck, & Kirsten, 2022). Data were then high-pass filtered at 0.1 Hz (−24 dB/oct roll-off) and low-pass filtered at 30 Hz (−48 dB/oct). Additionally, a notch filter was applied at the powerline frequency of 50 Hz. After using an ICA-based method to correct for eye blink artifacts, continuous data were segmented into epochs of 900 ms in length, thus equaling an (−100, 800 ms) interval relative to the onset of the presented word. Only target trials with correct responses and nontarget trials without a response were included. Epochs were baseline corrected to the 100-ms prestimulus interval and segments containing amplitudes exceeding ±100 μV were rejected. On average, for each participant, 271.41 artifact-free epochs (98.41% of all error-free trials; range: 242–282 epochs) remained in the dataset and were finally averaged separately for all combinations of the experimental factors (target) status, valence, arousal, word length, and word frequency.

2.5 Statistical analysis

Statistical analyses were conducted in R (R Core Team, 2021) using the tidyverse packages (Wickham et al., 2019) for data preparation, the afex package (Singmann, 2022) for repeated measures ANOVAs and emmeans (Lenth, 2022) for follow-up t-tests. For both behavioral and EEG data analysis, only significant main effects as well as interaction effects involving status × valence (related to attentional tuning toward the target valence) or valence × arousal (related to valence–arousal conflict theory) are reported. The full ANOVA tables can be found in Tables S2–S6, whereas the mean ERP amplitudes and respective standard errors are presented in Table S7.

Reaction times (RT) in target trials with correct responses were analyzed using a four-way repeated measures (rm) ANOVA with factors valence (positive/negative), arousal (low/high), word length (short/long), and word frequency (low/high). For the analysis of errors (i.e., misses in the target condition and false alarms in the nontarget condition), the four-way rm. ANOVA was extended by adding the factor status (target/nontarget). Finally, for the analysis of ERP components, mean amplitudes in the respective time windows were subjected to a six-way omnibus rm. ANOVA, additionally including the factor laterality (left/right). If no relevant modulation of the effects of interest by laterality was found, the left and right cluster were aggregated and analyzed in an analogous five-way ANOVA. As a general principle for all behavioral and ERP analyses, in case of significant higher order interactions, 2 × 2 subcells were identified by close inspection in which effects of interest (status × valence or valence × arousal interactions) were significant and numerically greater (i.e., had a greater F value) compared with corresponding neighboring 2 × 2 subcells. On the 2 × 2 level, Bonferroni–Holm corrected pairwise t-tests were used to statistically confirm these observations. If necessary and not already done in prior testing, Bonferroni–Holm corrected pairwise t-tests were additionally used to test for significant differences between conditions, as predicted in our hypotheses. For all analyses, p-values below .05 are considered significant and will be reported alongside the respective effect sizes (partial eta-square; small effect: ${\eta}_p^2$ = 0.02, medium effect: ${\eta}_p^2$ = 0.13, large effect: ${\eta}_p^2$ = 0.26; Cohen, 1988).

Time windows for the analyses of ERPs were defined using the approach of collapsed localizers (see Luck & Gaspelin, 2017). Lateral parieto-occipital P1 peaked at 108 ms in the grand–grand averaged ERP. Similar to Gibbons et al. (2023), a symmetrical 20-ms interval around the peak was chosen to determine P1 amplitude (100–118 ms). The left P1 cluster included electrodes PO7 and O1, with PO8 and O2 forming the right cluster. N170 peaked at 166 ms in the grand–grand averaged ERP. It was quantified as the mean amplitude in a 30-ms time window (150–180 ms) around the peak at symmetrical parieto-occipital clusters where the component reached its maximum (left: P7, PO7, O1; right: P8, PO8, O2; identical to Gibbons et al., 2023). In the averaged-referenced ERPs, an EPN-like complex was assessed as the mean amplitude between 150 and 270 ms at left (P7, PO7, O1) and right parieto-occipital clusters (P8, PO8, O2). Interestingly, this EPN-like deflection started earlier than usual (around 200 ms for word-related EPN; see Kissler et al., 2007) and extended over the entire 150–270 ms time window. Thus, instead of artificially separating the average-referenced ERP into different components in this time range, we decided to analyze the mean amplitude of the entire complex (also seeing that the conditions behaved very similarly in the ERP; Figures 3-6). No other EPN-like complex was observed beyond 270 ms. Previously, this complex has similarly been observed by Gibbons, Schmuck, and Kirsten (2022) and is thus referred to as N170/EPN (complex) in the following, due to its early onset.

3 RESULTS

3.1 Behavioral results

3.1.1 RT

The ANOVA with repeated measures on valence, arousal, word length, and word frequency yielded significant main effects for all four factors. Mean hit RT was significantly faster for positive (623 ms) compared with negative target words (638 ms), F(1,57) = 7.29, p = .009, ${\eta}_p^2$ = 0.11, as well as for high-arousal (627 ms) opposed to low-arousal target words (634 ms), F(1,57) = 8.36, p = .005, ${\eta}_p^2$ = 0.13. Additionally, shorter hit RTs were observed for short words (622 ms) compared with long words (638 ms), F(1,57) = 20.84, p < .001, ${\eta}_p^2$ = 0.27 and for frequent (614 ms) compared with rare words (646 ms), F(1,57) = 95.81, p < .001, ${\eta}_p^2$ = 0.63. The analysis revealed a significant two-way interaction between valence and arousal, F(1,57) = 20.44, p < .001, ${\eta}_p^2$ = 0.26. which was further qualified by the three-way interaction valence × arousal × length, F(1,57) = 7.84, p = .007, ${\eta}_p^2$ = 0.12. Investigating the levels of length separately revealed that the significant arousal × valence interaction was restricted to long words, F(1,57) = 22.55, p < .001, ${\eta}_p^2$ = 0.28, and was not observed in short words, F(1,57) = 2.17, p = .147. Orthogonal comparisons only among long words revealed that responses to long PH words were significantly faster than to long PL and NH words, both t(57) ≥ 6.23, p_c < .001, while the other two comparisons were not significant, t(57) ≤ 0.58, p_c > .999.

3.1.2 Errors

The five-way rm ANOVA for errors, now including the additional factor status, revealed significant main effects of frequency, F(1,57) = 35.99, p < .001, ${\eta}_p^2$ = 0.39, and status, F(1,57) = 12.57, p < .001, ${\eta}_p^2$ = 0.18. Low-frequency words led to more errors (3.0%) than high-frequency words (1.5%) while the miss rate on target trials (1.4%) was lower than the false alarm rate on nontarget trials (3.1%). For logical reasons and because of several higher order interactions involving status, misses, and false alarms were analyzed separately.

Misses

The only significant main effect in the four-way ANOVA indicated more missing responses for rare (1.9%) than for frequent target words (0.8%), F(1,57) = 14.89, p < .001, ${\eta}_p^2$ = 0.21. There was a significant valence × arousal interaction, F(1,57) = 7.49, p = .008, ${\eta}_p^2$ = 0.12, however, all four follow-up corrected pairwise orthogonal contrasts were not significant, t(57) ≤ 2.03, p_c ≥ .187.

False alarms

More errors were made for rare (4.0%) than for frequent nontarget words (2.2%), F(1,57) = 20.44, p < .001, ${\eta}_p^2$ = 0.26. There was a significant interaction between valence and arousal, F(1,57) = 6.29, p = .015, ${\eta}_p^2$ = 0.10. It was further qualified by a three-way interaction including length, F(1,57) = 6.06, p = .017, ${\eta}_p^2$ = 0.10, and finally, by a four-way interaction including frequency, F(1,57) = 4.68, p = .035, ${\eta}_p^2$ = 0.08. Inspection of the valence × arousal interaction separately for the four length × frequency combinations revealed that the significant term specifically resulted from the combination of long, rare nontarget words. In particular, long, rare PL nontargets produced significantly more false alarms (7.7%) compared with long, rare NL nontargets (2.3%), t(57) = 4.29, p_c < .001, as well as compared with long, rare PH nontargets (1.6%), t(57) = 5.16, p_c < .001, and short, rare PL nontargets (3.7%), t(57) = 3.30, p_c = .017.

3.2 ERP results

3.2.1 Mastoid-referenced P1 (100–118 ms)

The omnibus six-way rm ANOVA for P1 indicated neither a significant main effect of laterality, F(1,57) = 3.02, p = .088, nor any interaction including laterality and valence, F(1,57) < 1.77, p > .189; therefore the left and right cluster were aggregated. No main effects were found in the resulting five-way ANOVA, F(1,57) < 1.82, p > .180. The predicted interaction between status and valence was indeed observed, F(1,57) = 8.08, p = .006, ${\eta}_p^2$ = 0.12. P1 amplitude was significantly increased for negative words compared with positive words in the target condition, t(57) = 3.16, p_c = .005, d = 0.42, unlike for nontarget words, t(57) = 0.81, p_c = .421. The two-way interaction was further qualified by a three-way interaction including length, F(1,57) = 4.66, p = .035, ${\eta}_p^2$ = 0.08. Scrutiny of this interaction revealed that the term status × valence was significant in short words, F(1,57) = 10.75, p = .002, ${\eta}_p^2$ = 0.16, but not in long words, F(1,57) = 0.07, p = .793 (see Figure 1, panel A). The difference in P1 amplitude between negative and positive words was significant in short target words, t(57) = 4.16, p_c < .001, d = .55, as opposed to the difference in short nontarget words, t(57) = −0.82, p_c = .413.

