The role of ambiguous valence, origin, and activation in behavioral and eye-tracking measurements during N-back task Authors: Adrianna Wielgopolan, Radosław Wcześniak, Kamil K. Imbir Published: 2026 ABSTRACT This study investigates how emotionally ambiguous words can influence performance in the N-back task. In two experiments, we measured differences in behavior (performance accuracy and reaction times) and management of visual attention (eye-tracking measurement: number of fixations, mean duration of fixation) for ambiguous and unambiguous words. We hypothesized that words ambiguous on one of three emotional spaces – valence (positivity and negativity), origin (automaticity and reflectiveness), and activation (arousal and subjective significance) – would increase accuracy in the N-back task (Experiment 1), but also elicit more and longer fixations (Experiment 2) compared to the unidimensional, unambiguous words. The behavioral hypotheses were confirmed for words of ambiguous valence and origin, but not for activation ambiguity. Furthermore, all eye-tracking hypotheses were confirmed, finding differences between control words and all ambiguous groups of words. Our findings further show that ambiguity may be found in emotional spaces other than valence (i.e., origin), which may significantly influence our cognitive functioning. Most of the stimuli that we encounter – and, in consequence, our emotional responses to them – are unidimensional, allowing us to quickly assess their characteristics; however, this is not always the case. Some of our experiences are ambivalent (e.g.,1,2) or ambiguous3, and may not be classified as easily. Recent literature suggests that ambivalence is not as rare as initially thought, and while it does not occur as often as unidimensional affect, it is nevertheless within the norm of our emotional functioning4. Therefore, it seemed crucial to meticulously study the properties of ambivalence and other kinds of emotional ambiguity (on the spaces of origin and activation) and their influence on cognitive functioning. We studied those properties in two experiments – one behavioral and one eye-tracking – verifying how those ambiguities perceived in emotional words may affect performance in N-back tasks (with letter target stimuli). Ambivalence is usually defined as feeling positivity and negativity at the same time5–8, or as the co-occurrence of two opposing affects9; it is also often known as the mixed emotions phenomenon10,11. Our emotional states are so diverse and complex12 that we might feel joyous excitement and unpleasant fear, amusement and disgust13,14, or happiness and sadness7,15 all at the same moment. However, despite recent evidence that ambivalence may be perceived in emotions that were until now defined as unidimensional, such as basic emotions16, most of the literature still concerns only the unidimensional states. This may reflect a failure to provide the appropriate measures that would allow research participants to describe their ambivalent emotions. When we talk about emotions – especially in everyday life, not only in research – we usually intuitively use the categorical (discrete) approach (e.g.,17). When telling our friend how we feel, we would say that we are “sad” rather than we are feeling “very negative, lowly-arousing, basic, and automatic affect”. We need those categories, and we use them daily; however, sometimes, they might introduce a bit of simplification. The dimensional approach to emotions has the assumption that every emotion may be placed somewhere on a dimension: a characteristic in which intensity describes the emotion18–20. Using the example of feeling sad, if we assume that we have an underlying dimension of valence, ranging from extremely negative to extremely positive21,22, this sadness could be placed low on a valence dimension. It would be described as negative. There is an ongoing debate as to how many dimensions should there be, and which are the most important and universal20,22–24. Nevertheless, the dimensional approach allows us to describe emotions in detail, differentiate various kinds of one emotion (e.g., positive or negative surprise, or even to map the temporal changes in valence of the typical emotion, such as surprise;25,26), and generalize its specifics into the whole category of emotions (e.g., assessing the consequences of all high-arousing states). However, the dimensional approach is nuanced. The conceptualization of the dimensions will be crucial, because if something cannot be marked on those dimensions, and cannot be captured by them, it will not be reported and thus noticed. That is the case with ambivalence: the bipolar scale of valence with high negativity (unpleasantness) at one end and high positivity (pleasantness22 at the other does not allow for ambivalence. It could be argued that the middle point of the scale – the neutrality27– represents ambivalence. However, even if we assume that this middle point does not mean the “zero” point of no intensity of the two constructs depicted by the scale (as it was for most of the dimensional models, e.g.,22,28,29) and we agree that the middle point is the mixed emotions, there is still no way for this dimension to measure the intensity of the ambivalence. It therefore seems intuitive that all the complexity of mixed emotions cannot be marked on one tiny spot of the scale: mixed emotional states may constitute different proportions of positive and negative affect and have a varying structure26,30. Therefore, there was a need for a model that could account for the separate dimensions of the positivity and negativity31, account for their independence from one another32,33, and describe the phenomenon of ambivalence. One of the most successful models that takes all of these aspects into account is the Evaluative Space Model (ESM) postulated by Cacioppo and his team6,34,35. The authors acknowledged the fact that affective functioning is more complex and diverse than the bipolar dimension could capture. In the ESM, we may find a description of the independence of two opposite dimensions: positivity (approach system) and negativity (avoidance system). Each of these constructs may be activated without the other (they are functionally independent to some degree, so the increase in positivity does not necessarily mean the decrease in negativity and vice versa), or they may be activated at the same time (through reciprocal activation, independent activation, or co-activation8. In the ESM, the separate dimensions of the positivity and negativity are visualized as next to each other at a 90° angle. This creates a square space in which the ambivalence may be marked and differentiated, allowing people to indicate the intensity of both the positivity and negativity8. It therefore allows them to precisely describe emotions such as feeling bittersweet, longing, or having a disappointing win. Larsen et al. 7 found some specific situations in which participants of their study reported feeling ambivalent (on an average day they usually only used the bipolar categorization); furthermore, they determined that ambivalence may be elicited in a laboratory36, specifically in gambling scenarios with outcomes of disappointing wins and relieving losses, and after a short exposure to pictures of opposite valence8. The authors were even able to create a self-reported tool to describe the ambivalence36. Emotional ambiguity While the ESM described the possibility of the co-activation of two opposite characteristics – ambivalence8,35– that was a very specific phenomenon. However, this might not be the case. Brainerd et al. 37,38 reported attributional and semantic ambiguities, which significantly influenced cognitive processing. An important clarification, however, is that ambiguity was defined by Brainerd et al. as an uncertainty. Our understanding of ambiguity and ambivalence is thus different, and for this article we will refer to Cacioppo and Berntson’s 5,35 aforementioned definition of ambivalence: co-activation of the two opposite characteristics. Until recently, the only emotional dimensions that were seen as ambivalent were positivity and negativity. Wielgopolan and Imbir3,39 postulated that perhaps ambivalence is not a seldom phenomenon but rather an example of the relation between two variables. Those authors verified that other dimensions may persist in a similar relation, being opposite but not excluding each other. Therefore, the core assumption of the theory was that there is a construct of ambiguity. Ambiguity meant the significant intensity of any two opposite emotional dimensions, activated at the same time. Ambivalence was one kind of ambiguity, and until now, the only one that was studied and described (apart from the aforementioned non-emotional, semantic ambiguities37,40. Those ambiguities may be described on emotional spaces, built by two dimensions only, similarly to the space of valence (built by positivity and negativity) in Cacioppo’s theory6,35. The theory introduced two new spaces of ambiguity (apart from ambivalence): the origin and activation spaces, describing the ambiguities between dimensions of automaticity and reflectiveness, and arousal and subjective significance, respectively. Both the origin and activation spaces are built from dimensions in the understanding of the dual process theories, which usually differentiate the fast, effortless, “hot” system/mode of processing, and the slow, effortful, “cold” system/mode41–43. Jarymowicz and Imbir44 postulated that this differentiation may also be applied to emotions. The origin of emotions – how emotions appear, how they originate – differentiates all emotional states into two categories based on their amount of the cognitive component: (1) automatic emotions, which appear quickly, and require little or no cognitive processing, and (2) reflective