Details are in the caption following the image — **FIGURE 1**
Open in figure viewer PowerPoint

Interactive effect of target status and valence on P1 amplitude (100–118 ms): influences of word Length, Arousal, and Word Frequency. P1 (100–118 ms) is measured at a lateral posterior cluster (PO7, PO8, O1, O2). Averaged ERPs are shown as a function of status and valence separated by word length (panel A), arousal (panel B) and word frequency (panel C). Colored maps in panels A to C show the scalp distribution of the negative-positive amplitude difference in target words during the P1 time window. *p < .05; **p < .01, for the status × valence interaction in the respective condition.

Even though there was no modulation of the status × valence interaction by word frequency, in line with our fourth hypothesis, we conducted ANOVAs separately for high- and low-frequency words. While the interaction between status and valence was significant in high-frequency words, F(1,57) = 6.54, p = .013, ${\eta}_p^2$ = 0.10, this was not true for low-frequency words, F(1,57) = 1.50, p = .227 (see Figure 1, panel C). Specifically for high-frequency target words, we found an increased P1 amplitude for negative compared with positive words, t(57) = 3.36, p_c = .006, but not in any other combination of frequency and status, t(57) < 1.64, p_c > .319. In addition, a significant interaction between status, valence, and arousal was observed, F(1,57) = 4.27, p = .043, ${\eta}_p^2$ = 0.07. Subsequent analyses revealed that the status × valence interaction was significant in low-arousal words, F(1,57) = 10.47, p = .002, ${\eta}_p^2$ = 0.16, but not in high-arousal words, F(1,57) = 0.84, p = .364 (see Figure 1, panel B). Specifically for low-arousal words, mean P1 amplitude was larger for negative compared with positive words in the target condition, t(57) = 3.31, p_c = .006, d = 0.44, while this pattern was not present in nontarget words, t(57) = 1.62, p_c = .330. No such pattern was found in high-arousal target and nontarget words, t(57) < 1.37, p_c > .352.

Finally, the analysis yielded an interaction between valence, arousal, length, and frequency, F(1,57) = 6.67, p = .012, ${\eta}_p^2$ = 0.11. However, a careful inspection of this effect in the ERP figure revealed a noticeable difference in the ERP immediately after word onset in the time range of 0 to 60 ms. When we controlled for the mean voltage in this time window in the five-way ANOVA (see above) by adding it as covariate, the interaction between valence, arousal, length, and frequency in the P1 time window was no longer significant, F(1,57) = 2.27, p = .132, and was therefore not followed up further.

3.2.2 Mastoid-referenced N170 (150–180 ms)

The only significant main effect indicated larger N170 mean amplitude over the left cluster compared with the right cluster, F(1,57) = 29.62, p < .001, ${\eta}_p^2$ = 0.34. Furthermore, we observed a significant three-way interaction between valence, arousal, and frequency, F(1,57) = 4.41, p = .040, ${\eta}_p^2$ = 0.07.

Investigating the ESM-relevant valence × arousal term separately for the levels of frequency revealed that the interaction was not significant in high-frequency words, F(1,57) = 0.68, p = .413, as opposed to the low-frequency condition, F(1,57) = 4.79, p = .033, ${\eta}_p^2$ = 0.08. However, this latter two-way interaction was further modulated by laterality, F(1,57) = 4.18, p = .046, ${\eta}_p^2$ = 0.07. Separate analyses for each cluster revealed that the valence × arousal interaction for low-frequency words was only significant for the left, F(1,57) = 6.83, p = .011, ${\eta}_p^2$ = 0.11, but not for the right hemisphere, F(1,57) = 2.26, p = .139 (see Figure 2). Follow-up pairwise tests only for low-frequency words in the left cluster, where N170 was clearly most pronounced (see above), indicated that both the comparison between negative and positive words in high-arousal words, t(57) = −2.21, p_c = .062, as well as in low-arousal words fell short of significance, t(57) = 1.87, p_c = .066. When the low-frequency congruent PL and NH words were aggregated and compared with the incongruent words in the left hemisphere, we observed a larger N170 for congruent words, t(57) = 2.61, p_c = .023, d = .34. This pattern was not observed in the right hemisphere, t(57) = 1.50, p_c = .277.

3.2.3 Average-referenced N170/EPN (150–270 ms)

For the average-referenced N170/EPN complex, we observed the main effects of laterality, valence, and length. Similar to mastoid-referenced N170, the mean amplitude was larger over the left than the right hemisphere, F(1,57) = 37.38, p < .001, ${\eta}_p^2$ = 0.08, for negative compared with positive words, F(1,57) = 9.02, p = .004, ${\eta}_p^2$ = 0.14, and for short words compared with long words, F(1,57) = 67.17, p < .001, ${\eta}_p^2$ = 0.54. As laterality did not modulate our effects of interest, we continued with a five-way ANOVA. There was a significant two-way interaction between valence and arousal, F(1,57) = 7.64, p < .008, ${\eta}_p^2$ = 0.12. Mean amplitude was larger for NH compared with PH words, t(57) = 4.18, p_c < .001, d = 0.55, with no such difference for low-arousal words, t(57) = 0.02, p_c = .987 (see Figure 3).

In addition, the analysis indicated a significant three-way interaction between status, valence, and frequency, F(1,57) = 5.76, p = .020, ${\eta}_p^2$ = 0.09. This interaction was best examined separately by status, which yielded a significant interaction term valence × frequency for target words, F(1,57) = 10.54, p = .002, ${\eta}_p^2$ = 0.16, but not for nontargets, F(1,57) = 0.001, p = .974. Specifically for low-frequency target words, N170/EPN was increased for negative compared with positive words, t(57) = 2.85, p_c = .024, d = 0.37, a pattern which was not found for the other three combinations of status and frequency, t(57) < 1.11, p_c > .589 (see Figure 4).

Moreover, we found a significant three-way interaction between status, valence, and length, F(1,57) = 4.70, p = .034, ${\eta}_p^2$ = 0.08. Again, the interaction was best analyzed separately by status, indicating a significant two-way interaction between valence and length for target words, F(1,57) = 6.74, p = .012, ${\eta}_p^2$ = 0.11, as opposed to nontargets, F(1,57) = 0.01, p = .933. Orthogonal follow-up comparisons in target words revealed that short, negative words in particular had a larger amplitude when compared with long, negative, t(57) = 6.43, p_c < .001, d = 0.84, or short, positive words, t(57) = 2.30, p_c = .0497, d = 0.30 (see Figure 5). The three-way interaction was further modulated by the words' arousal, F(1,57) = 4.19, p = .045, ${\eta}_p^2$ = 0.07. Hence, the above pattern only held for the high-arousal condition in which short negative target words had a larger amplitude than short positive target words, t(57) = 3.06, p_c = .007, d = 0.40, while there was no such difference in low-arousal words, t(57) = 0.81, p_c = .420 (see Figure 6).

4 DISCUSSION

The present EEG study focused on the effects of selective attention on the early neural processing of valent word forms. Proceeding from the idea that early emotion effects rely on associations between word form and valence (hypothesis of valent word forms, see e.g., Gibbons et al., 2023; Hinojosa et al., 2020; Kissler et al., 2006), and from the concept of attentional tuning (e.g., Battistoni et al., 2017; Maunsell & Treue, 2006), we investigated whether selective attention to valence would facilitate the processing of visual word forms and whether this was influenced by lexical and affective word properties. Therefore, a valence-detection task under speed instructions requiring participants to selectively attend to one specific valence level of visual words was employed. In contrast to previous research (Gibbons et al., 2023; Schindler & Kissler, 2016) and to facilitate affective discrimination, the task was simplified to a two-level valence-detection task which only included emotional words (positive and negative) and no neutral words. The main finding regarding the concept of valent word forms was a greater valence discrimination in target compared with nontarget words at the lateral-posterior P1. As expected, this effect was modulated by word length and frequency, yielding the strongest valence discrimination in short and frequent target words, respectively. With relevance for the framework of valence–arousal conflict theory, which has a focus on valence × arousal interactions (see Robinson et al., 2004), N170 was larger for negative, high-arousal and positive, low-arousal words (NH and PL) as opposed to negative, low-arousal and positive, high-arousal words (NL and PH); this pattern was restricted to rare words. In addition, EPN was enhanced for NH compared with PH words while NL and PL words did not differ.

4.1 Behavioral results

The present study employed a two-level valence-detection task to maximize affective discrimination, simplifying the design of our earlier study (Gibbons et al., 2023). In comparison to the task including neutral words, participants here responded on average almost 150 ms faster and committed substantially fewer errors. Even though cross-study comparisons have to be interpreted with caution, this suggests that our task successfully facilitated the discrimination of the affective words. The more unambiguous discrimination should further improve attentional tuning processes compared with the three-level valence-detection task (see Introduction). Going beyond the mere comparison, our analysis revealed significant effects of both affective and lexical properties of words on RT. Participants responded faster to positive compared with negative words which is in line with a presumptive processing advantage for positive words in implicit and explicit word processing studies (Kauschke et al., 2019) and with the previous study by Gibbons et al. (2023). Independently, we found faster RT for high-arousal words (see also Estes & Adelman, 2008; but for a reverse pattern, see Kuperman et al., 2014), even though the actual decrease in RT was small. In contrast, large effects were revealed for the lexical factors of word length and frequency. Our results yielded faster responses to short compared with long words (~6 vs. 9 letters in our study, similar to New et al., 2006), but the greatest speed increase was found for frequent words. In line with the large effect, word frequency is among the strongest predictors of processing efficiency when performing word recognition tasks (Brysbaert et al., 2018). Indeed, participants in the present study had to recognize the word before they could decide whether it matched the target valence, a process that should be facilitated for high-frequency words.

Results further revealed an interaction between valence and arousal; unexpectedly, this was due to significantly faster responses specifically to PH words. In contrast, previous studies (Citron et al., 2014; Robinson et al., 2004; Yao et al., 2016) reported faster responses to congruent (PL and NH) than incongruent affective words (PH and NL), in line with the valence–arousal conflict theory (Robinson et al., 2004). Of note, this pattern has previously not been observed in all studies using LDT or affective categorization, in which independent advantages for positive and highly arousing words have occasionally been reported (Kever et al., 2019; Recio et al., 2014). In our study, the valence × arousal interaction might be explained by the task requirements. While the preattentive processing bias toward congruent compared with incongruent words might be strongest in tasks that do not require explicit emotional processing (e.g., LDT), selective attention and top-down tuning during our explicit task under speed instructions might have overridden the existing bias, at least its effects on response times. What is more, due to the strong top-down tuning toward valence, processing advantages for positive valence (and high arousal, Kever et al., 2019; Recio et al., 2014) might have been reinforced. It is likely that this does not apply to short words where detecting the target valence is already relatively fast (ceiling effect), but is especially advantageous in long words which take longer to encode (Oganian et al., 2016).

Regarding errors, there was a general bias in favor of responding, indicated by more false alarms in nontarget trials than misses in target trials. Given the easiness of the task, participants likely had a strong tendency toward a (speeded) response, thus leading to more false alarms. This is contrary to results from Gibbons et al. (2023) but can be explained by the lack of neutral words in the current experiment which particularly contributed to the higher miss rate in the previous study. Similar to RT, the predicted interaction between valence and arousal on errors was not observed, but rather the opposing pattern emerged. The congruent PL and NH words tended to elicit more misses and false alarms than PH and NL words. While it is difficult to interpret this finding within the framework of the valence–arousal conflict theory, methodological reasons might explain the observed pattern. Because our preselected pool of words showed the typical negative > positive arousal effect, we had to selectively remove NH as well as PL words during the selection of the final word sets, to match negative and positive words in terms of their arousal. Consequently, several words that are prototypical for their respective category were excluded, thus potentially causing participants to commit more misses and false alarms in response to the therefore less prototypical categories NH and PL (the congruent categories according to valence–arousal conflict theory). For false alarms, the interaction was further modulated by word length and frequency. In particular, long PL words led to more false alarms only in the rare condition. Again, this can be attributed to the selection of words as almost all errors were related to two less prototypical, rather rare PL words (begnadet [exceptionally gifted], verlässlich [reliable]). Considering the low number of total errors, the findings should be treated cautiously and warrant further research including different word stimuli.

4.2 Event-related potentials

4.2.1 P1 and early valence discrimination

For the lateral parieto-occipital P1 (100–118 ms), the expected greater valence discrimination in target than in nontarget words was confirmed. In particular, a significantly larger amplitude was found for negative compared with positive target words, while there was no difference in nontarget words. These results conceptually support previous findings of (target) status-dependent P1 valence discrimination from Gibbons et al. (2023). Furthermore, the observed valence discrimination in target words was modulated by lexical and affective word properties. In accordance with our fifth hypothesis, the effect emerged specifically in short words, but was not found in long words. Similarly, the status-dependent valence discrimination was only observed in high-frequency words, but not in low-frequency words. However, this effect was not significantly stronger in high-frequency words in comparison to low-frequency words.

So far, research has focused both on independent effects of lexical properties (word frequency and length, e.g., Hauk et al., 2006; Penolazzi et al., 2007) and interaction effects between lexical and affective word properties (e.g., word frequency and valence, Scott et al., 2009) on early word processing. Our study, however, is the first to test the influence of lexical properties within the context of the hypothesis of valent word forms (Bayer et al., 2019; see Gibbons et al., 2023; Keuper et al., 2014; Kissler et al., 2006). It states that early emotion effects prior to lexico-semantic processing are based on associative learning mechanisms, that is, the repeated linking of visual word forms with their valence. Consequently, selectively attending to a specific valence should not only preactivate neural representations of this valence (feature-based attention, e.g., Battistoni et al., 2017) but also spread to representations of associated word forms. Thus, selective attention (i.e., target status) should facilitate the early perceptual integration of presented words and modulate early ERP markers of word form processing. This was supported by a stronger P1 valence discrimination for target words.

Even more, this effect was particularly pronounced when visual forms and their valence were strongly associated, as we assumed it to be the case for short and high-frequency words (see Introduction). On the one hand, this provides further insights into the mechanisms behind early emotion effects. In the current experiment, the status-dependent valence effect interacted with lexical properties such as frequency and length which are themselves thought to operate on associative learning mechanisms (Brysbaert et al., 2018). The observed interaction therefore supports the notion that early emotion effects also involve associative learning mechanisms (see Hinojosa et al., 2020). On the other hand, it implies that future research should specifically take into account the influence of lexical properties when early emotion effects prior to lexico-semantic processing at the P1 amplitude are of interest.

In the present study, P1 was largest for negative (target) words and smallest for positive (target) words. Consistent with this finding, previous ERP studies on emotional words found an enhanced P1 amplitude for negative compared with positive words (Hofmann et al., 2009; Schindler et al., 2019; Zhang et al., 2014). Nevertheless, decreased amplitudes for negative in comparison to neutral words (Kissler & Herbert, 2013; Kuchinke et al., 2014; Scott et al., 2009) or positive words (Bayer et al., 2012) were also reported. It is difficult to directly compare the previous results to our findings due to the large differences between studies and our specific focus on selective attention to valence. However, in a highly similar study with a three-level valence-detection task (positive, neutral, negative), Gibbons et al. (2023) observed the opposite pattern, that is, larger P1 amplitude for positive than negative words. They suggested that to successfully distinguish positive and neutral words, participants might have particularly focused on the detection of positive valence, which then determined P1 amplitude (Gibbons et al., 2023).

Importantly, given that neutral words are not present in the current two-level task (positive, negative), there should no longer be any perceived difficulty in the distinction of words and participants might have adopted a different strategy. Consistent with a negativity bias in emotional word processing (Zhang et al., 2014; for review, see Norris, 2021), it could involve a stronger or dominant focus on negative valence. Even when (selectively) attending to positive valence, this might (at least partially) result in searching for the absence of any negative-related (or threat-related, Zhang et al., 2014) information. Thus, P1 could reflect the extent to which the presented word matches the pre-activated (and negatively biased) target template. This would nicely explain the larger amplitudes for negative target words that matched the template particularly well. In contrast, Gibbons et al. (2023) previously suggested that the perceived difficulty in the three-level detection task and the subsequent focus on positive valence modulated the P1 in terms of a greater amplitude for positive target words. These combined findings point to a crucial role of the attentional set (based on the overt task as well as implicit translations of this task into an efficient way to discriminate among the given set of categories) in early emotional word processing (see also Gibbons, Schmuck, & Kirsten, 2022, for further considerations on the concept of an “attentional set” in a valence-detection task using words, and for related behavioral and ERP effects).

Regarding the influence of affective factors, we observed greater P1 valence discrimination in low-arousal target words which was driven by strongly decreased amplitude for PL target words. This is in contrast to our hypothesis, which referred to a stronger status-dependent valence effect in the high-arousal condition. Therefore, it appears that the preactivation/tuning of word forms is not facilitated for high-arousal valent words, whose processing thus differs from the supposed mechanism for short and high-frequency valent words. Rather, valence and arousal in the current task might interact in a different way, which calls for alternative explanations. While the two dimensions are often conceptualized as independent (e.g., Russell & Barrett, 1999), it has been observed that processing of positive low-arousal stimuli and negative high-arousal stimuli at preattentive stages is facilitated (valence–arousal conflict theory, Robinson, 1998; Robinson et al., 2004). A very similar idea is expressed by ESM (e.g., Cacioppo et al., 1997) which assumes that negative stimuli are implicitly associated with high arousal and positive stimuli with low arousal. This association would explain the strongly decreased amplitude for PL target words. If there is indeed a predominant focus on negative valence in the current task (i.e., participants mainly try to discriminate negative from non-negative words), and if P1 reflects the extent to which the current word and a negative attentional template match, PL target words would be particularly easy to classify as non-negative. That is because they mismatch the negative attentional set, which implies an aspect of high arousal, in terms of both valence and arousal (see Gibbons, Schmuck, & Kirsten, 2022).

We have to note that ERP studies using a similar (three-level valence) paradigm with emotional words (Schindler & Kissler, 2016), pictures (Schindler & Straube, 2020; Schupp et al., 2007) or naturalistic faces (Schmuck et al., 2023) have not reported any P1 emotion–attention interactions. Apart from the exclusion of neutral words in our study, one reason could be the different stimulus categories (pictures vs. faces vs. words). For example, in contrast to words, pictures lack “clearly defined and with regard to identity and exact location predictable perceptual elements” (Gibbons et al., 2023) which are linked to a specific valence. Individual faces, in contrast, possess these predictable visual elements (e.g., mouth); nevertheless, they might vary too much in these elements and thus be less pre-activatable (Schmuck et al., 2023). Furthermore, different categories of words (nouns vs. adjectives, Herbert et al., 2008; Palazova et al., 2011) or variations in experimental tasks (e.g., pressing a button vs. counting) could have contributed to diverging results in the tasks using words.

4.2.2 N170 and valence × arousal congruency

Regarding N170, we observed an interaction between valence and arousal which was restricted to low-frequency words. PL and NH words combined, that is, congruent stimuli according to the valence–arousal conflict theory (Robinson et al., 2004), elicited a larger amplitude compared with incongruent (PH and NL) words. This pattern (congruent > incongruent) is consistent with a previous finding from Yao et al. (2016) during a lexical decision task, although they did not separate words according to frequency. Notably, we found no indication that the effect was modulated by (target) status. Indeed, Robinson et al. (2004) postulated that the evaluation and integration of valence and arousal are performed at preattentive levels which are primarily related to bottom-up processes. As such, they should be independent of selective attention and thus of any top-down influences in our task.