emotions, which have a slower origin, require a lot of cognitive processing, and are effortful45–47. An example of an automatic emotion may be fear – we do not need to process the situation to be scared; we may just see something terrifying, hear a noise, feel a touch, and we intuitively know that this is a threat and react accordingly. Conversely, a reflective emotion could be disappointment: in this case, we may take time to compare the current situation to the desired state, evaluate it, and see that it does not fit the criteria. The origin was initially only visualized as a single bipolar dimension, ranging from very automatic to very reflective44,48,49, similar to valence in some models21,22. However, the recent approach takes ambiguity into consideration and separates one bipolar scale into two unipolar ones: automaticity and reflectiveness3. As those two dimensions were negatively correlated with each other3,50,51, they seemed to be the perfect example of a different kind of ambivalence – an origin ambiguity. This ambiguity was successfully perceived by participants in the emotional load of words3 and, similar to ambivalence, in the assessment of the emotionality, processing it 52,53, and its consequences (in review; submitted). Meanwhile, the activation space is created by the dimensions of arousal22,54 and subjective significance55,56, which are – in the light of dual process theories – two types of activation. Arousal is an automatic kind of activation: physical energy directed towards some action, usually in response to external stimuli22,55. It does not require much cognitive effort or planning; very often, this is a first, intuitive reaction. On the other hand, subjective significance is a reflective activation: an effortful reaction based on planning, processing, and evaluating51,55,57. Those two kinds of activation explain the discrepancy in describing this construct in previous literature (at the same time as a bodily reaction and mental energy). In previous research, arousal and subjective significance had separate influences on cognitive functioning49,58, thus proving to be independent and distinguishable dimensions. As an emotional space, however, they have a different relationship than automaticity and reflectiveness; arousal and subjective significance, while separate, seem to be positively correlated with each other3,51. Arousal and subjective significance were, therefore, expected to have different characteristics than ambivalence and origin ambiguity. They were more similar to the blends of emotions59,60: mixed emotions consisted of two emotional states that had the same valence, but were different qualitatively (e.g., feeling two negative emotions at the same time: fear and anger). In previous experiments, words of activation ambiguity had their own properties and consequences52,53. The proposed theory could further broaden and explain the core assumptions of ESM theory and previous work on ambivalence. By studying new spaces of ambiguity, Wielgopolan and Imbir3,39 show that the core of ESM is not the fact that we have two specific dimensions of positivity and negativity, but the fact that they are each other’s opposites. It is not about the positivity and negativity, but about the relation between them, and that relation may be replicated amongst other dimensions. This may actually explain the mechanism behind ambivalence and the ESM, and allow us to further differentiate emotions and predict their consequences with higher precision. While the theories we presented mostly concern emotions as states experienced by people44, all of the aforementioned dimensions can be found in words3,51. Positivity, negativity, automaticity, reflectiveness, arousal, and subjective significance have all been perceived in the emotional load of words; in normative studies (conducted similarly to all previous studies that created norms for linguistic stimuli, e.g.,61–65), participants could evaluate all of these characteristics seen in a single word, and thus enable the researchers to create norms for each word3,51. Ambiguous words and cognitive processes Previous studies on ambiguous words Most psycholinguistic research concerning relations between emotions and language uses a dimensional approach66. Nevertheless, studying semantic ambiguity specifically is quite a recent interest among researchers, and emotional semantic ambiguity is still understudied4,67, sparking a very recent interest in the field. The emotional ambiguity hypothesis40,68 suggests that two different dimensions might be activated at the same time, based on the claim that emotional ambiguity presents overall distinct consequences for human behavior7,69,70. Ambiguity is an emotional state with distinct specificity; therefore, emotionally ambiguous words would have a distinct influence on cognitive processes. Brainerd40 found that the relation between valence and arousal is influenced by ambiguity as an additional variable. Following dualistic dimension theory44,46, it was claimed that ambivalent language stimuli contain more cognitive components, and thus require deeper, more nuanced, analytic cognitive processing. That claim was supported, yet in a limited way, by empirical evidence. When asked to rate the emotionality of ambiguous and unidimensional words, participants stopped their gaze significantly more and for a longer time on ambiguous rather than unidimensional words (Wielgopolan & Imbir, 2023b). If we assume that gaze is an indicator of attention (Beck & Lavie, 2005), the ambiguous words did require more of it. Furthermore, Brainerd37,71 indicated that more of the emotionally ambiguous words were recalled correctly than the emotionally equivocal words. Wielgopolan and Imbir (in review) also found that words containing a certain intensity of two opposite features, negatively correlated with each other, seem to be remembered better: when asked to memorize a list of words, participants could remember significantly more ambiguous words compared to the unidimensional ones. This effect was also replicated in a simple N-back task using words – the accuracy in recognizing the ambiguous words was significantly higher, and the reaction times significantly shorter, than for unidimensional stimuli. Perhaps their multidimensional nature may require more cognitive engagement and, thus, every aspect of those words is processed more carefully and deeply. This would suggest that words carrying an ambiguous emotional charge would require a more thorough cognitive appraisal. This appraisal would be visible both in task outcomes and physiological measures such as arousal or, specific to our study, eye-tracking indices such as fixation time. Nevertheless, until now, we only had the results of the simple rating task for ambiguous words (deciding whether those words are emotional or not52,53, simple remembering and recalling of the stimuli37,71, or N-back performed on the ambiguous words. There were no studies that could show the diffusive nature of the ambiguity, that is, the influence of ambiguous stimuli when they are not the center of our attention, but rather a distraction that has to be processed. The eye-tracking measurements While there are many different ways to measure cognitive processes, eye-tracking is one of the very few that enables researchers to determine how people manage their visual attention: when they look, for how long, or the dynamics of their gaze72. For example, eye-tracking measurements can show the fixations of the gaze, in other words, which locations on the screen participants stop to look at73,74. Further analysis can show how many fixations there were on each part of the screen, how long those fixations took on average, and whether it was a single fixation or there were more. All of this information may be linked to psychological variables; previous research shows that eye movements are related to attention75. Therefore, stimuli (in our case, words) that require more attention (longer processing) would elicit more fixations, lasting longer on average. Aims and hypotheses We wanted to further explore the links between emotional ambiguity and cognitive functioning. More precisely, our main aim was to investigate how different kinds of emotional ambiguity in distracting stimuli may influence a cognitive task (the N-back) performed at the same time. For one of the first times, we test words that are characterized by three kinds of ambiguity in their emotional load: ambivalence, origin ambiguity, and activation ambiguity. Although previous studies have shown some differences in the evaluation and perception of words of ambivalence and origin ambiguity compared to words of activation ambiguity, in the case of the N-back task, we do not expect any differences and hypothesize that all of the ambiguities – all kinds of mixed characteristics – will increase accuracy in the task. Therefore, we expected that the emotionality in words (distractor stimuli) would influence the remembering and recognizing of the target (neutral) stimuli in the task. In Experiment 1, we expected that the words of valence, origin, and activation ambiguities would significantly increase accuracy in the N-back task compared to the control-group words; we expected that the properties of the ambiguous stimuli would cause participants to process them longer, to analyze them (as they would seem more unequivocal; Wielgopolan & Imbir, 2023a, 2023b), and thus possibly evoke a more focused, analytical approach, increasing the accuracy and speed of completing the N-back task. Our previous experiments (in review) proved that evaluating words of different ambiguities did not differ significantly in reaction time. In this study, we also expected to fail to reject the null hypothesis regarding significant differences in reaction time between words of different ambiguities52. In