Importantly, in high-frequency words, the integration of incongruent combinations of valence and arousal might be strongly overlearned, due to the frequent occurrence of these stimuli in general, and thus resemble the integration of congruent words. Therefore, during N170 both congruent and incongruent high-frequency words are processed highly automatically with no additional processing benefit for either class of words. For low-frequency words, however, this integration is not overlearned and valence and arousal need to be integrated prior to further semantic processing. Consequently, a reduced N170 amplitude for incongruent words given their more difficult (and delayed) integration of valence and arousal was found. This congruency-related effect was only observed in the left hemisphere, in line with the overall larger N170 amplitude in the left cluster. This fits nicely with Dien (2009), who characterizes the word-related N170 as being left-lateralized and clearly distinguishes it from the right-lateralized face N170. Our findings further indicate that the integration of affective word properties can be located to the left hemisphere.

Finally, in contrast to our P1 findings, we did not find any modulation of the N170 component by (target) status. Gibbons et al. (2023) suggested that N170 might reflect the facilitated processing of affective target stimuli and the inhibition of affective nontarget stimuli. In particular, affective words elicited larger N170 amplitudes than neutral words under conditions of selective attention, while the pattern was reversed when words were not selectively attended and actually had to be ignored. These arousal-driven (emotional vs. neutral) effects, however, do not contradict the current results. As a general finding, N170 is reliably increased for a variety of negative and/or positive stimuli compared with neutral stimuli, thus reflecting its sensitivity to affective content. This can be observed both for the word N170 (e.g., Gibbons et al., 2023; Hofmann et al., 2009; Kissler & Herbert, 2013; Scott et al., 2009; Zhang et al., 2014) and the face N170 (Rellecke et al., 2012; Schindler & Bublatzky, 2020; Schmuck et al., 2023). It was further proposed that the word N170 could not differentiate between positive and negative affect of words when arousal is matched (Zhang et al., 2014). Since the two-level valence-detection task only includes positive and negative words that are overall matched on arousal (see Method), this would explain why there was neither a facilitation of affective target nor an inhibition of affective nontarget words. Interestingly, overall N170 tended to be larger for negative than positive words. This is consistent with a stronger focus toward negative affect in the present task but has to be regarded with caution given its non-significance.

4.2.3 N170/EPN and greater attention toward negative words

Following the mastoid-referenced N170 congruency effect, we observed a lateral-posterior negativity in average-referenced ERPs which started at 150 ms and resembled the EPN component. In comparison to the usual EPN time window in words (starting at around 200 ms, e.g., Bayer & Schacht, 2014; Kissler et al., 2007), the component onset in the current study was relatively early (but see Gibbons, Schmuck, & Kirsten, 2022). Importantly, the observed effects extended in a similar way throughout the entire 150–270 ms time window, which suggested that the complex was better analyzed as a whole. Hence, similarly to Gibbons, Schmuck, & Kirsten, 2022, we refer to the observed negativity as N170/EPN complex to emphasize the early onset during the N170 time window and to distinguish it from the typical EPN component.3 As expected, there was an interaction between valence and arousal which was characterized by larger negativity for congruent words than incongruent words. Unlike the disordinal interaction at N170, the modulation was driven by the high-arousal condition for which amplitudes were larger for negative than positive words. However, similar to processing during the N170, the effect was independent of status.

It would therefore stand to reason that high arousal facilitates valence discrimination (positive vs. negative) in a bottom-up fashion (for stronger valence effects specifically in high-arousal words, see, e.g., Gibbons, 2009). This idea is conceptually in line with findings that show an increased performance in discrimination tasks following arousing stimuli (Phelps et al., 2006) or under a state of increased arousal (Gelbard-Sagiv et al., 2018). Consistent with the general focus on negative valence in the current task, negative words in particular attracted more N170/EPN-associated processing resources. One may ask whether this reasoning contradicts our interpretation of the P1 finding, according to which the rejection of low-arousal positive target words as non-negative (which is ultimately also valence discrimination) should be particularly easy, as seen in the particularly small P1. However, in the case of P1, this is not the (bottom-up, pure) valence discrimination as described here. We rather suggested that it represents an initial mismatch of PL words with the presumably dominant negative attentional set (including the aspect of high arousal) and therefore a top-down driven process. This idea again implies that top-down effects of affect can have earlier effects on word-related ERPs than affect-related bottom-up processes, which of course requires further research.

While any valence × arousal interaction during the N170/EPN was independent of selective attention (similar to traditional N170, see above), valence and lexical features interacted only in target words. More specifically, there was an amplitude increase for negative compared with positive words that was limited to low-frequency targets, with a particularly large amplitude for negative, short target words. These findings go beyond the pure, bottom-up valence discrimination in high-arousal words, but indicate that the attentional set plays an important role. The restriction of valence effects (here: negative > positive) to target words is in accordance with the assumption that there is a focus on negative affect characterized by searching for negative affect in “attend to negative” trials and for the absence of negative affect in “attend to positive” trials. This idea is supported by the repeatedly observed increase of mean amplitudes for negative words at all three early components (P1, N170, N170/EPN).

However, the specific influence of lexical features requires a more detailed interpretation. First, the target valence discrimination (negative > positive) was only present in low-frequency words. One might assume that the highly automatic (and overlearned) processing of high-frequency words independent of their valence precludes any such discrimination. Only for low-frequency words, though, negative words might benefit more than positive words from the assumed focus on negative affect. Second, short negative target words elicited the largest mean N170/EPN amplitude. This can be interpreted as a result of the facilitated attentional tuning for short target words (see P1 discussion) and the stronger focus toward negative affect. In consequence, short negative target words which already had a processing advantage at P1 continue to be more intensively processed and/or attract greater attentional resources. Our analysis further revealed that this is particularly true for those high in arousal. Drawing back on our previous reasoning, we assumed that high arousal promotes the valence discrimination in words. When this bottom-up processing advantage for negative words is combined with the facilitated attentional tuning to short words, a maximal attentional capture, as indicated by the N170/EPN complex, by these high-arousal, negative, short target words seems highly plausible.

The question remains why we did not observe an independent effect of arousal on N170/EPN amplitude. First, the word-related EPN is usually observed and assessed between around 200 and 320 ms (Herbert et al., 2008; Kissler et al., 2007; e.g. Schacht & Sommer, 2009b). However, in the current study, the EPN-like complex emerged already at 150 ms, thus differing from previous experiments. Second, and more importantly, effects of arousal or emotion are typically driven by an increase for emotional, that is positive and negative, compared with neutral words (e.g., Citron et al., 2013; Gibbons, Kirsten, & Seib-Pfeifer, 2022; Kissler et al., 2007; Schacht & Sommer, 2009b) and less often by high-arousal compared with low-arousal words independent of their valence (Bayer et al., 2012). Therefore, the use of only positive and negative words might have made it more difficult to detect arousal-related effects given the more fine-grated differences in arousal. This might have been further complicated by the already mentioned methodological reasons, which included the removal of high-arousal negative words and low-arousal positive words and therefore reduced the variance of arousal scores. Third, instead of a mere N170/EPN increase for high-arousal words, an interaction between valence and arousal was observed. In addition to our previous interpretations, the finding of increased amplitudes for congruent words compared with incongruent words requires a more differentiated approach drawing on the valence–arousal conflict theory (Citron et al., 2013; see our results on N170 and Robinson et al., 2004) and warranting further research.

Additionally, and unrelated to the focus of the study, we observed a greater N170/EPN for short words compared with long words, which is consistent with prior results in this time range (Hauk & Pulvermüller, 2004). Unlike the more exogenous and perceptually driven P1 component that had a marginally higher amplitude for long words and the only marginal increase for short words at N170, the EPN reflects the analysis of more complex word information, which becomes increasingly difficult for longer words.

4.3 Conclusions, limitations, and future prospects

The present study provides new evidence for the hypothesis of valent words (Gibbons et al., 2023) and further explored early interactions between valence and arousal within the framework of valence–arousal conflict theory (Robinson et al., 2004). Findings revealed that at very early perceptual processing stages, task-induced selective attention to positive and negative valence facilitated discrimination of affective word stimuli. Importantly and predicted by theoretical considerations, the better P1 valence discrimination for target compared with nontarget words benefited from low word length and high word frequency. One might therefore suggest that P1 represents an early first match/mismatch with the task-induced attentional set and, consequently, pre-activated word forms. For the following N170, Gibbons et al. (2023) previously argued that it might indicate facilitated processing of affective target words and inhibition of affective nontarget words, compared with their neutral counterparts. Our results do not contradict this idea given the absence of neutral words in the task, which precludes any comparison between affective and neutral words. What is more, when only affective words are included, word N170 seems to indicate the automatic integration of affective stimuli features, that is, valence and arousal. The independence of target status further supports the assumed bottom-up, preattentive nature of this process (see Robinson et al., 2004). Conceptually, the automatic integration of affective features as well as the trend toward larger amplitude for short words fits nicely with the broader literature of N170 as a marker of the holistic processing of visually highly structured stimuli (N170 for faces: Eimer et al., 2011; Sagiv & Bentin, 2001; N170 for words: Simon et al., 2007; Dien, 2009). Finally, the increased average-referenced N170/EPN complex for negative compared with positive words was restricted to high-arousal rather than low-arousal words. This might indicate the earliest evidence of bottom-up (i.e., target status independent) valence discrimination which during early processing stages still relies on the greater automatic attraction of attentional resources by high-arousal words.