Experiment 2, we expected significantly more fixations and a longer mean fixation time for ambiguous words compared to unidimensional ones, in line with previous studies52. We expected that the ambiguous words would be more troubling for participants and the fact that they would sense the mixed emotional load in them would cause significantly more fixations (returning the eye gaze to the ambiguous stimulus) and longer processing time (looking at the stimulus for longer in order to analyze it). Experiment 1 Methods Participants Firstly, we checked the estimated sample size using G*Power software (version 3.1.9.476,77). We assumed the effect size of η2 = 0.05, based on the results from the previous studies employing the same stimuli52,53. It is important to remember that the present study has a different task than the previous ones; however, we decided to use the previous effect sizes as our stimuli are specific, and there were not many previous studies using ambiguous words (and therefore subtle manipulation of two affective characteristics in one word). As we wanted to conduct a repeated measures analysis of variance (ANOVA) with seven dependent measurements, the minimum sample (assuming the 95% power, two-sided α = 0.05, and an effect size of η2 = 0.05; f = 0.23) would be 40 participants. Taking into consideration the possible dropout rate, we decided to increase the group to 60 participants. We recruited 60 volunteers via social media, all of whom were young adults, usually students in Polish universities. Our requirements during the recruiting process were native knowledge of the Polish language (in order to easily understand the semantic meaning of the stimuli, but also their emotional load, which may have some cultural background; Wierzbicka, 1999) and normal or corrected-to-normal vision (wearing glasses or contact lenses). From the initial sample size of 60, we excluded four participants because of monotonous answers, very short reaction times, or very low accuracy in the task. Therefore, we analyzed a sample of 56 people (30 women, 26 men), aged 18 to 30 years (M = 21.34, SD = 1.12). We ensured that the data we gathered were anonymous and only analyzed the data in groups. We obtained informed consent for participating in the experiments from all participants; the experimenter stressed the fact that participants may withdraw from the study at any time without providing any explanation. Participants received a remuneration of zł 40 (about USD 10). All the procedures were conducted in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. We obtained the approval of the Research Ethics Committee of the Faculty of Psychology at the University of Warsaw (approval no. 14/11/22). Design We conducted the study in a repeated measures design using seven groups (words: control, moderate ambivalence, high ambivalence, moderate origin ambiguity, high origin ambiguity, moderate activation ambiguity, high activation ambiguity). Each participant completed the N-back task on 175 words (25 words in each group), presented in a completely randomized order (regardless of the blocks of ambiguities – the words were derived from the full list of 175 words). Materials N-back with distractors We used a 3-back version of the N-back task78–80, with letters as target stimuli. We presented them to participants one by one (using Times New Roman black font; the size of the stimuli was automatically sized to take up 10% of the screen). Participants were instructed to remember the stimuli and answer whether the stimulus currently presented on the screen was the same as the stimulus three stimuli ago. We used a paradigm of N-back with distractors (similar to the mixed paradigms of N-back and flankers81 in order to study the transfer of the emotionality of words (ambiguous or unidimensional) onto the target stimuli. In our version of the N-back task, words were presented as distractor stimuli with each trial (see Words materials section). With each trial, participants always saw one letter stimulus and one word stimulus. One of those two stimuli were presented at the bottom of the screen, and one at the top; however, placement was randomized (the letter might have been at the top and word at the bottom, or vice versa), so participants had to read both of them (even unconsciously). During analysis, we took into consideration the trials during which participants had to give their answer (decide whether the letter that they see right now is the same or different than the letter three letters ago). However, the distractor stimulus was the same throughout all three trials and presentation of letters (therefore, we excluded any potential influence of different words for each of the three remembered stimuli). We visualize the experimental procedure in Fig. 1. Fig. 1A visualization of the experimental procedure, showing the consecutive steps of the N-back task. A visualization of the experimental procedure, showing the consecutive steps of the N-back task. Words materials We used a list of 175 words, composed from the dataset containing the linguistic stimuli validated by their type and intensity of ambiguity, created in a previous study3. The chosen database is, to our knowledge, one of very few databases of ambivalent stimuli and the only one with stimuli ambiguous on the spaces of origin and activation. The list, prepared in previous experiments52, was used to ensure that all kinds of ambiguity were included. It consisted of seven groups (25 words each), differing in their kind of ambiguity (on one of three spaces: valence, origin, and activation) and the intensity of the respective ambiguity (low – control group, moderate, high). The ratings of the words were created in a procedure3 during which each participant assessed the extent to which each word (out of the randomly drawn set) is positive, negative, automatic, reflective, arousing, and subjectively significant, using separate scales ranging from 0 to 100. Additionally, scales were labeled with both written descriptions of the dimensions and pictures showing schematic, very simple examples of emotional reactions, e.g., Self-Assessment Manikins61. After gathering the responses, the ratings of the words were created as mean values on each dimension. As the indicator of ambiguity, we took the sum of the intensity of the two dimensions creating the ambiguity, e.g., the words of high valence ambiguity had high assessments of negativity and high assessments of positivity. We argue that this measurement of ambiguity is much more direct than using standard deviations (e.g., 37), which may be affected by many other factors, such as variability of the group itself – different individual characteristics; therefore, our measurement should be reliable and show the nature of the ambiguity. In order to verify the creation of the list of words, we conducted a one-way ANOVA analysis with seven dependent measurements in order to ensure that the word groups did not significantly differ in the number of letters and frequency of use82; we also ensured that the groups of different ambiguities were different from the control group on the dimensions, creating the space of ambiguity only. Summary characteristics of the words in terms of frequency of use, number of letters, and their respective ambiguity are presented below in Table 1. Table 1Summary characteristics of the words.EnglishPolishFrequency (Kazojć, 2011)OriginReflectivenessValenceArousalLettersdungeonloch37123,9053,5228,9757,524whorekurwa173135,3030,8038,7191,205graduateabsolwent20622,5283,9565,7632,579affrontafront9523,0772,4726,1386,676mimosamimoza3735,3837,1931,1931,486commandkomenda28612,1072,7124,3366,247mindumysł844116,6586,0587,3632,595mustachewąsy145328,1740,5631,6133,394cannonarmata12635,4837,4328,7081,176troutpstrąg9216,7379,0735,3179,736vomitwymiociny3219,2446,6229,2478,629cheersokrzyki285648,4244,2036,6078,327growlryk262246,6135,4432,6777,223utensilprzyrząd58525,2181,3750,1633,748fussinessawanturnictwo1754,1033,4028,0376,9413knowledgewiedza242033,8992,6892,6330,056theoryteoria164315,4485,3954,7823,786otterwydra8747,6343,8030,7033,535trustzaufanie257079,7638,5785,3323,718self-controlsamoopanowanie133,0078,6884,0522,3714kindnessżyczliwość72178,0032,1580,6334,5210cheerfulnesspogoda ducha5454,586,0523,3077,8027,3212prudencerozwaga11825,3787,5877,7924,637honestyuczciwość58568,6541,6277,7327,389maluszekmaluszek081,6913,5550,5931,068dolphindelfin30749,9634,5023,4819,706diagramschemat55616,5580,0546,7722,957chickenkurczak18723,4087,4715,5026,407calfcielę26738,8320,7815,5426,485aspectaspekt55529,3072,5249,1326,786routinerutyna13026,6271,3868,9327,526dresssukienka51149,4041,2531,7531,058screenekran358330,8057,5556,8031,605conceivingobmyślanie2031,7882,5777,6132,9610spreadrozpiętość7725,5257,8637,5035,0010positionstanowisko439130,3875,1075,6235,1010sentimentsentyment21770,6217,5250,9535,579beetlechrząszcz7520,7043,3017,4336,659streetulica216431,6556,3536,6236,855cradlekołyska5949,4839,3930,1337,437meatmięso290915,3855,0437,2638,835winewino421455,4840,0032,4840,704dragonsmok343827,3843,1920,4851,434clubklub87048,8732,4718,6051,804chunkprzystojniak7742,3040,1028,1455,9012mischiefpsota2245,0827,6432,2658,385discodyskoteka3266,0420,5025,7466,179punkpunk5149,4832,1533,7068,264perversionperwersja5953,6351,6832,0568,429sinzgrzeszenie162,4533,0531,5072,6411cottonbawełna6834,8251,8624,2710,457restodpoczynek123154,3051,9179,8616,7710sketchszkic64540,3561,2446,3023,775twiggałązka27541,2748,4729,5324,677nymphnimfa19261,8149,0514,8028,145headgłowa930234,8483,0062,2128,165thrushdrozd15136,0048,0024,9528,775casewalizka28835,3053,8335,3931,097titletytuł353934,0461,3944,0431,385lacekoronka6544,8352,2225,2832,337patiencecierpliwość185066,7155,4782,3933,7511patiencecierpliwość185066,7155,4782,3933,7511ethicalityetyczność034,1357,2714,3134,539memorypamięć736138,8782,3984,0035,136valetudinarianismsłabowitość268,7356,2057,1935,7311optionopcja5738,6876,1675,0536,165to