Highly interesting and unlike the previous study by Gibbons et al. (2023), the highest P1 amplitudes were observed for negative target words and the lowest for positive nontarget words. We suggest that this pattern may be due to a task-adapted processing strategy by the participants. In the absence of neutral words, there was no longer any perceived difficulty related to the distinction between positive and neutral words (see Gibbons et al., 2023). In contrast, participants might have searched for negative affect (in the attend-negative) or the absence of negative affect (in the attend-positive block) in the sense of a negativity bias (see Norris, 2021). Findings of a marginally larger N170 to negative affect and a greater N170/EPN particularly to negative target words also tentatively support the focus on negative affect. Summarizing our findings on P1 tuning, we should emphasize that these considerations related to the affective task context clarify that the present study with two levels of affect, reporting negative > positive P1 specifically for target words, does not contradict our prior study (Gibbons et al., 2023) with three levels of affect, reporting positive > negative P1 specifically for target words. Instead, we propose that both studies converge in showing that P1 valence discrimination in valence detection is increased for target compared with nontarget words.

Limitations of our study include the fact that mostly female undergraduate psychology students participated. Thus, future research should also include more male participants to make findings more generalizable and to be able to detect potential sex differences in emotional word processing (for a general overview of electrophysiological sex differences in language processing, see Ramos-Loyo et al., 2022). Additionally, we only used emotional adjectives in our task. Similar findings did previously emerge in a different word stimulus set of adjectives (see Gibbons et al., 2023), but it will be interesting to see whether these results can also be transferred to other word types such as nouns or verbs. For example, ERP differences were found between different types of words (see Palazova et al., 2011) and might partially explain diverging findings in the valence-detection task (see Schindler & Kissler, 2016). Another limitation regarding the word set is that we had to selectively remove NH as well as PL words during stimulus selection. This was necessary to match the valent words according to their arousal and exclude any potential confound with other word parameters. While the observed effects can thus be attributed with more certainty to either lexical or affective word properties, one might only speculate whether the inclusion of these highly prototypical, congruent NH and PL words might have resulted in stronger congruency effects. Previous studies, for example, have revealed the existence of such effects during the EPN (Citron et al., 2013; Recio et al., 2014). Furthermore, our final stimulus selection had to be adjusted after data collection given the high number of committed errors resulting from three words (see Methods). Even though this did not change the results as shown by control analyses with all 144 words, future word stimuli should be selected with great care in order to avoid any unintended consequences for the task. Additionally, linear mixed models could be used to account for the (random) effects of individual words and to investigate word features as continuous rather than categorical variables (thus avoiding to dichotomize word features by applying a median split). Finally, we note that the assessment of the EPN (complex) which is usually observed over occipito-temporal sites in average-referenced data (see Junghöfer et al., 2006) was not optimal. Even though it is also typically observed at the same electrodes which we analyzed in the current study, it would benefit from sampling of extremely lateral-posterior electrodes P9, P10, PO9, and PO10 not covered by our EEG setup (for a comparison, see Bayer et al., 2012; Kissler et al., 2009; Palazova et al., 2011).

While three-level valence-detection tasks have been employed using different stimuli including images (Schindler & Straube, 2020; Schupp et al., 2007) and faces (Schmuck et al., 2023), it might be interesting to see if the absence of neutral images or faces would also result in different stimulus processing. For faces specifically, attentional tuning and valence–arousal interactions could additionally be explored using various primary emotions, focusing for example on the less commonly studied emotions. Beyond merely extending the current task to other stimuli categories such as images or faces, future studies might try to manipulate specific stimuli properties in order to investigate early ERP effects of attentional tuning. Conceptually similar to the effects of word frequency in the present study, researchers could present familiar and unfamiliar faces when using face stimuli in such detection tasks. Given that effects of face familiarity have been reported for early perceptual components such as P1 and N170 (Huang et al., 2017; Jemel et al., 2003), an intriguing hypothesis might be that attentional tuning works specifically well for familiar faces, which should be reflected in these early ERP components. Finally, apart from psychology students, future studies could include specific population samples. It might be highly interesting to explore whether depressed individuals, whose attention is often biased toward negative affect (Peckham et al., 2010), show an even stronger focus to negative affect in the current task using words and if this bias still persists if neutral words are included (three-level valence-detection task, Gibbons et al., 2023).

AUTHOR CONTRIBUTIONS

Jonas Schmuck: Data curation; formal analysis; methodology; writing – original draft. Robert Schnuerch: Formal analysis; writing – review and editing. Emely Voltz: Investigation; writing – review and editing. Hannah Kirsten: Methodology; software. Henning Gibbons: Conceptualization; methodology; project administration; resources; supervision; writing – review and editing.

ACKNOWLEDGMENTS

We thank Maria Kist, Melanie Kordel, and Luzie Lerche for assistance with data acquisition.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Open Research