get a good sleepwyspanie151,8048,1078,4536,358customerklient113135,5953,9142,9537,056altarołtarz73439,3048,0534,1539,806farcefarsa14437,4251,1333,9647,925cougarpuma6734,2848,7826,2859,114kennelpsiarnia1237,0751,2536,5063,508barbarianbarbarzyńca51943,8650,5534,5074,6411pestszkodnik3336,5748,4136,8675,148crusadekrucjata13741,5759,8633,1478,488scionpotomek43760,1335,2565,8839,007snackprzekąska3764,0433,4344,5039,119mamamama993693,6018,6772,5339,274formationformowanie6224,6764,3937,0039,6110emperorcesarz339618,2863,7837,5640,006clerkurzędnik18776,7283,5654,5040,508pizzapizza12955,4319,0541,1042,485cakeciasto94953,0031,6437,0542,916bufferbufor4222,5567,6737,1743,005calmspokój13,45457,1047,7469,1543,696tolerancetolerancja13968,7142,0586,6743,8610hotelhotel192441,3943,4437,1144,445garbageśmieci257025,4845,7648,5245,296vitalitywitalność6240,7742,6872,8647,649boilerbojler1715,6033,5339,2548,136medicinemedycyna32233,9690,6170,7148,898funzabawa236068,3522,5458,6953,046feelinguczucie897584,699,8077,4853,687fantasykaprys39934,9434,1137,2854,226dentistdentysta16211,0687,0052,2858,508fevergorączka111139,0041,4444,4860,008jaundiceżółtaczka146,4469,5653,5062,289agonyagonia14353,2434,9148,6269,906horrorprzerażenie256443,4543,0340,7280,1311despairrozpacz280765,0033,6754,0989,097autumnjesień126650,4635,3930,7121,296fungus/ mushroomgrzyb33614,6546,908,0523,355twisttwist1353,0911,2061,9523,405nostalgianostalgia11860,8329,4854,1724,929blissbłogość14720,1351,8730,7525,337unitjednostka91929,2950,8347,1326,339gingin13430,2229,4031,0026,703gritpiasek304457,4738,2425,5028,386rulezasada124133,4573,1066,9528,506rulezasada124133,4573,1066,9528,506charm / glamoururok154251,877,6129,8129,104sumsuma90028,2170,9643,2229,254wickknot97745,0045,2012,0031,674profunditygłębia24355,2343,2336,9132,916mirrorlustro229247,2138,9542,3734,686furfutro125437,3837,1735,0035,285eveningwieczór10,94459,5041,4561,6535,757winterzima174945,4042,5653,1738,444rublerubel8033,4759,8435,4240,215shrineświątynia84236,1046,6531,2548,109bumblebeetrzmiel6437,8241,2323,6450,737casinokasyno14028,0050,7834,2854,726witchczarownica171234,0047,2023,4561,7010fetishfetysz3955,8633,7426,3467,756orgyorgia10860,0515,5330,9576,055geniusgeniusz93673,4765,6045,6313,607leekpor78552,0059,6022,5613,873nursepielęgniarka182251,3087,6519,5715,4612sourceźródło314153,2766,8733,2718,006brookpotok118958,0786,0034,5025,875stepkrok12,25944,3066,6756,1527,244waterwoda11,08753,6859,1273,9229,604bookksiążka337948,7171,8161,6729,867lightświatło19,41948,0558,3268,6830,647planetplaneta182249,3967,4363,6130,787pitchtonacja7553,7161,9335,3633,047patternwzorzec68052,0766,9762,4134,227makingrobienie31446,6264,3669,0034,248gulpłyk259647,0060,9542,4034,553thoughtmyśl28,41855,8574,1575,3134,854wisdommądrość145754,1577,3082,2635,007resignationrezygnacja29249,1761,4056,0035,6010describingopisywanie5150,7161,4355,2136,6410compatibilityzgodność20254,0062,3366,9437,008somethingcoś162,43646,6965,5646,5638,383winnerzwycięzca33644,0979,6540,7638,729polopolo59744,4075,3312,8851,534humiliationupokorzenie43470,2275,6211,5258,1911despotismdespotyzm7044,8459,5836,3262,379cellowiolonczela4449,4061,3333,8864,2011controllingkontrolowanie7520,3874,0965,6558,4613womankobieta28,50279,9340,3377,8958,467scourgeplaga19516,8864,8456,8059,045colossuskolos22128,5060,9561,4159,365birthdayurodziny114160,6224,4360,3859,438interestodsetki5819,2876,6165,0059,447coffintrumna51827,5548,8358,7260,226racewyścig45526,8655,4560,9562,646rescueuratowanie145873,5331,4772,1160,6310desireochota84367,4321,6159,6760,746pleasureprzyjemność512681,1719,6770,1161,1111bravobrawo105367,5832,4761,4761,215lumpguzek1523,5549,1856,1861,915obscenitysprośność1779,3227,5088,6262,309offspringpotomstwo12267,5734,5574,1462,489complaintnarzekanie6442,0041,8356,0060,0010fanaticismfanatyzm12437,1647,5256,0878,328lumpguzek1523,5549,1856,1861,915bulletpocisk213727,5066,5056,2066,456holidaysświęta12370,9018,8061,4563,506rowrząd559615,3859,2156,7576,984scourgeplaga19516,8864,8456,8059,045popularitypopularność47552,4441,7056,9364,4411calamitynieszczęście386847,2435,8157,0578,2412rigorrygor8327,8270,9557,4164,645 Table 1Summary characteristics of the words.EnglishPolishFrequency (Kazojć, 2011)OriginReflectivenessValenceArousalLettersdungeonloch37123,9053,5228,9757,524whorekurwa173135,3030,8038,7191,205graduateabsolwent20622,5283,9565,7632,579affrontafront9523,0772,4726,1386,676mimosamimoza3735,3837,1931,1931,486commandkomenda28612,1072,7124,3366,247mindumysł844116,6586,0587,3632,595mustachewąsy145328,1740,5631,6133,394cannonarmata12635,4837,4328,7081,176troutpstrąg9216,7379,0735,3179,736vomitwymiociny3219,2446,6229,2478,629cheersokrzyki285648,4244,2036,6078,327growlryk262246,6135,4432,6777,223utensilprzyrząd58525,2181,3750,1633,748fussinessawanturnictwo1754,1033,4028,0376,9413knowledgewiedza242033,8992,6892,6330,056theoryteoria164315,4485,3954,7823,786otterwydra8747,6343,8030,7033,535trustzaufanie257079,7638,5785,3323,718self-controlsamoopanowanie133,0078,6884,0522,3714kindnessżyczliwość72178,0032,1580,6334,5210cheerfulnesspogoda ducha5454,586,0523,3077,8027,3212prudencerozwaga11825,3787,5877,7924,637honestyuczciwość58568,6541,6277,7327,389maluszekmaluszek081,6913,5550,5931,068dolphindelfin30749,9634,5023,4819,706diagramschemat55616,5580,0546,7722,957chickenkurczak18723,4087,4715,5026,407calfcielę26738,8320,7815,5426,485aspectaspekt55529,3072,5249,1326,786routinerutyna13026,6271,3868,9327,526dresssukienka51149,4041,2531,7531,058screenekran358330,8057,5556,8031,605conceivingobmyślanie2031,7882,5777,6132,9610spreadrozpiętość7725,5257,8637,5035,0010positionstanowisko439130,3875,1075,6235,1010sentimentsentyment21770,6217,5250,9535,579beetlechrząszcz7520,7043,3017,4336,659streetulica216431,6556,3536,6236,855cradlekołyska5949,4839,3930,1337,437meatmięso290915,3855,0437,2638,835winewino421455,4840,0032,4840,704dragonsmok343827,3843,1920,4851,434clubklub87048,8732,4718,6051,804chunkprzystojniak7742,3040,1028,1455,9012mischiefpsota2245,0827,6432,2658,385discodyskoteka3266,0420,5025,7466,179punkpunk5149,4832,1533,7068,264perversionperwersja5953,6351,6832,0568,429sinzgrzeszenie162,4533,0531,5072,6411cottonbawełna6834,8251,8624,2710,457restodpoczynek123154,3051,9179,8616,7710sketchszkic64540,3561,2446,3023,775twiggałązka27541,2748,4729,5324,677nymphnimfa19261,8149,0514,8028,145headgłowa930234,8483,0062,2128,165thrushdrozd15136,0048,0024,9528,775casewalizka28835,3053,8335,3931,097titletytuł353934,0461,3944,0431,385lacekoronka6544,8352,2225,2832,337patiencecierpliwość185066,7155,4782,3933,7511patiencecierpliwość185066,7155,4782,3933,7511ethicalityetyczność034,1357,2714,3134,539memorypamięć736138,8782,3984,0035,136valetudinarianismsłabowitość268,7356,2057,1935,7311optionopcja5738,6876,1675,0536,165to get a good sleepwyspanie151,8048,1078,4536,358customerklient113135,5953,9142,9537,056altarołtarz73439,3048,0534,1539,806farcefarsa14437,4251,1333,9647,925cougarpuma6734,2848,7826,2859,114kennelpsiarnia1237,0751,2536,5063,508barbarianbarbarzyńca51943,8650,5534,5074,6411pestszkodnik3336,5748,4136,8675,148crusadekrucjata13741,5759,8633,1478,488scionpotomek43760,1335,2565,8839,007snackprzekąska3764,0433,4344,5039,119mamamama993693,6018,6772,5339,274formationformowanie6224,6764,3937,0039,6110emperorcesarz339618,2863,7837,5640,006clerkurzędnik18776,7283,5654,5040,508pizzapizza12955,4319,0541,1042,485cakeciasto94953,0031,6437,0542,916bufferbufor4222,5567,6737,1743,005calmspokój13,45457,1047,7469,1543,696tolerancetolerancja13968,7142,0586,6743,8610hotelhotel192441,3943,4437,1144,445garbageśmieci257025,4845,7648,5245,296vitalitywitalność6240,7742,6872,8647,649boilerbojler1715,6033,5339,2548,136medicinemedycyna32233,9690,6170,7148,898funzabawa236068,3522,5458,6953,046feelinguczucie897584,699,8077,4853,687fantasykaprys39934,9434,1137,2854,226dentistdentysta16211,0687,0052,2858,508fevergorączka111139,0041,4444,4860,008jaundiceżółtaczka146,4469,5653,5062,289agonyagonia14353,2434,9148,6269,906horrorprzerażenie256443,4543,0340,7280,1311despairrozpacz280765,0033,6754,0989,097autumnjesień126650,4635,3930,7121,296fungus/ mushroomgrzyb33614,6546,908,0523,355twisttwist1353,0911,2061,9523,405nostalgianostalgia11860,8329,4854,1724,929blissbłogość14720,1351,8730,7525,337unitjednostka91929,2950,8347,1326,339gingin13430,2229,4031,0026,703gritpiasek304457,4738,2425,5028,386rulezasada124133,4573,1066,9528,506rulezasada124133,4573,1066,9528,506charm / glamoururok154251,877,6129,8129,104sumsuma90028,2170,9643,2229,254wickknot97745,0045,2012,0031,674profunditygłębia24355,2343,2336,9132,916mirrorlustro229247,2138,9542,3734,686furfutro125437,3837,1735,0035,285eveningwieczór10,94459,5041,4561,6535,757winterzima174945,4042,5653,1738,444rublerubel8033,4759,8435,4240,215shrineświątynia84236,1046,6531,2548,109bumblebeetrzmiel6437,8241,2323,6450,737casinokasyno14028,0050,7834,2854,726witchczarownica171234,0047,2023,4561,7010fetishfetysz3955,8633,7426,3467,756orgyorgia10860,0515,5330,9576,055geniusgeniusz93673,4765,6045,6313,607leekpor78552,0059,6022,5613,873nursepielęgniarka182251,3087,6519,5715,4612sourceźródło314153,2766,8733,2718,006brookpotok118958,0786,0034,5025,875stepkrok12,25944,3066,6756,1527,244waterwoda11,08753,6859,1273,9229,604bookksiążka337948,7171,8161,6729,867lightświatło19,41948,0558,3268,6830,647planetplaneta182249,3967,4363,6130,787pitchtonacja7553,7161,9335,3633,047patternwzorzec68052,0766,9762,4134,227makingrobienie31446,6264,3669,0034,248gulpłyk259647,0060,9542,4034,553thoughtmyśl28,41855,8574,1575,3134,854wisdommądrość145754,1577,3082,2635,007resignationrezygnacja29249,1761,4056,0035,6010describingopisywanie5150,7161,4355,2136,6410compatibilityzgodność20254,0062,3366,9437,008somethingcoś162,43646,6965,5646,5638,383winnerzwycięzca33644,0979,6540,7638,729polopolo59744,4075,3312,8851,534humiliationupokorzenie43470,2275,6211,5258,1911despotismdespotyzm7044,8459,5836,3262,379cellowiolonczela4449,4061,3333,8864,2011controllingkontrolowanie7520,3874,0965,6558,4613womankobieta28,50279,9340,3377,8958,467scourgeplaga19516,8864,8456,8059,045colossuskolos22128,5060,9561,4159,365birthdayurodziny114160,6224,4360,3859,438interestodsetki5819,2876,6165,0059,447coffintrumna51827,5548,8358,7260,226racewyścig45526,8655,4560,9562,646rescueuratowanie145873,5331,4772,1160,6310desireochota84367,4321,6159,6760,746pleasureprzyjemność512681,1719,6770,1161,1111bravobrawo105367,5832,4761,4761,215lumpguzek1523,5549,1856,1861,915obscenitysprośność1779,3227,5088,6262,309offspringpotomstwo12267,5734,5574,1462,489complaintnarzekanie6442,0041,8356,0060,0010fanaticismfanatyzm12437,1647,5256,0878,328lumpguzek1523,5549,1856,1861,915bulletpocisk213727,5066,5056,2066,456holidaysświęta12370,9018,8061,4563,506rowrząd559615,3859,2156,7576,984scourgeplaga19516,8864,8456,8059,045popularitypopularność47552,4441,7056,9364,4411calamitynieszczęście386847,2435,8157,0578,2412rigorrygor8327,8270,9557,4164,645 Summary characteristics of the words. Procedure Our experiment was conducted online; we used the Gorilla platform for programming and conducting the N-back task. However, in order to ensure that the conditions of the experimental procedure would be as similar to the stationary lab as possible, we invited our participants to a Google Meet. As participants entered the meeting, they were asked to turn on their microphones and cameras for the whole duration; the experimenter also kept their camera on for the whole duration, but their microphone was silenced during the experiment, so it would not be a distraction. The participants could, therefore, see and hear the experimenter; they could ask questions and report any problems during the experiment. Furthermore, the experimenter could supervise the experiment just as they would in a stationary lab: monitor whether the participants skip the instructions, need help, are distracted, etc. The Google Meet meeting began with the experimenter greeting the participants and providing them with information about the study, including its general aim, the step-by-step procedure, and an estimated duration (approximately 15–20 min). The experimenter answered all potential questions about the procedure, and when the participants were ready, they were given the link to the Gorilla experiment. The first part of the Gorilla procedure was the training. Participants read the instructions explaining the N-back task, after which they were presented with 21 trials of single letters, presented either at the top or the bottom of the screen. Their task was to remember letters and answer whether the letter currently presented was the same as the letter presented three letters ago. To answer, participants used the keyboard arrows (the left arrow as “yes”, and the right arrow as “no”). The training session contained some additional features compared to the main task: the arrows were described on the screen, and after each trial, participants saw feedback, depending on whether their answer was correct (“That’s correct!”) or not (“That is not the right answer”). If participants were correct on less than 40% of trials, they would see instructions asking them to repeat the training, and encouraging them to ask the experimenter questions, if they had any. If participants were correct on at least 40% of trials, they were automatically directed to the main task. The main differences between the training and the main task (explained in an instruction) were the lack of the description of the arrows indicating the answers to the task, the lack of feedback, and the presence of the distracting stimuli. Distractors were the words (see Materials), presented one at a time in a randomized order. There were always two stimuli on the screen: one target (letter), and one distractor (word), presented in a randomized order either at the bottom or the top of the screen. Participants were asked to only answer regarding the letters; they completed the N-back task in approximately 10–15 min. After completing the N-back task, participants were asked to close the Gorilla tab in their browser and return to the experimenter, where they were thanked for their participation in the study and encouraged to ask questions about the procedure or their experience with it. Analysis plan We prepared the data by deleting reaction times shorter than 100ms and longer than 15s (21 trials in total). We also excluded four participants due to either very short reaction times, monotonous answers (clicking the same answer at least 10 times in a row, indicating a lack of attention and mechanical clicking through the task), or low accuracy in completing the task (lower than 40%), which resulted in a database consisting of 9779 trials. After that, the mean scores of accuracy and reaction times were created for each group of words for each participant. Results We conducted a repeated measures ANOVA with Greenhouse–Geisser correction and observed a significant effect between groups of words in task accuracy: F(4.89, 264.22) = 13.66, p < .001, η2 = 0.20, 90% CI: [0.12, 0.26]. We checked for pairwise comparisons, using the Bonferroni correction. All of the comparisons are presented in Table 2 and visualized in Fig. 2. Table 2All pairwise comparisons of the accuracy in seven groups of words.MSDt(df) p d95% CIControl group of wordsM = 0.62, SD = 0.10High ambivalence0.680.183.23(55)0.0010.43[0.16, 0.70]High Ambivalence0.680.173.31(55)< 0.0010.44[0.15, 0.70]Moderate Origin Ambiguity0.720.114.49(55)< 0.0010.6[0.31, 0.88]High Origin Ambiguity0.730.115.89(55)< 0.0010.79[0.49, 1.10]Moderate Activation Ambiguity0.60.121.16(55)0.130.16[-0.11, 0.42]High Activation Ambiguity0.630.170.14(55)0.440.02[-0.28, 0.24]Moderate ambivalenceM = 0.68, SD = 0.18Moderate Origin Ambiguity0.720.111.93(55)0.030.26[0.01, 0.52]High Origin Ambiguity0.730.112.06(55)0.020.28[0.01, 0.54]Moderate Activation Ambiguity0.600.123.96(55)< 0.0010.53[0.25, 0.81]High Activation Ambiguity0.630.173.26(55)< 0.0010.44[0.16, 0.71]High Ambivalence0.680.170.16(55)0.440.02[-0.24, 0.28]High ambivalenceM = 0.68, SD = 0.17Moderate Origin Ambiguity0.720.112.11(55)0.020.28[0.01, 0.55]High Origin Ambiguity0.730.112.15(55)0.020.29[0.02, 0.56]Moderate Activation Ambiguity0.600.124.34(55)< 0.0010.58[0.29, 0.86]High Activation Ambiguity0.630.173.05(55)0.0020.41[0.13, 0.68]Moderate origin ambiguityM = 0.72, SD = 0.11Moderate Activation Ambiguity0.600.126.17(55)< 0.0010.83[0.52, 1.13]High Activation Ambiguity0.630.175.37(55)< 0.0010.72[0.42, 1.10]High Origin Ambiguity0.730.110.48(55)0.950.01[-0.27, 0.26]High origin ambiguityM = 0.73, SD = 0.11Moderate Activation Ambiguity0.600.126.37(55)< 0.0010.85[0.54, 1.16]High Activation Ambiguity0.630.175.11(55)< 0.0010.69[0.39, 0.98]Moderate activationM = 0.60, SD = 0.12High Activation Ambiguity0.630.171.18(55)0.120.16[0.11, 0.42] Table 2All pairwise comparisons of the accuracy in seven groups of words.MSDt(df) p d95% CIControl group of wordsM = 0.62, SD = 0.10High ambivalence0.680.183.23(55)0.0010.43[0.16, 0.70]High Ambivalence0.680.173.31(55)< 0.0010.44[0.15, 0.70]Moderate Origin Ambiguity0.720.114.49(55)< 0.0010.6[0.31, 0.88]High Origin Ambiguity0.730.115.89(55)< 0.0010.79[0.49, 1.10]Moderate Activation Ambiguity0.60.121.16(55)0.130.16[-0.11, 0.42]High Activation Ambiguity0.630.170.14(55)0.440.02[-0.28, 0.24]Moderate ambivalenceM = 0.68, SD = 0.18Moderate Origin Ambiguity0.720.111.93(55)0.030.26[0.01, 0.52]High Origin Ambiguity0.730.112.06(55)0.020.28[0.01, 0.54]Moderate Activation Ambiguity0.600.123.96(55)< 0.0010.53[0.25, 0.81]High Activation Ambiguity0.630.173.26(55)< 0.0010.44[0.16, 0.71]High Ambivalence0.680.170.16(55)0.440.02[-0.24, 0.28]High ambivalenceM = 0.68, SD = 0.17Moderate Origin Ambiguity0.720.112.11(55)0.020.28[0.01, 0.55]High Origin Ambiguity0.730.112.15(55)0.020.29[0.02, 0.56]Moderate Activation Ambiguity0.600.124.34(55)< 0.0010.58[0.29, 0.86]High Activation Ambiguity0.630.173.05(55)0.0020.41[0.13, 0.68]Moderate origin ambiguityM = 0.72, SD = 0.11Moderate Activation Ambiguity0.600.126.17(55)< 0.0010.83[0.52, 1.13]High Activation Ambiguity0.630.175.37(55)< 0.0010.72[0.42, 1.10]High Origin Ambiguity0.730.110.48(55)0.950.01[-0.27, 0.26]High origin ambiguityM = 0.73, SD = 0.11Moderate Activation Ambiguity0.600.126.37(55)< 0.0010.85[0.54, 1.16]High Activation Ambiguity0.630.175.11(55)< 0.0010.69[0.39, 0.98]Moderate activationM = 0.60, SD = 0.12High Activation Ambiguity0.630.171.18(55)0.120.16[0.11, 0.42] All pairwise comparisons of the accuracy in seven groups of words. Control group of words M = 0.62, SD = 0.10 Moderate ambivalence M = 0.68, SD = 0.18 High ambivalence M = 0.68, SD = 0.17 Moderate origin ambiguity M = 0.72, SD = 0.11 High origin ambiguity M = 0.73, SD = 0.11 Moderate activation M = 0.60, SD = 0.12 Fig. 2Main effect of the accuracy of answers in the N-back task in Experiment 1. The error bars are standard deviations, the horizontal lines indicate statistically significant comparisons, with the asterisks for significance: * - p < .05, ** - p < .01,and *** - < .001. Main effect of the accuracy of answers in the N-back task in Experiment 1. The error bars are standard deviations, the horizontal lines indicate statistically significant comparisons, with the asterisks for significance: * - p < .05, ** - p < .01,and *** - < .001. We also conducted a repeated measures ANOVA to check for differences in reaction times (transformed into natural logarithms) for the seven groups of stimuli. However, there were no significant differences between the groups: F(6,330) = 0.51, p = .80, η2 = 0.009, CI: [0.01, 0.11]. None of the comparisons were statistically significant, as presented in Table 3. Table 3Pairwise comparisons of mean reaction times between groups of words.MSDt p d95% CIControl group of wordsM = 2146.21, SD = 1231.94Moderate