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Supporting Information

REFERENCES

Balota, D. A., Cortese, M. J., Sergent-Marshall, S. D., Spieler, D. H., & Yap, M. J. (2004). Visual word recognition of single-syllable words. Journal of Experimental Psychology: General, 133(2), 283–316. https://doi.org/10.1037/0096-3445.133.2.283
10.1037/0096-3445.133.2.283
PubMed Web of Science® Google Scholar
Banse, R. (1999). Automatic evaluation of self and significant others: Affective priming in close relationships. Journal of Social and Personal Relationships, 16(6), 803–821. https://doi.org/10.1177/0265407599166007
10.1177/0265407599166007
Web of Science® Google Scholar
Barrett, L. F., & Russell, J. A. (1998). Independence and bipolarity in the structure of current affect. Journal of Personality and Social Psychology, 74, 967–984. https://doi.org/10.1037/0022-3514.74.4.967
10.1037/0022-3514.74.4.967
Web of Science® Google Scholar
Battistoni, E., Stein, T., & Peelen, M. V. (2017). Preparatory attention in visual cortex. Annals of the New York Academy of Sciences, 1396(1), 92–107. https://doi.org/10.1111/nyas.13320
10.1111/nyas.13320
PubMed Web of Science® Google Scholar
Bayer, M., & Schacht, A. (2014). Event-related brain responses to emotional words, pictures, and faces—A cross-domain comparison. Frontiers in Psychology, 5, 1106. https://doi.org/10.3389/fpsyg.2014.01106
10.3389/fpsyg.2014.01106
PubMed Web of Science® Google Scholar
Bayer, M., Grass, A., & Schacht, A. (2019). Associated valence impacts early visual processing of letter strings: Evidence from ERPs in a cross-modal learning paradigm. Cognitive, Affective, & Behavioral Neuroscience, 19(1), 98–108. https://doi.org/10.3758/s13415-018-00647-2
10.3758/s13415-018-00647-2
PubMed Web of Science® Google Scholar
Bayer, M., Sommer, W., & Schacht, A. (2012). P1 and beyond: Functional separation of multiple emotion effects in word recognition. Psychophysiology, 49(7), 959–969. https://doi.org/10.1111/j.1469-8986.2012.01381.x
10.1111/j.1469-8986.2012.01381.x
CAS PubMed Web of Science® Google Scholar
Bradley, M. M., & Lang, P. J. (2000). Measuring emotion: Behavior, feeling, and physiology. In Cognitive neuroscience of emotion (pp. 242–276). Oxford University Press.
Google Scholar
Brosch, T., Pourtois, G., & Sander, D. (2010). The perception and categorisation of emotional stimuli: A review. Cognition & Emotion, 24(3), 377–400. https://doi.org/10.1080/02699930902975754
10.1080/02699930902975754
Web of Science® Google Scholar
Brysbaert, M., Mandera, P., & Keuleers, E. (2018). The word frequency effect in word processing: An updated review. Current Directions in Psychological Science, 27(1), 45–50. https://doi.org/10.1177/0963721417727521
10.1177/0963721417727521
Web of Science® Google Scholar
Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes and evaluative space: A critical review, with emphasis on the separability of positive and negative substrates. Psychological Bulletin, 115(3), 401–423. https://doi.org/10.1037/0033-2909.115.3.401
10.1037/0033-2909.115.3.401
Web of Science® Google Scholar
Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1997). Beyond bipolar conceptualizations and measures: The case of attitudes and evaluative space. Personality and Social Psychology Review, 1(1), 3–25. https://doi.org/10.1207/s15327957pspr0101_2
10.1207/s15327957pspr0101_2
CAS PubMed Google Scholar
Carretié, L., Hinojosa, J. A., Albert, J., López-Martín, S., De La Gándara, B. S., Igoa, J. M., & Sotillo, M. (2008). Modulation of ongoing cognitive processes by emotionally intense words. Psychophysiology, 45(2), 188–196. https://doi.org/10.1111/j.1469-8986.2007.00617.x
10.1111/j.1469-8986.2007.00617.x
PubMed Web of Science® Google Scholar
Citron, F. M. M. (2012). Neural correlates of written emotion word processing: A review of recent electrophysiological and hemodynamic neuroimaging studies. Brain and Language, 122(3), 211–226. https://doi.org/10.1016/j.bandl.2011.12.007
10.1016/j.bandl.2011.12.007
PubMed Web of Science® Google Scholar
Citron, F. M. M., Abugaber, D., & Herbert, C. (2016). Approach and withdrawal tendencies during written word processing: Effects of task, emotional valence, and emotional arousal. Frontiers in Psychology, 6, 1935. https://doi.org/10.3389/fpsyg.2015.01935
10.3389/fpsyg.2015.01935
PubMed Web of Science® Google Scholar
Citron, F. M. M., Weekes, B. S., & Ferstl, E. C. (2013). Effects of valence and arousal on written word recognition: Time course and ERP correlates. Neuroscience Letters, 533, 90–95. https://doi.org/10.1016/j.neulet.2012.10.054
10.1016/j.neulet.2012.10.054
CAS PubMed Web of Science® Google Scholar
Citron, F. M. M., Weekes, B. S., & Ferstl, E. C. (2014). Arousal and emotional valence interact in written word recognition. Language, Cognition and Neuroscience, 29(10), 1257–1267. https://doi.org/10.1080/23273798.2014.897734
10.1080/23273798.2014.897734
Web of Science® Google Scholar
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Routledge Academic.
10.1046/j.1526-4610.2001.111006343.x
Google Scholar
Delaney-Busch, N., Wilkie, G., & Kuperberg, G. (2016). Vivid: How valence and arousal influence word processing under different task demands. Cognitive, Affective, & Behavioral Neuroscience, 16(3), 415–432. https://doi.org/10.3758/s13415-016-0402-y
10.3758/s13415-016-0402-y
PubMed Web of Science® Google Scholar
Dien, J. (2009). The neurocognitive basis of reading single words as seen through early latency ERPs: A model of converging pathways. Biological Psychology, 80(1), 10–22. https://doi.org/10.1016/j.biopsycho.2008.04.013
10.1016/j.biopsycho.2008.04.013
PubMed Web of Science® Google Scholar
Eder, A. B., & Rothermund, K. (2010). Automatic influence of arousal information on evaluative processing: Valencearousal interactions in an affective Simon task. Cognition & Emotion, 24(6), 1053–1061. https://doi.org/10.1080/02699930903056836
10.1080/02699930903056836
Google Scholar
Eimer, M., Gosling, A., Nicholas, S., & Kiss, M. (2011). The N170 component and its links to configural face processing: A rapid neural adaptation study. Brain Research, 1376, 76–87. https://doi.org/10.1016/j.brainres.2010.12.046
10.1016/j.brainres.2010.12.046
CAS PubMed Web of Science® Google Scholar
Estes, Z., & Adelman, J. S. (2008). Automatic vigilance for negative words in lexical decision and naming: Comment on Larsen, Mercer, and Balota (2006). Emotion, 8(4), 441–444. https://doi.org/10.1037/1528-3542.8.4.441
10.1037/1528-3542.8.4.441
PubMed Web of Science® Google Scholar
Fritsch, N., & Kuchinke, L. (2013). Acquired affective associations induce emotion effects in word recognition: An ERP study. Brain and Language, 124(1), 75–83. https://doi.org/10.1016/j.bandl.2012.12.001
10.1016/j.bandl.2012.12.001
PubMed Web of Science® Google Scholar
Frühholz, S., Jellinghaus, A., & Herrmann, M. (2011). Time course of implicit processing and explicit processing of emotional faces and emotional words. Biological Psychology, 87(2), 265–274. https://doi.org/10.1016/j.biopsycho.2011.03.008
10.1016/j.biopsycho.2011.03.008
PubMed Web of Science® Google Scholar
Gelbard-Sagiv, H., Magidov, E., Sharon, H., Hendler, T., & Nir, Y. (2018). Noradrenaline modulates visual perception and late visually evoked activity. Current Biology, 28(14), 2239–2249.e6. https://doi.org/10.1016/j.cub.2018.05.051
10.1016/j.cub.2018.05.051
CAS PubMed Web of Science® Google Scholar
Gibbons, H. (2009). Evaluative priming from subliminal emotional words: Insights from event-related potentials and individual differences related to anxiety. Consciousness and Cognition, 18(2), 383–400. https://doi.org/10.1016/j.concog.2009.02.007
10.1016/j.concog.2009.02.007
PubMed Web of Science® Google Scholar
Gibbons, H., Kirsten, H., & Seib-Pfeifer, L.-E. (2022). Attention to affect 2.0: Multiple effects of emotion and attention on event-related potentials of visual word processing in a valence-detection task. Psychophysiology, 59(9), e14059. https://doi.org/10.1111/psyp.14059
10.1111/psyp.14059
PubMed Web of Science® Google Scholar
Gibbons, H., Kirsten, H., & Seib-Pfeifer, L.-E. (2023). Attentional tuning of valent word forms. International Journal of Psychophysiology, 184, 84–93. https://doi.org/10.1016/j.ijpsycho.2022.12.005
10.1016/j.ijpsycho.2022.12.005
PubMed Web of Science® Google Scholar
Gibbons, H., Schmuck, J., & Kirsten, H. (2022). Interactions of ignored and attended valence in a valence-detection task with emotional words support the model of evaluative space: An ERP study. Language, Cognition and Neuroscience, 38(10), 1430–1450. https://doi.org/10.1080/23273798.2022.2088817
10.1080/23273798.2022.2088817
Web of Science® Google Scholar
Hajcak, G., MacNamara, A., & Olvet, D. M. (2010). Event-related potentials, emotion, and emotion regulation: An integrative review. Developmental Neuropsychology, 35(2), 129–155. https://doi.org/10.1080/87565640903526504
10.1080/87565640903526504
PubMed Web of Science® Google Scholar
Hauk, O., & Pulvermüller, F. (2004). Effects of word length and frequency on the human event-related potential. Clinical Neurophysiology, 115(5), 1090–1103. https://doi.org/10.1016/j.clinph.2003.12.020
10.1016/j.clinph.2003.12.020
CAS PubMed Web of Science® Google Scholar
Hauk, O., Davis, M. H., Ford, M., Pulvermüller, F., & Marslen-Wilson, W. D. (2006). The time course of visual word recognition as revealed by linear regression analysis of ERP data. NeuroImage, 30(4), 1383–1400. https://doi.org/10.1016/j.neuroimage.2005.11.048
10.1016/j.neuroimage.2005.11.048
CAS PubMed Web of Science® Google Scholar
Hauk, O., Pulvermüller, F., Ford, M., Marslen-Wilson, W. D., & Davis, M. H. (2009). Can I have a quick word? Early electrophysiological manifestations of psycholinguistic processes revealed by event-related regression analysis of the EEG. Biological Psychology, 80(1), 64–74. https://doi.org/10.1016/j.biopsycho.2008.04.015
10.1016/j.biopsycho.2008.04.015
CAS PubMed Web of Science® Google Scholar
Herbert, C., Junghofer, M., & Kissler, J. (2008). Event related potentials to emotional adjectives during reading. Psychophysiology, 45(3), 487–498. https://doi.org/10.1111/j.1469-8986.2007.00638.x
10.1111/j.1469-8986.2007.00638.x
CAS PubMed Web of Science® Google Scholar
Hinojosa, J. A., Carretié, L., Méndez-Bértolo, C., Míguez, A., & Pozo, M. A. (2009). Arousal contributions to affective priming: Electrophysiological correlates. Emotion, 9(2), 164–171. https://doi.org/10.1037/a0014680
10.1037/a0014680
PubMed Web of Science® Google Scholar
Hinojosa, J. A., Moreno, E. M., & Ferré, P. (2020). Affective neurolinguistics: Towards a framework for reconciling language and emotion. Language, Cognition and Neuroscience, 35(7), 813–839. https://doi.org/10.1080/23273798.2019.1620957
10.1080/23273798.2019.1620957
Web of Science® Google Scholar
Hofmann, M. J., Kuchinke, L., Tamm, S., Võ, M. L. H., & Jacobs, A. M. (2009). Affective processing within 1/10th of a second: High arousal is necessary for early facilitative processing of negative but not positive words. Cognitive, Affective, & Behavioral Neuroscience, 9(4), 389–397. https://doi.org/10.3758/9.4.389
10.3758/9.4.389
PubMed Web of Science® Google Scholar
Homan, R. W., Herman, J., & Purdy, P. (1987). Cerebral location of international 10 system electrode placement. Electroencephalography and Clinical Neurophysiology, 66(4), 376–382. https://doi.org/10.1016/0013-4694(87)90206-9
10.1016/0013-4694(87)90206-9
CAS PubMed Google Scholar
Huang, W., Wu, X., Hu, L., Wang, L., Ding, Y., & Qu, Z. (2017). Revisiting the earliest electrophysiological correlate of familiar face recognition. International Journal of Psychophysiology, 120, 42–53. https://doi.org/10.1016/j.ijpsycho.2017.07.001
10.1016/j.ijpsycho.2017.07.001
PubMed Web of Science® Google Scholar
Jemel, B., Pisani, M., Calabria, M., Crommelinck, M., & Bruyer, R. (2003). Is the N170 for faces cognitively penetrable? Evidence from repetition priming of Mooney faces of familiar and unfamiliar persons. Cognitive Brain Research, 17(2), 431–446. https://doi.org/10.1016/S0926-6410(03)00145-9
10.1016/S0926-6410(03)00145-9
PubMed Google Scholar
Junghöfer, M., Peyk, P., Flaisch, T., & Schupp, H. T. (2006). Neuroimaging methods in affective neuroscience: Selected methodological issues. Progress in Brain Research, 156, 123–143.
10.1016/S0079-6123(06)56007-8
PubMed Web of Science® Google Scholar
Kauschke, C., Bahn, D., Vesker, M., & Schwarzer, G. (2019). The role of emotional valence for the processing of facial and verbal stimuli or negativity bias? Frontiers in Psychology, 10, 1654. https://doi.org/10.3389/fpsyg.2019.01654
10.3389/fpsyg.2019.01654
PubMed Google Scholar
Keuper, K., Zwanzger, P., Nordt, M., Eden, A., Laeger, I., Zwitserlood, P., Kissler, J., Junghöfer, M., & Dobel, C. (2014). How “love” and “hate” differ from “sleep”: Using combined electro/magnetoencephalographic data to reveal the sources of early cortical responses to emotional words. Human Brain Mapping, 35(3), 875–888. https://doi.org/10.1002/hbm.22220
10.1002/hbm.22220
PubMed Web of Science® Google Scholar
Kever, A., Grynberg, D., Szmalec, A., Smalle, E., & Vermeulen, N. (2019). “Passion” versus “patience”: The effects of valence and arousal on constructive word recognition. Cognition and Emotion, 33(6), 1302–1309. https://doi.org/10.1080/02699931.2018.1561419
10.1080/02699931.2018.1561419
PubMed Web of Science® Google Scholar
Kissler, J., & Herbert, C. (2013). Emotion, Etmnooi, or Emitoon? Faster lexical access to emotional than to neutral words during reading. Biological Psychology, 92(3), 464–479. https://doi.org/10.1016/j.biopsycho.2012.09.004
10.1016/j.biopsycho.2012.09.004
PubMed Web of Science® Google Scholar
Kissler, J., Assadollahi, R., & Herbert, C. (2006). Emotional and semantic networks in visual word processing: Insights from ERP studies. Progress in Brain Research, 156, 147–183. https://doi.org/10.1016/S0079-6123(06)56008-X
10.1016/S0079-6123(06)56008-X
PubMed Web of Science® Google Scholar
Kissler, J., Herbert, C., Peyk, P., & Junghofer, M. (2007). Buzzwords: Early cortical responses to emotional words during reading. Psychological Science, 18(6), 475–480. https://doi.org/10.1111/j.1467-9280.2007.01924.x