Ambivalence2020.89925.221.250.220.16[-0.09, 0.42]High Ambivalence1984.13923.131.460.150.19[-0.09, 0.41]Moderate Origin Ambiguity2037.70871.470.920.360.12[-0.14, 0.37]High Origin Ambiguity2078.70971.400.510.610.07[-0.19, 0.32]Moderate Activation Ambiguity2057.521040.480.830.410.11[-0.15, 0.36]High Activation Ambiguity2013.65980.881.350.190.17[-0.08, 0.43]Moderate ambivalenceM = 2020.89, SD = 925.22High Ambivalence1984.13923.130.480.630.06[-0.19, 0.32]Moderate Origin2037.70871.470.190.850.02[-0.28, 0.23]High Origin2078.70971.400.670.510.09[-0.34, 0.17]Moderate Activation2057.521040.480.420.680.05[-0.31, 0.20]High Activation2013.65980.880.110.920.01[-0.24, 0.27]High ambivalenceM = 1984.13, SD = 923.13Moderate Origin2037.70871.470.550.590.07[-0.32, 0.18]High Origin2078.70971.400.920.360.12[-0.37, 0.14]Moderate Activation2057.521040.480.820.420.11[-0.36, 0.15]High Activation2013.65980.880.460.650.06[-0.31, 0.19]Moderate origin ambiguityM = 2037.70, SD = 871.47High Origin2078.70971.400.390.70.05[-0.30, 0.20]Moderate Activation2057.521040.480.210.840.03[-0.28, 0.23]High Activation2013.65980.880.290.770.04[-0.22, 0.29]High origin ambiguityM = 2078.70, SD = 971.40Moderate Activation2057.521040.480.190.850.03[-0.23, 0.27]High Activation2013.65980.880.630.530.08[-0.17, 0.33]Moderate activationM = 2057.52, SD = 1040.48High Activation2013.65980.880.470.640.06[-0.19, 0.31] Table 3Pairwise comparisons of mean reaction times between groups of words.MSDt p d95% CIControl group of wordsM = 2146.21, SD = 1231.94Moderate Ambivalence2020.89925.221.250.220.16[-0.09, 0.42]High Ambivalence1984.13923.131.460.150.19[-0.09, 0.41]Moderate Origin Ambiguity2037.70871.470.920.360.12[-0.14, 0.37]High Origin Ambiguity2078.70971.400.510.610.07[-0.19, 0.32]Moderate Activation Ambiguity2057.521040.480.830.410.11[-0.15, 0.36]High Activation Ambiguity2013.65980.881.350.190.17[-0.08, 0.43]Moderate ambivalenceM = 2020.89, SD = 925.22High Ambivalence1984.13923.130.480.630.06[-0.19, 0.32]Moderate Origin2037.70871.470.190.850.02[-0.28, 0.23]High Origin2078.70971.400.670.510.09[-0.34, 0.17]Moderate Activation2057.521040.480.420.680.05[-0.31, 0.20]High Activation2013.65980.880.110.920.01[-0.24, 0.27]High ambivalenceM = 1984.13, SD = 923.13Moderate Origin2037.70871.470.550.590.07[-0.32, 0.18]High Origin2078.70971.400.920.360.12[-0.37, 0.14]Moderate Activation2057.521040.480.820.420.11[-0.36, 0.15]High Activation2013.65980.880.460.650.06[-0.31, 0.19]Moderate origin ambiguityM = 2037.70, SD = 871.47High Origin2078.70971.400.390.70.05[-0.30, 0.20]Moderate Activation2057.521040.480.210.840.03[-0.28, 0.23]High Activation2013.65980.880.290.770.04[-0.22, 0.29]High origin ambiguityM = 2078.70, SD = 971.40Moderate Activation2057.521040.480.190.850.03[-0.23, 0.27]High Activation2013.65980.880.630.530.08[-0.17, 0.33]Moderate activationM = 2057.52, SD = 1040.48High Activation2013.65980.880.470.640.06[-0.19, 0.31] Pairwise comparisons of mean reaction times between groups of words. Control group of words M = 2146.21, SD = 1231.94 Moderate ambivalence M = 2020.89, SD = 925.22 High ambivalence M = 1984.13, SD = 923.13 Moderate origin ambiguity M = 2037.70, SD = 871.47 High origin ambiguity M = 2078.70, SD = 971.40 Moderate activation M = 2057.52, SD = 1040.48 Experiment 2 Methods Participants As our design was identical to the design in Experiment 1, our minimal sample size computed a priori was the same (40 people). For Experiment 2, we decided to recruit 44 people (22 women, 22 men); however, due to the low quality of data (caused by a lot of head movement from the participants), we had to exclude three participants. That left us with a sample of 41 people (20 women, 21 men) who fulfilled the required criteria (identical to Experiment 1). We informed participants about following the ethical guidance (identical to Experiment 1). Once again, we obtained informed consent from participants (with the possibility to withdraw it at any moment), we analyzed results in groups only, and we offered remuneration of zł 60 (about 14.50 USD) for completing the experiment. Furthermore, we understood that the webcam eye-tracking may evoke suspicion of being recorded, so we explicitly explained to participants that we would only obtain numerical data from them in an Excel file. All the procedures were conducted in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. We obtained the approval of the Research Ethics Committee of the Faculty of Psychology at the University of Warsaw (approval no. 14/11/22). Design The experimental design in Experiment 2 was identical to the design in Experiment 1: a repeated measures ANOVA with seven groups (words: control, moderate ambivalence, high ambivalence, moderate origin ambiguity, high origin ambiguity, moderate activation ambiguity, high activation ambiguity). Materials The materials were identical to the ones used in Experiment 1. Procedure The procedure in Experiment 2 was similar to the procedure described in Experiment 1. Our participants were invited to the online meeting, where they were greeted by the experimenter, who explained the study details, with an important addition about the eye-tracking procedure. The experimenter told participants that there was no recording during the experiment – we only obtained numerical data in an Excel file. The experimenter also warned participants about the calibration process: the fact that participants need to follow the instructions, adjust the camera to them (sit at the right distance, fit their face in the frame so their webcam could follow their gaze), and not move. The experimenter also told participants that the eye-tracking platform will repeat the calibration process if needed, and even stop the experiment if it detects movement and the need for additional calibration. After providing these instructions, the experimenter gave participants the link to the RealEye platform, where they could take part in the experiment. The procedure started with a calibration process: first, participants were presented with a view from their camera and asked to sit so that their faces fit in a small green frame – in the right position and distance. After that, participants were asked to follow 40 little red dots, appearing one by one on a screen; they should track them with both their eye gaze and the mouse pointer. The participants were then asked to stare directly into one of the four sequentially presented dots until it exploded. If the calibration was unsuccessful, the experimenter started helping participants: asking them to adjust their position, change the lighting in the room, or try to stay more still during the process. If the calibration was successful, participants were automatically directed to the N-back training and the main task – identical to Experiment 1. Taking into consideration the fact that our procedure was relatively long, we inserted one additional calibration – after half of the main task. Participants were instructed to rest for a little while (not longer than about a minute), sit comfortably, and repeat the calibration procedure. That was supposed to ensure the data were of high quality, which is even more crucial during the online procedure. After completing the second half of the experiment, participants were asked to close the tab and return to the experimenter, who thanked them for their participation and answered any questions they had. The whole procedure took approximately 20–25 min. Results For the webcam-based eye-tracking experiment, we analyzed the number of fixations on a distractor word, and their mean duration. As a fixation, we took into consideration all the stops of a gaze on an area of the word and the screen surrounding it (18% of the screen) that lasted at least 100 milliseconds83. We analyzed the differences in mean number of fixations between groups using a repeated measures ANOVA with seven dependent measurements: F(4.23, 160.72) = 4.15, p = .003, η2 = 0.10; the effect is reported with a Greenhouse–Geisser correction due to a lack of sphericity in the data (checked with Mauchly’s test). The pairwise comparisons, checked with t-tests with Bonferroni correction, are presented in Table 4 and in Fig. 3. Table 4Pairwise comparisons of mean number of fixations between groups of words.ComparisonMSDt(df) p d95% CIControl group of wordsM = 1.18, SD = 0.29Moderate Ambivalence1.540.753.18 (38)0.0030.51[0.17, 0.84]High Ambivalence1.640.654.47 (38)< 0.0010.72[0.34, 1.06]Moderate Origin Ambiguity1.770.545.73 (38)< 0.0010.92[0.54, 1.29]High Origin Ambiguity1.780.694.82 (38)< 0.0010.77[0.41, 1.12]Moderate Activation Ambiguity1.620.653.72 (38)< 0.0010.60[0.25, 0.93]High Activation Ambiguity1.660.674.21 (38)< 0.0010.68[0.32, 1.02]Moderate ambivalenceM = 1.54, SD = 0.75High ambivalence1.640.651.07(38)0.29-0.17[-0.49, 0.15]Moderate origin ambiguity1.770.541.43(38)0.16-0.23[-0.55, 0.09]High origin ambiguity1.780.691.42(38)0.16-0.23[-0.55, 0.09]Moderate activation ambiguity1.620.650.44(38)0.66-0.07[-0.39, 0.24]High activation ambiguity1.660.670.74(38)0.46-0.12[-0.43, 0.20]High ambivalenceM = 1.64, SD = 0.65Moderate origin ambiguity1.770.540.87(38)0.390.14[-0.45, 0.18]High origin ambiguity1.780.691.02(38)0.310.16[-0.48, 0.15]Moderate activation ambiguity1.620.650.11(38)0.910.02[-0.30, 0.33]High activation ambiguity1.660.670.17(38)0.870.03[-0.34, 0.29]Moderate origin ambiguityM = 1.77, SD = 0.54High origin ambiguity1.780.690.05(38)0.960.16[-0.32, 0.31]Moderate activation ambiguity1.620.650.99(38)0.330.12[-0.16, 0.48]High activation ambiguity1.660.670.72(38)0.470.16[-0.20, 0.43]High origin ambiguityM = 1.78, SD = 0.69Moderate activation ambiguity1.620.651.01(38)0.320.16[-0.16, 0.48]High activation ambiguity1.660.670.68(38)0.500.11[-0.21, 0.42]Moderate activation ambiguityM = 1.62, SD = 0.67High activation ambiguity1.660.670.31(38)0.760.05[-0.36, 0.27] Table 4Pairwise comparisons of mean number of fixations between groups of words.ComparisonMSDt(df) p d95% CIControl group of wordsM = 1.18, SD = 0.29Moderate Ambivalence1.540.753.18 (38)0.0030.51[0.17, 0.84]High Ambivalence1.640.654.47 (38)< 0.0010.72[0.34, 1.06]Moderate Origin Ambiguity1.770.545.73 (38)< 0.0010.92[0.54, 1.29]High Origin Ambiguity1.780.694.82 (38)< 0.0010.77[0.41, 1.12]Moderate Activation Ambiguity1.620.653.72 (38)< 0.0010.60[0.25, 0.93]High Activation Ambiguity1.660.674.21 (38)< 0.0010.68[0.32, 1.02]Moderate ambivalenceM = 1.54, SD = 0.75High ambivalence1.640.651.07(38)0.29-0.17[-0.49, 0.15]Moderate origin ambiguity1.770.541.43(38)0.16-0.23[-0.55, 0.09]High origin ambiguity1.780.691.42(38)0.16-0.23[-0.55, 0.09]Moderate activation ambiguity1.620.650.44(38)0.66-0.07[-0.39, 0.24]High activation ambiguity1.660.670.74(38)0.46-0.12[-0.43, 0.20]High ambivalenceM = 1.64, SD = 0.65Moderate origin ambiguity1.770.540.87(38)0.390.14[-0.45, 0.18]High origin ambiguity1.780.691.02(38)0.310.16[-0.48, 0.15]Moderate activation ambiguity1.620.650.11(38)0.910.02[-0.30, 0.33]High activation ambiguity1.660.670.17(38)0.870.03[-0.34, 0.29]Moderate origin ambiguityM = 1.77, SD = 0.54High origin ambiguity1.780.690.05(38)0.960.16[-0.32, 0.31]Moderate activation ambiguity1.620.650.99(38)0.330.12[-0.16, 0.48]High activation ambiguity1.660.670.72(38)0.470.16[-0.20, 0.43]High origin ambiguityM = 1.78, SD = 0.69Moderate activation ambiguity1.620.651.01(38)0.320.16[-0.16, 0.48]High activation ambiguity1.660.670.68(38)0.500.11[-0.21, 0.42]Moderate activation ambiguityM = 1.62, SD = 0.67High activation ambiguity1.660.670.31(38)0.760.05[-0.36, 0.27] Pairwise comparisons of mean number of fixations between groups of words. Control group of words M = 1.18, SD = 0.29 Moderate ambivalence M = 1.54, SD = 0.75 1.07 (38) 1.43 (38) 1.42 (38) 0.44 (38) 0.74 (38) High ambivalence M = 1.64, SD = 0.65 0.87 (38) 1.02 (38) 0.11 (38) 0.17 (38) Moderate origin ambiguity M = 1.77, SD = 0.54 0.05 (38) 0.99 (38) 0.72 (38) High origin ambiguity M = 1.78, SD = 0.69 1.01 (38) 0.68 (38) Moderate activation ambiguity M = 1.62, SD = 0.67 0.31 (38) Fig. 3Main effect of the number of fixations on distractor words during the N-back task in Experiment 2. The error bars are standard deviations, the horizontal lines indicate statistically significant comparisons, with the asterisks for significance: * - p < .05, ** - p < .01, and *** - p < .001. Main effect of the number of fixations on distractor words during the N-back task in Experiment 2. The error bars are standard deviations, the horizontal lines indicate statistically significant comparisons, with the asterisks for significance: * - p < .05, ** - p < .01, and *** - p < .001. Similarly, we also compared the mean duration of fixation between the groups, obtaining a significant main effect in a repeated measures ANOVA: F(3.07, 116.83) = 14.09, p < .002, η2 = 0.27; the effect is reported with a Greenhouse–Geisser correction due to a lack of sphericity in the data (checked with Mauchly’s test). The pairwise comparisons are presented in Table 5 and visualized in Fig. 4. Table 5Pairwise comparisons of mean duration of fixations between groups of words.ComparisonMSDt(df) p d95% CIControl group of wordsM = 292.30, SD = 63.48Moderate Ambivalence504.03190.347.02 (38)0.003-1.12[-1.52, -0.72]High Ambivalence526.97197.531.55 (38)< 0.04-1.25[-0.56, 0.07]Moderate Origin Ambiguity681.34287.197.65 (38)< 0.001-1.36[-1.64, -0.80]High Origin Ambiguity467.54158.118.51 (38)< 0.001-1.37[-1.80, -0.92]Moderate Activation Ambiguity521.04171.948.53 (38)< 0.001-1.05[-1.43, -0.65]High Activation Ambiguity523.96250.106.53 (38)< 0.001-1.43[0.32, 1.02]Moderate ambivalenceM = 504.03, SD = 190.34High ambivalence526.97197.534.75(38)< 0.0010.76[0.40, 1.11]Moderate origin ambiguity681.34287.193.05(38)0.002-0.82[-0.82, -0.15]High origin ambiguity467.54158.110.97(38)0.17-0.16[-0.16, 0.47]Moderate activation ambiguity521.04171.940.43(38)0.34-0.16[-0.38, 0.24]High activation ambiguity523.96250.100.39(38)0.35-0.06[-0.38, 0.25]High ambivalenceM = 526.97, SD = 197.53Moderate origin ambiguity681.34287.196.84(38)< 0.001-1.10[-1.49, -0.69]High origin ambiguity467.54158.116.34(38)< 0.001-1.01[-1.40, -0.62]Moderate activation ambiguity521.04171.948.65(38)< 0.001-1.39[-1.82, -0.94]High activation ambiguity523.96250.107.96(38)< 0.001-1.28[-1.69, -0.85]Moderate origin ambiguityM = 681.34, SD = 287.19High origin ambiguity467.54158.113.80(38)< 0.001-1.69[-1.69, -0.31]Moderate activation ambiguity521.04171.942.81(38)0.0040.61[-0.26, 0.95]High activation ambiguity523.96250.102.35(38)0.010.45[0.12, 0.78]High origin ambiguityM = 467.54, SD = 158.11Moderate activation ambiguity521.04171.942.27(38)0.010.38[0.05, 0.70]High activation ambiguity523.96250.100.68(38)0.02-0.39[-0.68, -0.04]Moderate activation ambiguityM = 521.04, SD = 171.94High activation ambiguity523.96250.100.10(38)0.46-0.02[-0.33, 0.30] Table 5Pairwise comparisons of mean duration of fixations between groups of words.ComparisonMSDt(df) p d95% CIControl group of wordsM = 292.30, SD = 63.48Moderate Ambivalence504.03190.347.02 (38)0.003-1.12[-1.52, -0.72]High Ambivalence526.97197.531.55 (38)< 0.04-1.25[-0.56, 0.07]Moderate Origin Ambiguity681.34287.197.65 (38)< 0.001-1.36[-1.64, -0.80]High Origin Ambiguity467.54158.118.51 (38)< 0.001-1.37[-1.80, -0.92]Moderate Activation Ambiguity521.04171.948.53 (38)< 0.001-1.05[-1.43, -0.65]High Activation Ambiguity523.96250.106.53 (38)< 0.001-1.43[0.32, 1.02]Moderate ambivalenceM = 504.03, SD = 190.34High ambivalence526.97197.534.75(38)< 0.0010.76[0.40, 1.11]Moderate origin ambiguity681.34287.193.05(38)0.002-0.82[-0.82, -0.15]High origin ambiguity467.54158.110.97(38)0.17-0.16[-0.16, 0.47]Moderate activation ambiguity521.04171.940.43(38)0.34-0.16[-0.38, 0.24]High activation ambiguity523.96250.100.39(38)0.35-0.06[-0.38, 0.25]High ambivalenceM = 526.97, SD = 197.53Moderate origin ambiguity681.34287.196.84(38)< 0.001-1.10[-1.49, -0.69]High origin ambiguity467.54158.116.34(38)< 0.001-1.01[-1.40, -0.62]Moderate activation ambiguity521.04171.948.65(38)< 0.001-1.39[-1.82, -0.94]High activation ambiguity523.96250.107.96(38)< 0.001-1.28[-1.69, -0.85]Moderate origin ambiguityM = 681.34, SD = 287.19High origin ambiguity467.54158.113.80(38)< 0.001-1.69[-1.69, -0.31]Moderate activation ambiguity521.04171.942.81(38)0.0040.61[-0.26, 0.95]High activation ambiguity523.96250.102.35(38)0.010.45[0.12, 0.78]High origin ambiguityM = 467.54, SD = 158.11Moderate activation ambiguity521.04171.942.27(38)0.010.38[0.05, 0.70]High activation ambiguity523.96250.100.68(38)0.02-0.39[-0.68, -0.04]Moderate activation ambiguityM = 521.04, SD = 171.94High activation ambiguity523.96250.100.10(38)0.46-0.02[-0.33, 0.30] Pairwise comparisons of mean duration of fixations between groups of words. Control group of words M = 292.30, SD = 63.48 Moderate ambivalence M = 504.03, SD = 190.34 4.75 (38) 3.05 (38) 0.97 (38) 0.43 (38) 0.39 (38) High ambivalence M = 526.97, SD = 197.53 6.84 (38) 6.34 (38) 8.65 (38) 7.96 (38) Moderate origin ambiguity M = 681.34, SD = 287.19 3.80 (38) 2.81 (38) 2.35 (38) High origin ambiguity M = 467.54, SD = 158.11 2.27 (38) 0.68 (38) Moderate activation ambiguity M = 521.04, SD = 171.94 0.10 (38) Fig. 4Main effect of the mean duration (in milliseconds) of fixation on distractor words during the N-back task in Experiment 2. The error bars are standard deviations, the horizontal lines indicate statistically significant comparisons, with the asterisks for significance: * - p < .05, ** - p < .01, and *** - p < .001. Main effect of the mean duration (in milliseconds) of fixation on distractor words during the N-back task in Experiment 2. The error bars are standard deviations, the horizontal lines indicate statistically significant comparisons, with the asterisks for significance: * - p < .05, ** - p < .01, and *** - p < .001. Discussion Our results partially confirmed our hypotheses: we obtained higher accuracy for ambivalence and origin ambiguity trials compared to control words. Moreover, as expected, there were significantly more numerous and longer fixations for ambiguous words compared to control words. Therefore, our results verify the hypothesis that words of ambiguous valence, origin, and activation have their own characteristics and may significantly influence performance in a cognitive task, even when they are not the target stimuli. First and foremost, in Experiment 1, we only confirmed the hypothesis regarding task accuracy for ambivalence and origin ambiguity. We observed that ambivalence and origin ambiguity increased the accuracy significantly compared with the control words and activation ambiguity. Therefore, the ambiguous stimuli are not only remembered better38 but also, when processed involuntarily, apparently increase memory and recall for other stimuli. It seems that this process may be caused by the properties of the ambiguous stimuli: by being both unequivocal and mixed, those stimuli may evoke more attention to detail, and perhaps involve more attention and more vigilance towards stimuli. We hypothesize that if things are not simple and equivocal, more cognitive resources should be spent on processing the situation, which would explain the increased accuracy in the N-back task only for the words of valence and origin ambiguity. In previous research, the activation dimensions had various influences over cognition: either disrupting it (mainly arousal; work in review), enhancing it (subjective significance49, or interacting with each other (57; work in review). It seems that activation plays a differential role in moderating cognitive processes, and the nature of its influence depends on the specifics of the task. For the N-back task, there was no difference between the accuracy of words ambiguous on the space of activation and unidimensional, control words. Words of activation ambiguity did not improve performance; perhaps the two different kinds