10.1111/j.1467-9280.2007.01924.x
PubMed Web of Science® Google Scholar
Kissler, J., Herbert, C., Winkler, I., & Junghofer, M. (2009). Emotion and attention in visual word processing study. Biological Psychology, 80(1), 75–83. https://doi.org/10.1016/j.biopsycho.2008.03.004
10.1016/j.biopsycho.2008.03.004
PubMed Web of Science® Google Scholar
Kuchinke, L., Fritsch, N., & Müller, C. J. (2015). Evaluative conditioning of positive and negative valence affects P1 and N1 in verbal processing. Brain Research, 1624, 405–413. https://doi.org/10.1016/j.brainres.2015.07.059
10.1016/j.brainres.2015.07.059
CAS PubMed Web of Science® Google Scholar
Kuchinke, L., Krause, B., Fritsch, N., & Briesemeister, B. B. (2014). A familiar font drives early emotional effects in word recognition. Brain and Language, 137, 142–147. https://doi.org/10.1016/j.bandl.2014.08.007
10.1016/j.bandl.2014.08.007
PubMed Web of Science® Google Scholar
Kulke, L., Bayer, M., Grimm, A.-M., & Schacht, A. (2019). Differential effects of learned associations with words and pseudowords on event-related brain potentials. Neuropsychologia, 124, 182–191. https://doi.org/10.1016/j.neuropsychologia.2018.12.012
10.1016/j.neuropsychologia.2018.12.012
PubMed Web of Science® Google Scholar
Kuperman, V., Estes, Z., Brysbaert, M., & Warriner, A. B. (2014). Emotion and language: Valence and arousal affect word recognition. Journal of Experimental Psychology: General, 143(3), 1065–1081. https://doi.org/10.1037/a0035669
10.1037/a0035669
PubMed Web of Science® Google Scholar
Lenth, R. (2022). emmeans: Estimated marginal means, aka least-squares means. R package version 1.8.4-1.
Google Scholar
Luck, S. J., & Gaspelin, N. (2017). How to get statistically significant effects in any ERP experiment (and why you shouldn't). Psychophysiology, 54(1), 146–157. https://doi.org/10.1111/psyp.12639
10.1111/psyp.12639
PubMed Web of Science® Google Scholar
Mather, M. (2007). Emotional arousal and memory binding: An object-based framework. Perspectives on Psychological Science, 2(1), 33–52. https://doi.org/10.1111/j.1745-6916.2007.00028.x
10.1111/j.1745-6916.2007.00028.x
PubMed Web of Science® Google Scholar
Maunsell, J. H. R., & Treue, S. (2006). Feature-based attention in visual cortex. Trends in Neurosciences, 29(6), 317–322. https://doi.org/10.1016/j.tins.2006.04.001
10.1016/j.tins.2006.04.001
CAS PubMed Web of Science® Google Scholar
Méndez-Bértolo, C., Pozo, M. A., & Hinojosa, J. A. (2011). Word frequency modulates the processing of emotional words: Convergent behavioral and electrophysiological data. Neuroscience Letters, 494(3), 250–254. https://doi.org/10.1016/j.neulet.2011.03.026
10.1016/j.neulet.2011.03.026
CAS PubMed Web of Science® Google Scholar
New, B., Ferrand, L., Pallier, C., & Brysbaert, M. (2006). Reexamining the word length effect in visual word recognition: New evidence from the English Lexicon Project. Psychonomic Bulletin & Review, 13(1), 45–52. https://doi.org/10.3758/BF03193811
10.3758/BF03193811
PubMed Web of Science® Google Scholar
Norris, C. J. (2021). The negativity bias, revisited: Evidence from neuroscience measures and an individual differences approach. Social Neuroscience, 16(1), 68–82. https://doi.org/10.1080/17470919.2019.1696225
10.1080/17470919.2019.1696225
PubMed Web of Science® Google Scholar
Norris, C. J., Gollan, J., Berntson, G. G., & Cacioppo, J. T. (2010). The current status of research on the structure of evaluative space. Biological Psychology, 84(3), 422–436. https://doi.org/10.1016/j.biopsycho.2010.03.011
10.1016/j.biopsycho.2010.03.011
PubMed Web of Science® Google Scholar
Oganian, Y., Froehlich, E., Schlickeiser, U., Hofmann, M. J., Heekeren, H. R., & Jacobs, A. M. (2016). Slower perception followed by faster lexical decision in longer words: A diffusion model analysis. Frontiers in Psychology, 6, 1958. https://doi.org/10.3389/fpsyg.2015.01958
10.3389/fpsyg.2015.01958
PubMed Google Scholar
Olofsson, J. K., Nordin, S., Sequeira, H., & Polich, J. (2008). Affective picture processing: An integrative review of ERP findings. Biological Psychology, 77(3), 247–265. https://doi.org/10.1016/j.biopsycho.2007.11.006
10.1016/j.biopsycho.2007.11.006
PubMed Web of Science® Google Scholar
Palazova, M., Mantwill, K., Sommer, W., & Schacht, A. (2011). Are effects of emotion in single words non-lexical? Evidence from event-related brain potentials. Neuropsychologia, 49(9), 2766–2775. https://doi.org/10.1016/j.neuropsychologia.2011.06.005
10.1016/j.neuropsychologia.2011.06.005
PubMed Web of Science® Google Scholar
Peckham, A. D., McHugh, R. K., & Otto, M. W. (2010). A meta-analysis of the magnitude of biased attention in depression. Depression and Anxiety, 27(12), 1135–1142. https://doi.org/10.1002/da.20755
10.1002/da.20755
PubMed Web of Science® Google Scholar
Peelen, M. V., & Kastner, S. (2011). A neural basis for real-world visual search in human occipitotemporal cortex. Proceedings of the National Academy of Sciences of the United States of America, 108(29), 12125–12130. https://doi.org/10.1073/pnas.1101042108
10.1073/pnas.1101042108
CAS PubMed Web of Science® Google Scholar
Penolazzi, B., Hauk, O., & Pulvermüller, F. (2007). Early semantic context integration and lexical access as revealed by event-related brain potentials. Biological Psychology, 74(3), 374–388. https://doi.org/10.1016/j.biopsycho.2006.09.008
10.1016/j.biopsycho.2006.09.008
PubMed Web of Science® Google Scholar
Phelps, E. A., Ling, S., & Carrasco, M. (2006). Emotion facilitates perception and potentiates the perceptual benefits of attention. Psychological Science, 17(4), 292–299. https://doi.org/10.1111/j.1467-9280.2006.01701.x
10.1111/j.1467-9280.2006.01701.x
PubMed Web of Science® Google Scholar
Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Review of Neuroscience, 13(1), 25–42. https://doi.org/10.1146/annurev.ne.13.030190.000325
10.1146/annurev.ne.13.030190.000325
CAS PubMed Web of Science® Google Scholar
Pourtois, G., & Vuilleumier, P. (2006). Dynamics of emotional effects on spatial attention in the human visual cortex. In Progress in brain research (Vol. 156, pp. 67–91). Elsevier. https://doi.org/10.1016/S0079-6123(06)56004-2
10.1016/S0079-6123(06)56004-2
Google Scholar
R Core Team. (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/
Google Scholar
Ramos-Loyo, J., González-Garrido, A. A., Llamas-Alonso, L. A., & Sequeira, H. (2022). Sex differences in cognitive processing: An integrative review of electrophysiological findings. Biological Psychology, 172, 108370. https://doi.org/10.1016/j.biopsycho.2022.108370
10.1016/j.biopsycho.2022.108370
PubMed Web of Science® Google Scholar
Recio, G., Conrad, M., Hansen, L. B., & Jacobs, A. M. (2014). On pleasure and thrill: The interplay between arousal and valence during visual word recognition. Brain and Language, 134, 34–43. https://doi.org/10.1016/j.bandl.2014.03.009
10.1016/j.bandl.2014.03.009
PubMed Web of Science® Google Scholar
Rellecke, J., Sommer, W., & Schacht, A. (2012). Does processing of emotional facial expressions depend on intention? Time-resolved evidence from event-related brain potentials. Biological Psychology, 90(1), 23–32. https://doi.org/10.1016/j.biopsycho.2012.02.002
10.1016/j.biopsycho.2012.02.002
PubMed Web of Science® Google Scholar
Robinson, M. D. (1998). Running from William James' Bear: A review of preattentive mechanisms and their contributions to emotional experience. Cognition & Emotion, 12(5), 667–696. https://doi.org/10.1080/026999398379493
10.1080/026999398379493
Web of Science® Google Scholar
Robinson, M. D., Storbeck, J., Meier, B. P., & Kirkeby, B. S. (2004). Watch out! That could be dangerous: Valence-arousal interactions in evaluative processing. Personality and Social Psychology Bulletin, 30(11), 1472–1484. https://doi.org/10.1177/0146167204266647
10.1177/0146167204266647
PubMed Web of Science® Google Scholar
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39(6), 1161–1178. https://doi.org/10.1037/h0077714
10.1037/h0077714
Web of Science® Google Scholar
Russell, J. A., & Barrett, L. F. (1999). Core affect, prototypical emotional episodes, and other things called emotion: Dissecting the elephant. Journal of Personality and Social Psychology, 76(5), 805–819. https://doi.org/10.1037/0022-3514.76.5.805
10.1037/0022-3514.76.5.805
CAS PubMed Web of Science® Google Scholar
Sabatinelli, D., Fortune, E. E., Li, Q., Siddiqui, A., Krafft, C., Oliver, W. T., Beck, S., & Jeffries, J. (2011). Emotional perception: Meta-analyses of face and natural scene processing. NeuroImage, 54(3), 2524–2533. https://doi.org/10.1016/j.neuroimage.2010.10.011
10.1016/j.neuroimage.2010.10.011
PubMed Web of Science® Google Scholar
Sagiv, N., & Bentin, S. (2001). Structural encoding of human and schematic faces: Holistic and part-based processes. Journal of Cognitive Neuroscience, 13(7), 937–951. https://doi.org/10.1162/089892901753165854
10.1162/089892901753165854
CAS PubMed Web of Science® Google Scholar
Schacht, A., & Sommer, W. (2009a). Emotions in word and face processing: Early and late cortical responses. Brain and Cognition, 69(3), 538–550. https://doi.org/10.1016/j.bandc.2008.11.005
10.1016/j.bandc.2008.11.005
PubMed Web of Science® Google Scholar
Schacht, A., & Sommer, W. (2009b). Time course and task dependence of emotion effects in word processing. Cognitive, Affective, & Behavioral Neuroscience, 9(1), 28–43. https://doi.org/10.3758/CABN.9.1.28
10.3758/CABN.9.1.28
PubMed Web of Science® Google Scholar
Schindler, S., & Bublatzky, F. (2020). Attention and emotion: An integrative review of emotional face processing as a function of attention. Cortex, 130, 362–386. https://doi.org/10.1016/j.cortex.2020.06.010
10.1016/j.cortex.2020.06.010
PubMed Web of Science® Google Scholar
Schindler, S., & Kissler, J. (2016). Selective visual attention to emotional words: Early parallel frontal and visual activations followed by interactive effects in visual cortex. Human Brain Mapping, 37(10), 3575–3587. https://doi.org/10.1002/hbm.23261
10.1002/hbm.23261
PubMed Web of Science® Google Scholar
Schindler, S., & Straube, T. (2020). Selective visual attention to emotional pictures: Interactions of task-relevance and emotion are restricted to the late positive potential. Psychophysiology, 57(9), e13585. https://doi.org/10.1111/psyp.13585
10.1111/psyp.13585
PubMed Web of Science® Google Scholar
Schindler, S., Vormbrock, R., & Kissler, J. (2019). Emotion in context: How sender predictability and identity affect processing of words as imminent personality feedback. Frontiers in Psychology, 10, 94. https://doi.org/10.3389/fpsyg.2019.00094
10.3389/fpsyg.2019.00094
PubMed Web of Science® Google Scholar
Schmuck, J., Schnuerch, R., Kirsten, H., Shivani, V., & Gibbons, H. (2023). The influence of selective attention to specific emotions on the processing of faces as revealed by event-related brain potentials. Psychophysiology, 60, e14325. https://doi.org/10.1111/psyp.14325
10.1111/psyp.14325
PubMed Web of Science® Google Scholar
Schupp, H. T., Stockburger, J., Codispoti, M., Junghöfer, M., Weike, A. I., & Hamm, A. O. (2007). Selective visual attention to emotion. Journal of Neuroscience, 27(5), 1082–1089. https://doi.org/10.1523/JNEUROSCI.3223-06.2007
10.1523/JNEUROSCI.3223-06.2007
CAS PubMed Web of Science® Google Scholar
Schwibbe, M., Räder, K., Schwibbe, K., Borchardt, M., & Geiken-Pophanken, G. (1994). Zum emotionalen Gehalt von Substantiven, Adjektiven und Verben [The emotional meaning of nouns, adjectives, and verbs]. In W. Hager & M. Hasselhorn (Eds.), Handbuch deutschsprachiger Wortnormen [Handbook of German Word Norms] (pp. 272–284). Hogrefe.
Google Scholar
Scott, G. G., O'Donnell, P. J., Leuthold, H., & Sereno, S. C. (2009). Early emotion word processing: Evidence from event-related potentials. Biological Psychology, 80(1), 95–104. https://doi.org/10.1016/j.biopsycho.2008.03.010
10.1016/j.biopsycho.2008.03.010
PubMed Web of Science® Google Scholar
Simon, G., Petit, L., Bernard, C., & Rebaï, M. (2007). N170 ERPs could represent a logographic processing strategy in visual word recognition. Behavioral and Brain Functions, 3(1), 21. https://doi.org/10.1186/1744-9081-3-21
10.1186/1744-9081-3-21
PubMed Google Scholar
Singmann, H., Bolker, B., Westfall, J., Aust, F., & Ben-Shachar, M. (2022). afex: Analysis of factorial experiments. R package version 1.1-1.
Google Scholar
Vuilleumier, P., & Pourtois, G. (2007). Distributed and interactive brain mechanisms during emotion face perception: Evidence from functional neuroimaging. Neuropsychologia, 45(1), 174–194. https://doi.org/10.1016/j.neuropsychologia.2006.06.003
10.1016/j.neuropsychologia.2006.06.003
PubMed Web of Science® Google Scholar
Võ, M. L. H., Conrad, M., Kuchinke, L., Urton, K., Hofmann, M. J., & Jacobs, A. M. (2009). The Berlin affective word list reloaded (BAWL-R). Behavior Research Methods, 41(2), 534–538. https://doi.org/10.3758/BRM.41.2.534
10.3758/BRM.41.2.534
PubMed Web of Science® Google Scholar
Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L. D., François, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T. L., Miller, E., Bache, S. M., Müller, K., Ooms, J., Robinson, D., Seidel, D. P., Spinu, V., … Yutani, H. (2019). Welcome to the Tidyverse. Journal of Open Source Software, 4(43), 1686. https://doi.org/10.21105/joss.01686
10.21105/joss.01686
Google Scholar
Yao, Z., Yu, D., Wang, L., Zhu, X., Guo, J., & Wang, Z. (2016). Effects of valence and arousal on emotional word processing are modulated by concreteness: Behavioral and ERP evidence from a lexical decision task. International Journal of Psychophysiology, 110, 231–242. https://doi.org/10.1016/j.ijpsycho.2016.07.499
10.1016/j.ijpsycho.2016.07.499
PubMed Web of Science® Google Scholar
Zhang, D., He, W., Wang, T., Luo, W., Zhu, X., Gu, R., Li, H., & Luo, Y. (2014). Three stages of emotional word processing: An ERP study with rapid serial visual presentation. Social Cognitive and Affective Neuroscience, 9(12), 1897–1903. https://doi.org/10.1093/scan/nst188
10.1093/scan/nst188
PubMed Web of Science® Google Scholar