of activation were too arousing for this task, requiring the constant remembering and recalling of stimuli and fast answers. In effect, we obtained no improvement in the accuracy of completing the task. Perhaps the properties of activation ambiguity in the emotional load of words are rather difficult to observe, and require some more detailed, sensitive method of measurement to observe the process, and not only the result (e.g., EEG measurement). In Experiment 1, for the first time, we shifted the focus from remembering the ambiguous stimuli themselves to observing their influence as distractor stimuli in the background, as those stimuli were unconsciously processed by the participant while performing the N-back task. It seems that processing those unequivocal, mixed stimuli (even unintentionally) might actually benefit the working memory and increase its performance. These findings are especially important as this is one of the first experiments to consider not only the consequences of ambivalence (which is still understudied but somehow present in studies70,84,85 but also the consequences of origin and activation ambiguity. Moreover, we did not obtain any significant results for reaction time; the improving role of ambivalence and origin ambiguity was only visible in the task accuracy. Perhaps more complicated tasks could better catch those differences. In Experiment 2, we observed differences in both the number and mean duration of fixations between control words and all ambiguous words. The number of fixations was significantly higher for ambiguous words than control words; participants returned their eyes to the ambiguous words, perhaps because their complex and unequivocal nature elicited the need for more resources and longer processing time to fully understand them. Similarly, the mean duration of a fixation was significantly higher for ambiguous words than for control words. These results are in line with previous findings52,53, indicating that the ambiguity in words may be perceived and that there is a need for directing more attention to this kind of stimulus. However, it is important to note the counterintuitive results in Experiments 1 and 2: while in Experiment 1 the ambiguous words did not increase reaction times, they did elicit significantly longer (and more numerous) fixations in Experiment 2. That seems peculiar, as longer fixations are usually an indicator of longer and more effortful processing (in the case of our studies, resulting, as we presented, in higher accuracy). While the underlying mechanism of these results certainly requires more research, perhaps it indicates that participants were searching for a way to compensate for the increased cognitive load (the ambiguity in words) by changing their strategy in completing the task, e.g., by adopting a more efficient encoding or retrieval process of the letters. The distractor words could take up so much of their attention that participants had to complete the task faster and adjust to that requirement. The N-back task is a very specific task, as it requires constant remembering, encoding, and recalling of the stimuli86; therefore, an increased cognitive load in a task that is already rather demanding might have created a situation in which participants were forced to develop a better strategy. It is, however, especially interesting that this effect may be observed when the ambiguous words are just distractor stimuli. Those ambiguous, mixed stimuli use attentional resources and stop participants’ gaze, but at the same time (as we saw in the behavioral results for the valence and origin ambiguities), increase accuracy in the performed task. Therefore, despite taking a lot of resources away from the instructed task, those stimuli do not disrupt the process; perhaps they allow participants to switch their processing mode (from intuitive to more deliberative? 87,88) and improve their performance. That leads us to an even more complex effect, which was produced by the words of activation ambiguity. Those words did not provide any behavioral results – accuracy in their groups did not differ significantly from the control words – but they seemed no different from the valence and origin ambiguities in eye-tracking results. We posit that activation ambiguity may be somewhat halfway between the unidimensional and ambiguous emotional load. Participants seem to perceive it similarly to valence and origin ambiguity (stopping their gaze, returning their gaze to those stimuli), but that does not show in their behavior. Perhaps the ability to notice the ambiguity of dimensions positively correlated to each other (contrary to the ambiguities negatively correlated, such as valence and origin) is an even more difficult process, and is therefore even more subtle. We have already mentioned the need for studies with measurements (like eye-tracking) that are more susceptible to the processing of stimuli, e.g., EEG measurement, or maybe something both behavioral and processual, such as mouse-tracking. A clear and foremost limitation of the study is that both experiments were conducted online, so the environmental conditions may have slightly varied between participants. That said, conditions were meticulously prepared so that apart from meeting face-to-face, the study would proceed as close as possible to laboratory conditions. Research assistants collecting the data clearly asked participants to take part in both experiments in conditions allowing undisturbed completion of cognitive tasks. A working camera and microphone on both sides was a necessary condition for every meeting. We believe that adjustment not only helped us control for environmental conditions, but also facilitated any contact between researcher and participants while making a meeting atmosphere resembling (as close as possible) one taking place in the laboratory. Requirements for participants were clearly stated before participating and thoroughly revised during the course of every meeting. On a couple of occasions, participants who had clearly not adhered to requirements were asked by the assistant to take part in the experiment on a different occasion. Nevertheless, those incidents did not cause any dropouts from the study. In all cases, participants took part in the experiment on a different date and under adjusted sufficient conditions. Any possibly or slightly distracting stimuli, e.g., background noise, were noted by assistants and accounted for during data analysis (e.g., if a reaction time was significantly different around the timestamp of a noise noted by experimenter, we deleted the trial as being uncharacteristic for the participant; there were not many such trials, cf. Results sections). It is also worth indicating that previous experiments conducted in our interest area took place both online and in a laboratory, and no significant differences were found between those two conditions while including very similar cognitive tasks, same group specifics, and, in most cases, the same person collecting the data. Regarding devices used in eye-tracking webcam-based experiments, we may present data suggesting that online-collected webcam data sets are comparable in quality to laboratory-collected ones89,90. A further limitation is that reported studies were conducted on homogenous groups of adolescent and young adult students with Polish as a first language. It is highly recommended that future studies include more diverse groups as well as between-group comparisons based on the fact that our emotional cognitive functioning undergoes significant changes in different stages of life. As our life progresses, we tend to show an increased preference for positive stimuli91,92, thus, we more often feel positive emotions (Gurera, 2019). Middle-aged and older adults also tend to be more passive both cognitively and behaviorally93, which may lead to more overall stability and wellbeing among them, including emotional wellbeing94,95. While Experiment 2 involved the additional measure of eye-tracking, Experiment 1 was purely behavioral, which is a limitation regarding such subtle yet nuanced cognitive phenomena of ambiguity perception. Further studies in the field should possibly involve measures of electro-physiological changes (such as EEG). This addition would help us understand mechanisms underlying ambiguity perception in semantic stimuli. Further broadening the database of words used for the study is also recommended, as our current stimuli collection includes only nouns. Another step would be to isolate ambiguous verbs and nouns and provide affective norms of them, thus making any future studies more related to the actual pragmatic use of language in everyday communication. Our experiments were supposed to further study and verify the theory that emotionally ambiguous stimuli may influence many different facets of our cognitive functioning. Our findings showed that two kinds of ambiguity (valence and origin) in words improved accuracy in the N-back task. All of the ambiguous words elicited distinctive patterns of managing visual attention. Our studies used only one cognitive task (and a specific, rather difficult one that required using many cognitive functions at the same time; 78), but perhaps future studies could further assess whether even unintentional processing of ambiguous words alters our cognitive processing. We argue that it may be an especially important result in a situation in which we are flooded with information and overloaded with stimuli (e.g., while browsing the internet). Our findings may also be a cue for composing memorable communication – if we remember not only the ambiguous stimuli37,38, but also the accompanying information better, we could use this effect (which we only showed in laboratory conditions) in practice in many different areas of linguistics, journalism, or education.