1 Please note that N1 and word N170 are often used interchangeably in the literature on word processing. For example, previous studies on N1 effects of words (Kissler et al., 2009; Scott et al., 2009) showed a time window and topography similar to the word N170 in our study. Following our previous study (Gibbons et al., 2023) and the component peak in the current study, we will use the term N170 in the following. Importantly and unless stated otherwise, N170 thus refers to the word N170, which has to be distinguished from the face N170 (see Dien, 2009) which is rather specific to facial stimuli and related to their structural encoding and configural face representation (e.g., Eimer et al., 2011).
2 One might argue that a more specific prediction would be justified or even required, referring to greater P1 for positive compared to negative target, but not nontarget words, which would replicate Gibbons et al. (2023). However, the present study employs a different set of words that, moreover, no longer contains neutral words. This significant contextual change might alter the participants' focus on positive vs. negative words (see discussion). Therefore, the most basic conclusion to be drawn from the hypothesis of valent word forms and from Gibbons et al.'s (2023) P1 findings refers to greater P1 valence discrimination in target than nontarget words.
3 Note however that an unusually early EPN in words could also be due to the specific properties of a valence detection task. Selective attention to specific levels of valence might particularly accelerate the processing of target affect, compared to previous ERP studies on affective words that sometimes did not require attention to affect at all. Thus, the N170 and EPN components might partly merge in this specific task under speed instructions.

Citing Literature

Volume62, Issue1

January 2025

e14748

The influence of lexical word properties on selective attention to emotional words: Support for the attentional tuning of valent word forms

Abstract